MATLAB REAL-TIME WORKSHOP 7 - TARGET LANGUAGE COMPILER User`s manual

Computation
Visualization
Programming
External Interfaces
Version 6
How to Contact The MathWorks:
www.mathworks.com
comp.soft-sys.matlab
Web
Newsgroup
info@mathworks.com
Technical support
Product enhancement suggestions
Bug reports
Documentation error reports
Order status, license renewals, passcodes
Sales, pricing, and general information
508-647-7000
Phone
508-647-7001
Fax
The MathWorks, Inc.
3 Apple Hill Drive
Natick, MA 01760-2098
Mail
support@mathworks.com
suggest@mathworks.com
bugs@mathworks.com
doc@mathworks.com
service@mathworks.com
For contact information about worldwide offices, see the MathWorks Web site.
MATLAB External Interfaces
 COPYRIGHT 1984 - 2001 by The MathWorks, Inc.
The software described in this document is furnished under a license agreement. The software may be used
or copied only under the terms of the license agreement. No part of this manual may be photocopied or reproduced in any form without prior written consent from The MathWorks, Inc.
FEDERAL ACQUISITION: This provision applies to all acquisitions of the Program and Documentation by
or for the federal government of the United States. By accepting delivery of the Program, the government
hereby agrees that this software qualifies as "commercial" computer software within the meaning of FAR
Part 12.212, DFARS Part 227.7202-1, DFARS Part 227.7202-3, DFARS Part 252.227-7013, and DFARS Part
252.227-7014. The terms and conditions of The MathWorks, Inc. Software License Agreement shall pertain
to the government’s use and disclosure of the Program and Documentation, and shall supersede any
conflicting contractual terms or conditions. If this license fails to meet the government’s minimum needs or
is inconsistent in any respect with federal procurement law, the government agrees to return the Program
and Documentation, unused, to MathWorks.
MATLAB, Simulink, Stateflow, Handle Graphics, and Real-Time Workshop are registered trademarks, and
Target Language Compiler is a trademark of The MathWorks, Inc.
Other product or brand names are trademarks or registered trademarks of their respective holders.
Printing History: December 1996 First printing
July 1997
Revised for 5.1 (online only)
January 1998
Second printing Revised for MATLAB 5.2
October 1998
Third printing
Revised for MATLAB 5.3 (Release 11)
November 2000 Fourth printing Revised and renamed for MATLAB 6.0
(Release 12)
June 2001
Online only
Revised for MATLAB 6.1 (Release 12.1)
Contents
Calling C and Fortran Programs from MATLAB
1
Introducing MEX-Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-3
Using MEX-Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-3
The Distinction Between mx and mex Prefixes . . . . . . . . . . . . . 1-4
MATLAB Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The MATLAB Array . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Data Storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Data Types in MATLAB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Using Data Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1-6
1-6
1-6
1-7
1-9
Building MEX-Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Compiler Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Testing Your Configuration on UNIX . . . . . . . . . . . . . . . . . . .
Testing Your Configuration on Windows . . . . . . . . . . . . . . . . .
Specifying an Options File . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1-11
1-11
1-12
1-14
1-17
Custom Building MEX-Files . . . . . . . . . . . . . . . . . . . . . . . . . . .
Who Should Read This Chapter . . . . . . . . . . . . . . . . . . . . . . . .
MEX Script Switches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Default Options File on UNIX . . . . . . . . . . . . . . . . . . . . . . . . .
Default Options File on Windows . . . . . . . . . . . . . . . . . . . . . . .
Custom Building on UNIX . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Custom Building on Windows . . . . . . . . . . . . . . . . . . . . . . . . . .
1-20
1-20
1-20
1-22
1-23
1-24
1-26
Troubleshooting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Configuration Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Understanding MEX-File Problems . . . . . . . . . . . . . . . . . . . . .
Compiler and Platform-Specific Issues . . . . . . . . . . . . . . . . . .
Memory Management Compatibility Issues . . . . . . . . . . . . . .
1-32
1-32
1-33
1-37
1-37
Additional Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-42
Files and Directories - UNIX Systems . . . . . . . . . . . . . . . . . . . 1-42
Files and Directories - Windows Systems . . . . . . . . . . . . . . . . 1-44
i
Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-46
Technical Support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-48
Creating C Language MEX-Files
2
C MEX-Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-3
The Components of a C MEX-File . . . . . . . . . . . . . . . . . . . . . . . . 2-3
Required Arguments to a MEX-File . . . . . . . . . . . . . . . . . . . . . . 2-5
Examples of C MEX-Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-7
A First Example — Passing a Scalar . . . . . . . . . . . . . . . . . . . . . 2-7
Passing Strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-11
Passing Two or More Inputs or Outputs . . . . . . . . . . . . . . . . . . 2-13
Passing Structures and Cell Arrays . . . . . . . . . . . . . . . . . . . . . 2-16
Handling Complex Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-21
Handling 8-,16-, and 32-Bit Data . . . . . . . . . . . . . . . . . . . . . . . 2-23
Manipulating Multidimensional Numerical Arrays . . . . . . . . 2-25
Handling Sparse Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-29
Calling Functions from C MEX-Files . . . . . . . . . . . . . . . . . . . . 2-33
Advanced Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Help Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Linking Multiple Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Workspace for MEX-File Functions . . . . . . . . . . . . . . . . . . . . .
Memory Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Using LAPACK and BLAS Functions . . . . . . . . . . . . . . . . . . . .
2-37
2-37
2-37
2-37
2-38
2-40
Debugging C Language MEX-Files . . . . . . . . . . . . . . . . . . . . . 2-47
Debugging on UNIX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-47
Debugging on Windows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-48
ii
Contents
Creating Fortran MEX-Files
3
Fortran MEX-Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-3
The Components of a Fortran MEX-File . . . . . . . . . . . . . . . . . . 3-3
The %val Construct . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-8
Examples of Fortran MEX-Files . . . . . . . . . . . . . . . . . . . . . . . . 3-9
A First Example — Passing a Scalar . . . . . . . . . . . . . . . . . . . . 3-10
Passing Strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-12
Passing Arrays of Strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-14
Passing Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-17
Passing Two or More Inputs or Outputs . . . . . . . . . . . . . . . . . . 3-19
Handling Complex Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-22
Dynamically Allocating Memory . . . . . . . . . . . . . . . . . . . . . . . . 3-25
Handling Sparse Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-28
Calling Functions from Fortran MEX-Files . . . . . . . . . . . . . . . 3-32
Advanced Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Help Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Linking Multiple Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Workspace for MEX-File Functions . . . . . . . . . . . . . . . . . . . . .
Memory Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3-36
3-36
3-36
3-36
3-37
Debugging Fortran Language MEX-Files . . . . . . . . . . . . . . . 3-38
Debugging on UNIX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-38
Debugging on Windows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-39
Calling MATLAB from C and Fortran Programs
4
Using the MATLAB Engine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-3
The Engine Library . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-3
GUI-Intensive Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-5
Examples of Calling Engine Functions . . . . . . . . . . . . . . . . . . 4-6
Calling MATLAB From a C Application . . . . . . . . . . . . . . . . . . . 4-6
iii
Calling MATLAB From a Fortran Application . . . . . . . . . . . . 4-11
Attaching to an Existing MATLAB Session . . . . . . . . . . . . . . . 4-15
Compiling and Linking Engine Programs . . . . . . . . . . . . . .
Masking Floating-Point Exceptions . . . . . . . . . . . . . . . . . . . . .
Compiling and Linking on UNIX . . . . . . . . . . . . . . . . . . . . . . .
Compiling and Linking on Windows . . . . . . . . . . . . . . . . . . . . .
4-17
4-17
4-18
4-19
Calling Java from MATLAB
5
iv
Contents
Using Java from MATLAB: An Overview . . . . . . . . . . . . . . . .
Java Interface Is Integral to MATLAB . . . . . . . . . . . . . . . . . . . .
Benefits of the MATLAB Java Interface . . . . . . . . . . . . . . . . . .
Who Should Use the MATLAB Java Interface . . . . . . . . . . . . . .
To Learn More About Java Programming . . . . . . . . . . . . . . . . .
Platform Support for the Java Virtual Machine . . . . . . . . . . . .
5-3
5-3
5-3
5-3
5-3
5-4
Bringing Java Classes into MATLAB . . . . . . . . . . . . . . . . . . . .
Sources of Java Classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Defining New Java Classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Making Java Classes Available to MATLAB . . . . . . . . . . . . . . .
Loading Java Class Definitions . . . . . . . . . . . . . . . . . . . . . . . . . .
Simplifying Java Class Names . . . . . . . . . . . . . . . . . . . . . . . . . .
5-5
5-5
5-5
5-6
5-7
5-8
Creating and Using Java Objects . . . . . . . . . . . . . . . . . . . . . .
Constructing Java Objects . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Concatenating Java Objects . . . . . . . . . . . . . . . . . . . . . . . . . . .
Saving and Loading Java Objects to MAT-Files . . . . . . . . . . .
Finding the Public Data Fields of an Object . . . . . . . . . . . . . .
Accessing Private and Public Data . . . . . . . . . . . . . . . . . . . . . .
Determining the Class of an Object . . . . . . . . . . . . . . . . . . . . .
5-10
5-10
5-12
5-14
5-15
5-16
5-17
Invoking Methods on Java Objects . . . . . . . . . . . . . . . . . . . .
Using Java and MATLAB Calling Syntax . . . . . . . . . . . . . . . .
Invoking Static Methods on Java Classes . . . . . . . . . . . . . . . .
Obtaining Information About Methods . . . . . . . . . . . . . . . . . . .
5-19
5-19
5-21
5-22
Java Methods That Affect MATLAB Commands . . . . . . . . . . . 5-26
How MATLAB Handles Undefined Methods . . . . . . . . . . . . . . 5-27
How MATLAB Handles Java Exceptions . . . . . . . . . . . . . . . . . 5-28
Working with Java Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . .
How MATLAB Represents the Java Array . . . . . . . . . . . . . . .
Creating an Array of Objects Within MATLAB . . . . . . . . . . . .
Accessing Elements of a Java Array . . . . . . . . . . . . . . . . . . . . .
Assigning to a Java Array . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Concatenating Java Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Creating a New Array Reference . . . . . . . . . . . . . . . . . . . . . . .
Creating a Copy of a Java Array . . . . . . . . . . . . . . . . . . . . . . . .
5-29
5-30
5-34
5-36
5-39
5-43
5-44
5-45
Passing Data to a Java Method . . . . . . . . . . . . . . . . . . . . . . . .
Conversion of MATLAB Argument Data . . . . . . . . . . . . . . . . .
Passing Built-In Data Types . . . . . . . . . . . . . . . . . . . . . . . . . . .
Passing String Arguments . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Passing Java Objects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Other Data Conversion Topics . . . . . . . . . . . . . . . . . . . . . . . . .
Passing Data to Overloaded Methods . . . . . . . . . . . . . . . . . . . .
5-46
5-46
5-48
5-49
5-50
5-53
5-54
Handling Data Returned from a Java Method . . . . . . . . . .
Conversion of Java Return Data . . . . . . . . . . . . . . . . . . . . . . . .
Built-In Data Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Java Objects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Converting Objects to MATLAB Data Types . . . . . . . . . . . . . .
5-56
5-56
5-57
5-57
5-57
Introduction to Programming Examples . . . . . . . . . . . . . . . 5-62
Example – Reading a URL . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-63
Description of URLdemo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-63
Running the Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-64
Example – Finding an Internet Protocol Address . . . . . . . . 5-66
Description of resolveip . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-66
Running the Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-67
v
Example – Communicating Through a Serial Port . . . . . . . 5-68
Description of Serial Example . . . . . . . . . . . . . . . . . . . . . . . . . . 5-69
Running the serialexample Program . . . . . . . . . . . . . . . . . . . . 5-72
Example – Creating and Using a Phone Book . . . . . . . . . . .
Description of Function phonebook . . . . . . . . . . . . . . . . . . . . . .
Description of Function pb_lookup . . . . . . . . . . . . . . . . . . . . . .
Description of Function pb_add . . . . . . . . . . . . . . . . . . . . . . . . .
Description of Function pb_remove . . . . . . . . . . . . . . . . . . . . . .
Description of Function pb_change . . . . . . . . . . . . . . . . . . . . . .
Description of Function pb_listall . . . . . . . . . . . . . . . . . . . . . . .
Description of Function pb_display . . . . . . . . . . . . . . . . . . . . . .
Description of Function pb_keyfilter . . . . . . . . . . . . . . . . . . . . .
Running the phonebook Program . . . . . . . . . . . . . . . . . . . . . . .
5-73
5-74
5-78
5-79
5-80
5-80
5-82
5-82
5-83
5-83
Importing and Exporting Data
6
vi
Contents
Using MAT-Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Importing Data to MATLAB . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Exporting Data from MATLAB . . . . . . . . . . . . . . . . . . . . . . . . . .
Exchanging Data Files Between Platforms . . . . . . . . . . . . . . . .
Reading and Writing MAT-Files . . . . . . . . . . . . . . . . . . . . . . . . .
Finding Associated Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6-3
6-3
6-4
6-5
6-6
6-8
Examples of MAT-Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Creating a MAT-File in C . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Reading a MAT-File in C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Creating a MAT-File in Fortran . . . . . . . . . . . . . . . . . . . . . . . .
Reading a MAT-File in Fortran . . . . . . . . . . . . . . . . . . . . . . . . .
6-11
6-11
6-16
6-21
6-24
Compiling and Linking MAT-File Programs . . . . . . . . . . . .
Masking Floating Point Exceptions . . . . . . . . . . . . . . . . . . . . .
Compiling and Linking on UNIX . . . . . . . . . . . . . . . . . . . . . . .
Compiling and Linking on Windows . . . . . . . . . . . . . . . . . . . . .
6-28
6-28
6-29
6-30
ActiveX and DDE Support
7
Introducing MATLAB ActiveX Integration . . . . . . . . . . . . . . 7-3
ActiveX Concepts and Terminology . . . . . . . . . . . . . . . . . . . . . . 7-3
MATLAB ActiveX Support Overview . . . . . . . . . . . . . . . . . . . . . 7-4
MATLAB ActiveX Client Support . . . . . . . . . . . . . . . . . . . . . . . 7-6
Using ActiveX Objects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-6
Writing Event Handlers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-8
Additional ActiveX Client Information . . . . . . . . . . . . . . . . .
Releasing Interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Using ActiveX Collections . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Converting Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Using MATLAB As a DCOM Server Client . . . . . . . . . . . . . . .
MATLAB ActiveX Support Limitations . . . . . . . . . . . . . . . . . .
MATLAB Sample Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Using MATLAB As an Automation Client . . . . . . . . . . . . . . . .
7-10
7-10
7-10
7-11
7-12
7-13
7-13
7-13
MATLAB ActiveX Automation Server Support . . . . . . . . . . 7-16
MATLAB ActiveX Automation Methods . . . . . . . . . . . . . . . . . . 7-17
MATLAB ActiveX Automation Properties . . . . . . . . . . . . . . . . 7-20
Additional ActiveX Server Information . . . . . . . . . . . . . . . .
Launching the MATLAB ActiveX Server . . . . . . . . . . . . . . . . .
Specifying a Shared or Dedicated Server . . . . . . . . . . . . . . . . .
Using MATLAB As a DCOM Server . . . . . . . . . . . . . . . . . . . . .
7-21
7-21
7-21
7-22
Dynamic Data Exchange (DDE) . . . . . . . . . . . . . . . . . . . . . . .
DDE Concepts and Terminology . . . . . . . . . . . . . . . . . . . . . . . .
Accessing MATLAB As a Server . . . . . . . . . . . . . . . . . . . . . . . .
The DDE Name Hierarchy . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Example: Using Visual Basic and the MATLAB DDE Server
Using MATLAB As a Client . . . . . . . . . . . . . . . . . . . . . . . . . . .
DDE Advisory Links . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7-23
7-23
7-25
7-26
7-29
7-31
7-32
vii
Serial Port I/O
8
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
What Is MATLAB’s Serial Port Interface? . . . . . . . . . . . . . . . . .
Supported Serial Port Interface Standards . . . . . . . . . . . . . . . .
Supported Platforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Using the Examples with Your Device . . . . . . . . . . . . . . . . . . . .
8-2
8-2
8-2
8-2
8-3
Overview of the Serial Port . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-4
What Is Serial Communication? . . . . . . . . . . . . . . . . . . . . . . . . . 8-4
The Serial Port Interface Standard . . . . . . . . . . . . . . . . . . . . . . 8-4
Connecting Two Devices with a Serial Cable . . . . . . . . . . . . . . . 8-5
Serial Port Signals and Pin Assignments . . . . . . . . . . . . . . . . . . 8-6
Serial Data Format . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-11
Finding Serial Port Information for Your Platform . . . . . . . . . 8-15
Selected Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-17
Getting Started with Serial I/O . . . . . . . . . . . . . . . . . . . . . . . .
Example: Getting Started . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The Serial Port Session . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Configuring and Returning Properties . . . . . . . . . . . . . . . . . . .
8-18
8-18
8-19
8-20
Creating a Serial Port Object . . . . . . . . . . . . . . . . . . . . . . . . .
Configuring Properties During Object Creation . . . . . . . . . . .
The Serial Port Object Display . . . . . . . . . . . . . . . . . . . . . . . . .
Creating an Array of Serial Port Objects . . . . . . . . . . . . . . . . .
8-24
8-25
8-25
8-26
Connecting to the Device . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-27
Configuring Communication Settings . . . . . . . . . . . . . . . . . . 8-28
Writing and Reading Data . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Example: Introduction to Writing and Reading Data . . . . . . .
Controlling Access to the MATLAB Command Line . . . . . . . .
Writing Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Reading Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Example: Writing and Reading Text Data . . . . . . . . . . . . . . . .
Example: Parsing Input Data Using strread . . . . . . . . . . . . . .
Example: Reading Binary Data . . . . . . . . . . . . . . . . . . . . . . . . .
viii Contents
8-29
8-29
8-29
8-31
8-36
8-42
8-44
8-45
Events and Callbacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Example: Introduction to Events and Callbacks . . . . . . . . . . .
Event Types and Callback Properties . . . . . . . . . . . . . . . . . . . .
Storing Event Information . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Creating and Executing Callback Functions . . . . . . . . . . . . . .
Enabling Callback Functions After They Error . . . . . . . . . . . .
Example: Using Events and Callbacks . . . . . . . . . . . . . . . . . . .
8-48
8-48
8-49
8-50
8-52
8-53
8-54
Using Control Pins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-56
Signaling the Presence of Connected Devices . . . . . . . . . . . . . 8-56
Controlling the Flow of Data: Handshaking . . . . . . . . . . . . . . 8-59
Debugging: Recording Information to Disk . . . . . . . . . . . . .
Example: Introduction to Recording Information . . . . . . . . . .
Creating Multiple Record Files . . . . . . . . . . . . . . . . . . . . . . . . .
Specifying a Filename . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The Record File Format . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Example: Recording Information to Disk . . . . . . . . . . . . . . . . .
8-62
8-62
8-63
8-63
8-64
8-65
Saving and Loading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-68
Using Serial Port Objects on Different Platforms . . . . . . . . . . 8-68
Disconnecting and Cleaning Up . . . . . . . . . . . . . . . . . . . . . . . 8-69
Disconnecting a Serial Port Object . . . . . . . . . . . . . . . . . . . . . . 8-69
Cleaning Up the MATLAB Environment . . . . . . . . . . . . . . . . . 8-69
Property Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-70
The Property Reference Page Format . . . . . . . . . . . . . . . . . . . . 8-70
Serial Port Object Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-71
ix
x
Contents
1
Calling C and Fortran
Programs from MATLAB
Introducing MEX-Files . . . . . . . . . . . . . . . 1-3
Using MEX-Files . . . . . . . . . . . . . . . . . . . 1-3
The Distinction Between mx and mex Prefixes . . . . . . . 1-4
MATLAB Data . . .
The MATLAB Array .
Data Storage . . . .
Data Types in MATLAB
Using Data Types . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Building MEX-Files . . . . . . . .
Compiler Requirements . . . . . . .
Testing Your Configuration on UNIX .
Testing Your Configuration on Windows
Specifying an Options File . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 1-11
. 1-11
. 1-12
. 1-14
. 1-17
Custom Building MEX-Files .
Who Should Read This Chapter .
MEX Script Switches . . . . .
Default Options File on UNIX .
Default Options File on Windows
Custom Building on UNIX . . .
Custom Building on Windows .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 1-20
. 1-20
. 1-20
. 1-22
. 1-23
. 1-24
. 1-26
Troubleshooting . . . . . . . . . .
Configuration Issues . . . . . . . . .
Understanding MEX-File Problems . . .
Compiler and Platform-Specific Issues . .
Memory Management Compatibility Issues
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 1-32
. 1-32
. 1-33
. 1-37
. 1-37
Additional Information . . . . . .
Files and Directories - UNIX Systems .
Files and Directories - Windows Systems
Examples . . . . . . . . . . . . .
Technical Support . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 1-42
. 1-42
. 1-44
. 1-46
. 1-48
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1-6
1-6
1-6
1-7
1-9
1
Calling C and Fortran Programs from MATLAB
Although MATLAB® is a complete, self-contained environment for
programming and manipulating data, it is often useful to interact with data
and programs external to the MATLAB environment. MATLAB provides an
interface to external programs written in the C and Fortran languages.
This section discusses the following topics:
• “Introducing MEX-Files”
• “MATLAB Data”
• “Building MEX-Files”
• “Custom Building MEX-Files”
• “Troubleshooting”
• “Additional Information”
For additional information and support in building your applications, see the
section, “Additional Information” on page 1-42.
Note In platform independent discussions that refer to directory paths, this
book uses the UNIX convention. For example, a general reference to the mex
directory is <matlab>/extern/examples/mex.
1-2
Introducing MEX-Files
Introducing MEX-Files
You can call your own C or Fortran subroutines from MATLAB as if they were
built-in functions. MATLAB callable C and Fortran programs are referred to
as MEX-files. MEX-files are dynamically linked subroutines that the MATLAB
interpreter can automatically load and execute.
MEX-files have several applications:
• Large pre-existing C and Fortran programs can be called from MATLAB
without having to be rewritten as M-files.
• Bottleneck computations (usually for-loops) that do not run fast enough in
MATLAB can be recoded in C or Fortran for efficiency.
MEX-files are not appropriate for all applications. MATLAB is a
high-productivity system whose specialty is eliminating time-consuming,
low-level programming in compiled languages like Fortran or C. In general,
most programming should be done in MATLAB. Don’t use the MEX facility
unless your application requires it.
Using MEX-Files
MEX-files are subroutines produced from C or Fortran source code. They
behave just like M-files and built-in functions. While M-files have a
platform-independent extension, .m, MATLAB identifies MEX-files by
platform-specific extensions. This table lists the platform-specific extensions
for MEX-files.
Table 1-1: MEX-File Extensions
Platform
MEX-File Extension
Alpha
mexaxp
HP, version 10.20
mexhp7
HP, version 11.x
mexhpux
IBM RS/6000
mexrs6
Linux
mexglx
SGI, SGI64
mexsg
1-3
1
Calling C and Fortran Programs from MATLAB
Table 1-1: MEX-File Extensions (Continued)
Platform
MEX-File Extension
Solaris
mexsol
Windows
dll
You can call MEX-files exactly as you would call any M-function. For example,
a MEX-file called conv2.mex on your disk in the MATLAB datafun toolbox
directory performs a 2-D convolution of matrices. conv2.m only contains the
help text documentation. If you invoke the function conv2 from inside
MATLAB, the interpreter looks through the list of directories on MATLAB’s
search path. It scans each directory looking for the first occurrence of a file
named conv2 with the corresponding filename extension from the table or .m.
When it finds one, it loads the file and executes it. MEX-files take precedence
over M-files when like-named files exist in the same directory. However, help
text documentation is still read from the .m file.
The Distinction Between mx and mex Prefixes
Routines in the API that are prefixed with mx allow you to create, access,
manipulate, and destroy mxArrays. Routines prefixed with mex perform
operations back in the MATLAB environment.
mx Routines
The array access and creation library provides a set of array access and
creation routines for manipulating MATLAB arrays. These subroutines, which
are fully documented in the online API reference pages, always start with the
prefix mx. For example, mxGetPi retrieves the pointer to the imaginary data
inside the array.
Although most of the routines in the array access and creation library let you
manipulate the MATLAB array, there are two exceptions — the IEEE routines
and memory management routines. For example, mxGetNaN returns a double,
not an mxArray.
mex Routines
Routines that begin with the mex prefix perform operations back in the
MATLAB environment. For example, the mexEvalString routine evaluates a
string in the MATLAB workspace.
1-4
Introducing MEX-Files
Note mex routines are only available in MEX-functions.
1-5
1
Calling C and Fortran Programs from MATLAB
MATLAB Data
Before you can program MEX-files, you must understand how MATLAB
represents the many data types it supports. This section discusses the
following topics:
• “The MATLAB Array”
• “Data Storage”
• “Data Types in MATLAB”
• “Using Data Types”
The MATLAB Array
The MATLAB language works with only a single object type: the MATLAB
array. All MATLAB variables, including scalars, vectors, matrices, strings, cell
arrays, structures, and objects are stored as MATLAB arrays. In C, the
MATLAB array is declared to be of type mxArray. The mxArray structure
contains, among other things:
• Its type
• Its dimensions
• The data associated with this array
• If numeric, whether the variable is real or complex
• If sparse, its indices and nonzero maximum elements
• If a structure or object, the number of fields and field names
Data Storage
All MATLAB data is stored columnwise, which is how Fortran stores matrices.
MATLAB uses this convention because it was originally written in Fortran. For
example, given the matrix
a=['house'; 'floor'; 'porch']
a =
house
floor
porch
1-6
MATLAB Data
its dimensions are
size(a)
ans =
3
5
and its data is stored as
h
f
p
o
l
o
u
o
r
s
o
c
e
r
h
Data Types in MATLAB
Complex Double-Precision Matrices
The most common data type in MATLAB is the complex double-precision,
nonsparse matrix. These matrices are of type double and have dimensions
m-by-n, where m is the number of rows and n is the number of columns. The data
is stored as two vectors of double-precision numbers – one contains the real
data and one contains the imaginary data. The pointers to this data are
referred to as pr (pointer to real data) and pi (pointer to imaginary data),
respectively. A real-only, double-precision matrix is one whose pi is NULL.
Numeric Matrices
MATLAB also supports other types of numeric matrices. These are
single-precision floating-point and 8-, 16-, and 32-bit integers, both signed and
unsigned. The data is stored in two vectors in the same manner as
double-precision matrices.
MATLAB Strings
MATLAB strings are of type char and are stored the same way as unsigned
16-bit integers except there is no imaginary data component. Each character in
the string is stored as 16-bit ASCII Unicode. Unlike C, MATLAB strings are
not null terminated.
Sparse Matrices
Sparse matrices have a different storage convention than full matrices in
MATLAB. The parameters pr and pi are still arrays of double-precision
numbers, but there are three additional parameters, nzmax, ir, and jc:
1-7
1
Calling C and Fortran Programs from MATLAB
• nzmax is an integer that contains the length of ir, pr, and, if it exists, pi. It
is the maximum possible number of nonzero elements in the sparse matrix.
• ir points to an integer array of length nzmax containing the row indices of the
corresponding elements in pr and pi.
• jc points to an integer array of length N+1 that contains column index
information. For j, in the range 0 ≤ j ≤ N-1, jc[j] is the index in ir and pr
(and pi if it exists) of the first nonzero entry in the jth column and
jc[j+1] - 1 index of the last nonzero entry. As a result, jc[N] is also equal
to nnz, the number of nonzero entries in the matrix. If nnz is less than nzmax,
then more nonzero entries can be inserted in the array without allocating
additional storage.
Cell Arrays
Cell arrays are a collection of MATLAB arrays where each mxArray is referred
to as a cell. This allows MATLAB arrays of different types to be stored together.
Cell arrays are stored in a similar manner to numeric matrices, except the data
portion contains a single vector of pointers to mxArrays. Members of this vector
are called cells. Each cell can be of any supported data type, even another cell
array.
Structures
A 1-by-1 structure is stored in the same manner as a 1-by-n cell array where n
is the number of fields in the structure. Members of the data vector are called
fields. Each field is associated with a name stored in the mxArray.
Objects
Objects are stored and accessed the same way as structures. In MATLAB,
objects are named structures with registered methods. Outside MATLAB, an
object is a structure that contains storage for an additional classname that
identifies the name of the object.
Multidimensional Arrays
MATLAB arrays of any type can be multidimensional. A vector of integers is
stored where each element is the size of the corresponding dimension. The
storage of the data is the same as matrices.
1-8
MATLAB Data
Logical Arrays
Any noncomplex numeric or sparse array can be flagged as logical. The storage
for a logical array is the same as the storage for a nonlogical array.
Empty Arrays
MATLAB arrays of any type can be empty. An empty mxArray is one with at
least one dimension equal to zero. For example, a double-precision mxArray of
type double, where m and n equal 0 and pr is NULL, is an empty array.
Using Data Types
The six fundamental data types in MATLAB are double, char, sparse, uint8,
cell, and struct. You can write MEX-files, MAT-file applications, and engine
applications in C that accept any data type supported by MATLAB. In Fortran,
only the creation of double-precision n-by-m arrays and strings are supported.
You can treat C and Fortran MEX-files, once compiled, exactly like
M-functions.
The explore Example
There is an example MEX-file included with MATLAB, called explore, that
identifies the data type of an input variable. The source file for this example is
in the <matlab>/extern/examples/mex directory, where <matlab> represents
the top-level directory where MATLAB is installed on your system. For
example, typing
cd([matlabroot '/extern/examples/mex']);
x = 2;
explore(x);
produces this result
-----------------------------------------------Name: x
Dimensions: 1x1
Class Name: double
-----------------------------------------------(1,1) = 2
1-9
1
Calling C and Fortran Programs from MATLAB
explore accepts any data type. Try using explore with these examples.
explore([1 2 3 4 5])
explore 1 2 3 4 5
explore({1 2 3 4 5})
explore(int8([1 2 3 4 5]))
explore {1 2 3 4 5}
explore(sparse(eye(5)))
explore(struct('name', 'Joe Jones', 'ext', 7332))
explore(1, 2, 3, 4, 5)
1-10
Building MEX-Files
Building MEX-Files
This section covers the following topics:
• “Compiler Requirements”
• “Testing Your Configuration on UNIX”
• “Testing Your Configuration on Windows”
• “Specifying an Options File”
Compiler Requirements
Your installed version of MATLAB contains all the tools you need to work with
the API. MATLAB includes a C compiler for the PC called Lcc, but does not
include a Fortran compiler. If you choose to use your own C compiler, it must
be an ANSI C compiler. Also, if you are working on a Microsoft Windows
platform, your compiler must be able to create 32-bit windows dynamically
linked libraries (DLLs).
MATLAB supports many compilers and provides preconfigured files, called
options files, designed specifically for these compilers. The Options Files table
lists all supported compilers and their corresponding options files. The purpose
of supporting this large collection of compilers is to provide you with the
flexibility to use the tool of your choice. However, in many cases, you simply can
use the provided Lcc compiler with your C code to produce your applications.
The MathWorks also maintains a list of compilers supported by MATLAB at
the following location on the web: http://www.mathworks.com/support/
tech-notes/v5/1600/1601.shtml.
Note The MathWorks provides an option (setup) for the mex script that lets
you easily choose or switch your compiler.
The following sections contain configuration information for creating
MEX-files on UNIX and Windows systems. More detailed information about
the mex script is provided in “Custom Building MEX-Files” on page 1-20. In
addition, there is a section on “Troubleshooting” on page 1-32, if you are having
difficulties creating MEX-files.
1-11
1
Calling C and Fortran Programs from MATLAB
Testing Your Configuration on UNIX
The quickest way to check if your system is set up properly to create MEX-files
is by trying the actual process. There is C source code for an example,
yprime.c, and its Fortran counterpart, yprimef.F and yprimefg.F, included in
the <matlab>/extern/examples/mex directory, where <matlab> represents
the top-level directory where MATLAB is installed on your system.
To compile and link the example source files, yprime.c or yprimef.F and
yprimefg.F, on UNIX, you must first copy the file(s) to a local directory, and
then change directory (cd) to that local directory.
At the MATLAB prompt, type
mex yprime.c
This uses the system compiler to create the MEX-file called yprime with the
appropriate extension for your system.
You can now call yprime as if it were an M-function.
yprime(1,1:4)
ans =
2.0000
8.9685
4.0000
-1.0947
To try the Fortran version of the sample program with your Fortran compiler,
at the MATLAB prompt, type
mex yprimef.F yprimefg.F
In addition to running the mex script from the MATLAB prompt, you can also
run the script from the system prompt.
Selecting a Compiler
To change your default compiler, you select a different options file. You can do
this anytime by using the command
mex -setup
Using the 'mex -setup' command selects an options file that is
placed in ~/matlab and used by default for 'mex'. An options
file in the current working directory or specified on the
command line overrides the default options file in ~/matlab.
1-12
Building MEX-Files
Options files control which compiler to use, the compiler and
link command options, and the runtime libraries to link
against.
To override the default options file, use the 'mex -f' command
(see 'mex -help' for more information).
The options files available for mex are:
1: <matlab>/bin/gccopts.sh :
Template Options file for building gcc MEXfiles
2: <matlab>/bin/mexopts.sh :
Template Options file for building MEXfiles using the
system ANSI compiler
Enter the number of the options file to use as your default options
file:
Select the proper options file for your system by entering its number and
pressing Return. If an options file doesn’t exist in your MATLAB directory, the
system displays a message stating that the options file is being copied to your
user-specific matlab directory. If an options file already exists in your matlab
directory, the system prompts you to overwrite it.
Note The setup option creates a user-specific matlab directory in your
individual home directory and copies the appropriate options file to the
directory. (If the directory already exists, a new one is not created.) This
matlab directory is used for your individual options files only; each user can
have his or her own default options files (other MATLAB products may place
options files in this directory). Do not confuse these user-specific matlab
directories with the system matlab directory, where MATLAB is installed. To
see the name of this directory on your machine, use the MATLAB command
prefdir.
Using the setup option resets your default compiler so that the new compiler is
used every time you use the mex script.
1-13
1
Calling C and Fortran Programs from MATLAB
Testing Your Configuration on Windows
Before you can create MEX-files on the Windows platform, you must configure
the default options file, mexopts.bat, for your compiler. The switch, setup,
provides an easy way for you to configure the default options file. To configure
or change the options file at anytime, run
mex -setup
from either the MATLAB or DOS command prompt.
Selecting a Compiler
MATLAB includes a C compiler, Lcc, that you can use to create C MEX-files.
The mex script will use the Lcc compiler automatically if you do not have a C or
C++ compiler of your own already installed on your system and you try to
compile a C MEX-file. Naturally, if you need to compile Fortran programs, you
must supply your own supported Fortran compiler.
The mex script uses the filename extension to determine the type of compiler to
use for creating your MEX-files. For example,
mex test1.f
would use your Fortran compiler and
mex test2.c
would use your C compiler.
On Systems without a Compiler. If you do not have your own C or C++ compiler on
your system, the mex utility automatically configures itself for the included Lcc
compiler. So, to create a C MEX-file on these systems, you can simply enter
mex filename.c
This simple method of creating MEX-files works for the majority of users.
If using the included Lcc compiler satisfies your needs, you can skip ahead in
this section to “Building the MEX-File on Windows” on page 1-16.
On Systems with a Compiler. On systems where there is a C, C++, or Fortran
compiler, you can select which compiler you want to use. Once you choose your
compiler, that compiler becomes your default compiler and you no longer have
1-14
Building MEX-Files
to select one when you compile MEX-files. To select a compiler or change to
existing default compiler, use mex –setup.
This example shows the process of setting your default compiler to the
Microsoft Visual C++ Version 6.0 compiler.
mex -setup
Please choose your compiler for building external interface (MEX)
files.
Would you like mex to locate installed compilers [y]/n? n
Select a compiler:
[1] Borland C++Builder version 5.0
[2] Borland C++Builder version 4.0
[3] Borland C++Builder version 3.0
[4] Borland C/C++ version 5.02
[5] Borland C/C++ version 5.0
[6] Compaq Visual Fortran version 6.1
[7] Digital Visual Fortran version 5.0
[8] Lcc C version 2.4
[9] Microsoft Visual C/C++ version 6.0
[10] Microsoft Visual C/C++ version 5.0
[11] WATCOM C/C++ version 11
[12] WATCOM C/C++ version 10.6
[0] None
Compiler: 9
Your machine has a Microsoft Visual C/C++ compiler located at
D:\Applications\Microsoft Visual Studio. Do you want to use this
compiler [y]/n? y
Please verify your choices:
Compiler: Microsoft Visual C/C++ 6.0
Location: D:\Applications\Microsoft Visual Studio
Are these correct?([y]/n): y
1-15
1
Calling C and Fortran Programs from MATLAB
The default options file:
"C:\WINNT\Profiles\username\ApplicationData\MathWorks\MATLAB\R12
\mexopts.bat" is being updated from ...
If the specified compiler cannot be located, you are given the message:
The default location for compiler-name is directory-name,
but that directory does not exist on this machine.
Use directory-name anyway [y]/n?
Using the setup option sets your default compiler so that the new compiler is
used every time you use the mex script.
Building the MEX-File on Windows
There is example C source code, yprime.c, and its Fortran counterpart,
yprimef.f and yprimefg.f, included in the <matlab>\extern\examples\mex
directory, where <matlab> represents the top-level directory where MATLAB
is installed on your system.
To compile and link the example source file on Windows, at the MATLAB
prompt, type
cd([matlabroot '\extern\examples\mex'])
mex yprime.c
This should create the MEX-file called yprime with the .DLL extension, which
corresponds to the Windows platform.
You can now call yprime as if it were an M-function.
yprime(1,1:4)
ans =
2.0000
8.9685
4.0000
-1.0947
To try the Fortran version of the sample program with your Fortran compiler,
switch to your Fortran compiler using mex -setup. Then, at the MATLAB
prompt, type
cd([matlabroot '\extern\examples\mex'])
mex yprimef.f yprimefg.f
1-16
Building MEX-Files
In addition to running the mex script from the MATLAB prompt, you can also
run the script from the system prompt.
Specifying an Options File
You can use the -f option to specify an options file on either UNIX or Windows.
To use the -f option, at the MATLAB prompt type
mex filename -f <optionsfile>
and specify the name of the options file along with its pathname. The Options
Files table, below, contains a list of the options files included with MATLAB.
There are several situations when it may be necessary to specify an options file
every time you use the mex script. These include:
• (Windows and UNIX) You want to use a different compiler (and not use the
-setup option), or you want to compile MAT or engine stand-alone programs.
• (UNIX) You do not want to use the system C compiler.
Preconfigured Options Files
MATLAB includes some preconfigured options files that you can use with
particular compilers. The Options Files table lists the compilers whose options
files are included with this release of MATLAB.
Table 1-2: Options Files
Platform
Compiler
Options File
Windows
Borland C++, Version 5.0 & 5.2
bccopts.bat
Borland C++Builder 3.0 (Borland
C++, Version 5.3)
bcc53opts.bat
Borland C++Builder 4.0 (Borland
C++, Version 5.4)
bcc54opts.bat
Borland C++Builder 5.0 (Borland
C++, Version 5.5)
bcc55opts.bat
Lcc C Compiler, bundled with
MATLAB
lccopts.bat
1-17
1
Calling C and Fortran Programs from MATLAB
Table 1-2: Options Files (Continued)
Platform
1-18
Compiler
Options File
Microsoft C/C++, Version 5.0
msvc50opts.bat
Microsoft C/C++, Version 6.0
msvc60opts.bat
Watcom C/C++, Version 10.6
watcopts.bat
Watcom C/C++, Version 11
wat11copts.bat
DIGITAL Visual Fortran,
Version 5.0
df50opts.bat
Compaq Visual Fortran,
Version 6.1
df60opts.bat
Borland C, Version 5.0 & 5.2, for
Engine and MAT stand-alone
programs
bccengmatopts.bat
Borland C, Version 5.3, for Engine
and MAT stand-alone programs
bcc53engmatopts.bat
Borland C, Version 5.4, for Engine
and MAT stand-alone programs
bcc54engmatopts.bat
Borland C, Version 5.5, for Engine
and MAT stand-alone programs
bcc55engmatopts.bat
Lcc C compiler for Engine and
MAT stand-alone programs,
lccengmatopts.bat
Microsoft Visual C for Engine and
MAT stand-alone programs,
Version 5.0
msvc50engmatopts.bat
Microsoft Visual C for Engine and
MAT stand-alone programs,
Version 6.0
msvc60engmatopts.bat
Building MEX-Files
Table 1-2: Options Files (Continued)
Platform
UNIX
Compiler
Options File
Watcom C for Engine and MAT
stand-alone programs,
Version 10.6
watengmatopts.bat
Watcom C for Engine and MAT
stand-alone programs,
Version 11
wat11engmatopts.bat
DIGITAL Visual Fortran for MAT
stand-alone programs,
Version 5.0
df50engmatopts.bat
Compaq Visual Fortran for MAT
stand-alone programs,
Version 6.1
df60engmatopts.bat
System ANSI Compiler
mexopts.sh
GCC
gccopts.sh
System C++ Compiler
cxxopts.sh
System ANSI Compiler for
Engine stand-alone programs
engopts.sh
System ANSI Compiler for MAT
stand-alone programs
matopts.sh
An up-to-date list of options files is available from our FTP server, ftp://
ftp.mathworks.com/pub/tech-support/docexamples/apiguide/R12/bin.
For a list of all the compilers supported by MATLAB, access the MathWorks
Technical Support Web site at http://www.mathworks.com/support.
Note The next section, “Custom Building MEX-Files” on page 1-20, contains
specific information on how to modify options files for particular systems.
1-19
1
Calling C and Fortran Programs from MATLAB
Custom Building MEX-Files
This section discusses in detail the process that the MEX-file build script uses.
It covers the following topics:
• “Who Should Read This Chapter”
• “MEX Script Switches”
• “Default Options File on UNIX”
• “Default Options File on Windows”
• “Custom Building on UNIX”
• “Custom Building on Windows”
Who Should Read This Chapter
In general, the defaults that come with MATLAB should be sufficient for
building most MEX-files. There are reasons that you might need more detailed
information, such as:
• You want to use an Integrated Development Environment (IDE), rather than
the provided script, to build MEX-files.
• You want to create a new options file, for example, to use a compiler that is
not directly supported.
• You want to exercise more control over the build process than the script uses.
The script, in general, uses two stages (or three, for Microsoft Windows) to
build MEX-files. These are the compile stage and the link stage. In between
these two stages, Windows compilers must perform some additional steps to
prepare for linking (the prelink stage).
MEX Script Switches
The mex script has a set of switches (also called options) that you can use to
modify the link and compile stages. The MEX Script Switches table lists the
available switches and their uses. Each switch is available on both UNIX and
Windows unless otherwise noted.
For customizing the build process, you should modify the options file, which
contains the compiler-specific flags corresponding to the general compile,
prelink, and link steps required on your system. The options file consists of a
1-20
Custom Building MEX-Files
series of variable assignments; each variable represents a different logical
piece of the build process.
Table 1-3: MEX Script Switches
Switch
Function
@<rsp_file>
Include the contents of the text file <rsp_file> as
command line arguments to the mex script.
-argcheck
Perform argument checking on MATLAB API
functions (C functions only).
-c
Compile only; do not link.
-D<name>[#<def>]
Define C preprocessor macro <name> [as having
value <def>]. (Note: UNIX also allows
-D<name>[=<def>].)
-f <file>
Use <file> as the options file; <file> is a full
pathname if it is not in current directory.
-g
Build an executable with debugging symbols
included.
-h[elp]
Help; lists the switches and their functions.
-I<pathname>
Include <pathname> in the compiler include search
path.
-inline
Inlines matrix accessor functions (mx*). The
generated MEX-function may not be compatible
with future versions of MATLAB.
-l<file>
(UNIX) Link against library lib<file>.
-L<pathname>
(UNIX) Include <pathname> in the list of directories
to search for libraries.
1-21
1
Calling C and Fortran Programs from MATLAB
Table 1-3: MEX Script Switches (Continued)
Switch
Function
<name>#<def>
Override options file setting for variable <name>.
This option is equivalent to <ENV_VAR>#<val>,
which temporarily sets the environment variable
<ENV_VAR> to <val> for the duration of the call to
mex. <val> can refer to another environment
variable by prepending the name of the variable
with a $, e.g., COMPFLAGS#"$COMPFLAGS -myswitch".
<name>=<def>
(UNIX) Override options file setting for variable
<name>.
-O
Build an optimized executable.
-outdir <name>
Place all output files in directory <name>.
-output <name>
Create an executable named <name>. (An
appropriate executable extension is automatically
appended.)
-setup
Set up default options file. This switch should be the
only argument passed.
-U<name>
Undefine C preprocessor macro <name>.
-v
Verbose; print all compiler and linker settings.
-V4
Compile MATLAB 4-compatible MEX-file.
Default Options File on UNIX
The default MEX options file provided with MATLAB is located in <matlab>/
bin. The mex script searches for an options file called mexopts.sh in the
following order:
• The current directory
• $HOME/.matlab/R12
• <matlab>/bin
1-22
Custom Building MEX-Files
mex uses the first occurrence of the options file it finds. If no options file is
found, mex displays an error message. You can directly specify the name of the
options file using the -f switch.
For specific information on the default settings for the MATLAB supported
compilers, you can examine the options file in <matlab>/bin/mexopts.sh, or
you can invoke the mex script in verbose mode (-v). Verbose mode will print the
exact compiler options, prelink commands (if appropriate), and linker options
used in the build process for each compiler. “Custom Building on UNIX” on
page 1-24 gives an overview of the high-level build process.
Default Options File on Windows
The default MEX options file is placed in your user profile directory after you
configure your system by running mex -setup. The mex script searches for an
options file called mexopts.bat in the following order:
• The current directory
• The user profile directory. See the following section, “The User Profile
Directory”, for more information about this directory
• <matlab>\bin\win32\mexopts
mex uses the first occurrence of the options file it finds. If no options file is
found, mex searches your machine for a supported C compiler and
automatically configures itself to use that compiler. Also, during the
configuration process, it copies the compiler’s default options file to the user
profile directory. If multiple compilers are found, you are prompted to select
one.
For specific information on the default settings for the MATLAB supported
compilers, you can examine the options file, mexopts.bat, or you can invoke the
mex script in verbose mode (-v). Verbose mode will print the exact compiler
options, prelink commands, if appropriate, and linker options used in the build
process for each compiler. “Custom Building on Windows” on page 1-26 gives
an overview of the high-level build process.
The User Profile Directory
The Windows user profile directory is a directory that contains user-specific
information such as desktop appearance, recently used files, and Start menu
items. The mex and mbuild utilities store their respective options files,
mexopts.bat and compopts.bat, which are created during the –setup process,
1-23
1
Calling C and Fortran Programs from MATLAB
in a subdirectory of your user profile directory, named Application
Data\MathWorks\MATLAB. Under Windows NT and Windows 95/98 with user
profiles enabled, your user profile directory is
%windir%\Profiles\username. Under Windows 95/98 with user profiles
disabled, your user profile directory is %windir%. Under Windows 95/98, you
can determine whether or not user profiles are enabled by using the
Passwords control panel.
Custom Building on UNIX
On UNIX systems, there are two stages in MEX-file building: compiling and
linking.
Compile Stage
The compile stage must:
• Add <matlab>/extern/include to the list of directories in which to find
header files (-I<matlab>/extern/include)
• Define the preprocessor macro MATLAB_MEX_FILE (-DMATLAB_MEX_FILE)
• (C MEX-files only) Compile the source file, which contains version
information for the MEX-file, <matlab>/extern/src/mexversion.c
Link Stage
The link stage must:
• Instruct the linker to build a shared library
• Link all objects from compiled source files (including mexversion.c)
• (Fortran MEX-files only) Link in the precompiled versioning source file,
<matlab>/extern/lib/$Arch/version4.o
• Export the symbols mexFunction and mexVersion (these symbols represent
functions called by MATLAB)
For Fortran MEX-files, the symbols are all lower case and may have appended
underscores. For specific information, invoke the mex script in verbose mode
and examine the output.
1-24
Custom Building MEX-Files
Build Options
For customizing the build process, you should modify the options file. The
options file contains the compiler-specific flags corresponding to the general
steps outlined above. The options file consists of a series of variable
assignments; each variable represents a different logical piece of the build
process. The options files provided with MATLAB are located in <matlab>/bin.
The section, “Default Options File on UNIX” on page 1-22, describes how the
mex script looks for an options file.
To aid in providing flexibility, there are two sets of options in the options file
that can be turned on and off with switches to the mex script. These sets of
options correspond to building in debug mode and building in optimization
mode. They are represented by the variables DEBUGFLAGS and OPTIMFLAGS,
respectively, one pair for each driver that is invoked (CDEBUGFLAGS for the C
compiler, FDEBUGFLAGS for the Fortran compiler, and LDDEBUGFLAGS for the
linker; similarly for the OPTIMFLAGS).
• If you build in optimization mode (the default), the mex script will include the
OPTIMFLAGS options in the compile and link stages.
• If you build in debug mode, the mex script will include the DEBUGFLAGS
options in the compile and link stages, but will not include the OPTIMFLAGS
options.
• You can include both sets of options by specifying both the optimization and
debugging flags to the mex script (-O and -g, respectively).
Aside from these special variables, the mex options file defines the executable
invoked for each of the three modes (C compile, Fortran compile, link) and the
flags for each stage. You can also provide explicit lists of libraries that must be
linked in to all MEX-files containing source files of each language.
The variables can be summed up as follows.
Variable
C Compiler
Fortran Compiler
Linker
Executable
CC
FC
LD
Flags
CFLAGS
FFLAGS
LDFLAGS
Optimization
COPTIMFLAGS
FOPTIMFLAGS
LDOPTIMFLAGS
1-25
1
Calling C and Fortran Programs from MATLAB
Variable
C Compiler
Fortran Compiler
Linker
Debugging
CDEBUGFLAGS
FDEBUGFLAGS
LDDEBUGFLAGS
Additional libraries
CLIBS
FLIBS
----
For specifics on the default settings for these variables, you can:
• Examine the options file in <matlab>/bin/mexopts.sh (or the options file
you are using), or
• Invoke the mex script in verbose mode
Custom Building on Windows
There are three stages to MEX-file building for both C and Fortran on Windows
– compiling, prelinking, and linking.
Compile Stage
For the compile stage, a mex options file must:
• Set up paths to the compiler using the COMPILER (e.g., Watcom), PATH,
INCLUDE, and LIB environment variables. If your compiler always has the
environment variables set (e.g., in AUTOEXEC.BAT), you can remark them out
in the options file.
• Define the name of the compiler, using the COMPILER environment variable,
if needed.
• Define the compiler switches in the COMPFLAGS environment variable.
a The switch to create a DLL is required for MEX-files.
b For stand-alone programs, the switch to create an exe is required.
c
The -c switch (compile only; do not link) is recommended.
d The switch to specify 8-byte alignment.
e
Any other switch specific to the environment can be used.
• Define preprocessor macro, with -D, MATLAB_MEX_FILE is required.
• Set up optimizer switches and/or debug switches using OPTIMFLAGS and
DEBUGFLAGS. These are mutually exclusive: the OPTIMFLAGS are the default,
1-26
Custom Building MEX-Files
and the DEBUGFLAGS are used if you set the -g switch on the mex command
line.
Prelink Stage
The prelink stage dynamically creates import libraries to import the required
function into the MEX, MAT, or engine file. All MEX-files link against
MATLAB only. MAT stand-alone programs link against libmx.dll (array
access library) and libmat.dll (MAT-functions). Engine stand-alone programs
link against libmx.dll (array access library) and libeng.dll for engine
functions. MATLAB and each DLL have corresponding .def files of the same
names located in the <matlab>\extern\include directory.
Link Stage
Finally, for the link stage, a mex options file must:
• Define the name of the linker in the LINKER environment variable.
• Define the LINKFLAGS environment variable that must contain:
- The switch to create a DLL for MEX-files, or the switch to create an exe
for stand-alone programs.
- Export of the entry point to the MEX-file as mexFunction for C or
MEXFUNCTION@16 for DIGITAL Visual Fortran.
- The import library(s) created in the PRELINK_CMDS stage.
- Any other link switch specific to the compiler that can be used.
• Define the linking optimization switches and debugging switches in
LINKEROPTIMFLAGS and LINKDEBUGFLAGS. As in the compile stage, these two
are mutually exclusive: the default is optimization, and the -g switch
invokes the debug switches.
• Define the link-file identifier in the LINK_FILE environment variable, if
needed. For example, Watcom uses file to identify that the name following
is a file and not a command.
• Define the link-library identifier in the LINK_LIB environment variable, if
needed. For example, Watcom uses library to identify the name following is
a library and not a command.
• Optionally, set up an output identifier and name with the output switch in
the NAME_OUTPUT environment variable. The environment variable MEX_NAME
contains the name of the first program in the command line. This must be set
1-27
1
Calling C and Fortran Programs from MATLAB
for -output to work. If this environment is not set, the compiler default is to
use the name of the first program in the command line. Even if this is set, it
can be overridden by specifying the mex -output switch.
Linking DLLs to MEX-Files
To link a DLL to a MEX-file, list the DLL’s .lib file on the command line.
Versioning MEX-Files
The mex script can build your MEX-file with a resource file that contains
versioning and other essential information. The resource file is called
mexversion.rc and resides in the extern\include directory. To support
versioning, there are two new commands in the options files, RC_COMPILER and
RC_LINKER, to provide the resource compiler and linker commands. It is
assumed that:
• If a compiler command is given, the compiled resource will be linked into the
MEX-file using the standard link command.
• If a linker command is given, the resource file will be linked to the MEX-file
after it is built using that command.
Compiling MEX-Files with the Microsoft Visual C++ IDE
Note This section provides information on how to compile MEX-files in the
Microsoft Visual C++ (MSVC) IDE; it is not totally inclusive. This section
assumes that you know how to use the IDE. If you need more information on
using the MSVC IDE, refer to the corresponding Microsoft documentation.
To build MEX-files with the Microsoft Visual C++ integrated development
environment:
1 Create a project and insert your MEX source and mexversion.rc into it.
2 Create a .DEF file to export the MEX entry point. For example
LIBRARY MYFILE.DLL
EXPORTS mexFunction
<-- for a C MEX-file
or
EXPORTS MEXFUNCTION@16
1-28
<-- for a Fortran MEX-file
Custom Building MEX-Files
3 Add the .DEF file to the project.
4 Locate the .LIB files for the compiler version you are using under
matlabroot\extern\lib\win32\microsoft. For example, for version 6.0,
these files are in the msvc60 subdirectory.
5 From this directory, add libmx.lib, libmex.lib, and libmat.lib to the
library modules in the LINK settings option.
6 Add the MATLAB include directory, MATLAB\EXTERN\INCLUDE to the
include path in the Settings C/C++ Preprocessor option.
7 Add MATLAB_MEX_FILE to the C/C++ Preprocessor option by selecting
Settings from the Build menu, selecting C/C++, and then typing
,MATLAB_MEX_FILE after the last entry in the Preprocessor definitions
field.
8 To debug the MEX-file using the IDE, put MATLAB.EXE in the Settings
Debug option as the Executable for debug session.
If you are using a compiler other than the Microsoft Visual C/C++ compiler, the
process for building MEX files is similar to that described above. In step 4,
locate the .LIB files for the compiler you are using in a subdirectory of
matlabroot\extern\lib\win32. For example, for version 5.4 of the Borland C/
C++ compiler, look in matlabroot\extern\lib\win32\borland\bc54.
Using the Add-In for Visual Studio
The MathWorks provides a MATLAB add-in for the Visual Studio®
development system that lets you work easily within Microsoft Visual C/C++
(MSVC). The MATLAB add-in for Visual Studio greatly simplifies using M-files
in the MSVC environment. The add-in automates the integration of M-files
into Visual C++ projects. It is fully integrated with the MSVC environment.
The add-in for Visual Studio is automatically installed on your system when
you run either mbuild -setup or mex -setup and select Microsoft Visual C/C++
version 5 or 6. However, there are several steps you must follow in order to use
the add-in:
1 To build MEX-files with the add-in for Visual Studio, run the following
command at the MATLAB command prompt.
1-29
1
Calling C and Fortran Programs from MATLAB
mex -setup
Follow the menus and choose either Microsoft Visual C/C++ 5.0 or 6.0. This
configures mex to use the selected Microsoft compiler and also installs the
necessary add-in files in your Microsoft Visual C/C++ directories.
2 To configure the MATLAB add-in for Visual Studio to work with Microsoft
Visual C/C++:
a Select Tools -> Customize from the MSVC menu.
b Click on the Add-ins and Macro Files tab.
c
Check MATLAB for Visual Studio on the Add-ins and Macro Files list
and click Close. The floating MATLAB add-in for Visual Studio toolbar
appears. The checkmark directs MSVC to automatically load the add-in
when you start MSVC again.
Configuring on Windows 95, Windows 98, and Windows ME Systems
Windows 95 and 98. To run the MATLAB add-in for Visual Studio on Windows
95 or Windows 98 systems, add this line to your config.sys file.
shell=c:\command.com /e:32768 /p
Windows ME. To run the MATLAB add-in for Visual Studio on Windows ME
systems, do the following:
1 Find C:\windows\system\conagent.exe in the Windows Explorer.
2 Right click on the conagent.exe icon.
3 Select Properties from the context menu. This brings up the
CONAGENT.EXE Properties window.
4 Select the Memory tab in the CONAGENT.EXE Properties window.
5 Set the Initial Environment field to 4096.
6 Click Apply.
7 Click OK.
1-30
Custom Building MEX-Files
For additional information on the MATLAB add-in for Visual Studio:
• See the MATLABAddin.hlp file in the <matlab>\bin\win32 directory, or
• Click on the Help icon in the MATLAB add-in for Visual Studio toolbar.
Help Icon
1-31
1
Calling C and Fortran Programs from MATLAB
Troubleshooting
This section explains how to troubleshoot some of the more common problems
you may encounter. It addresses the following topics:
• “Configuration Issues”
• “Understanding MEX-File Problems”
• “Compiler and Platform-Specific Issues”
• “Memory Management Compatibility Issues”
Configuration Issues
This section focuses on some common problems that might occur when creating
MEX-files.
Search Path Problem on Windows
Under Windows, if you move the MATLAB executable without reinstalling
MATLAB, you may need to modify mex.bat to point to the new MATLAB
location.
MATLAB Pathnames Containing Spaces on Windows
If you have problems building MEX-files on Windows and there is a space in
any of the directory names within the MATLAB path, you need to either
reinstall MATLAB into a pathname that contains no spaces or rename the
directory that contains the space. For example, if you install MATLAB under
the Program Files directory, you may have difficulty building MEX-files with
certain C compilers. Also, if you install MATLAB in a directory such as MATLAB
V5.2, you may have difficulty.
DLLs Not on Path on Windows
MATLAB will fail to load MEX-files if it cannot find all DLLs referenced by the
MEX-file; the DLLs must be on the DOS path or in the same directory as the
MEX-file. This is also true for third-party DLLs.
Internal Error When Using mex -setup (PC).
Some antivirus software packages may conflict with the mex -setup process or
other mex commands. If you get an error message of the following form in
response to a mex command,
1-32
Troubleshooting
mex.bat: internal error in sub get_compiler_info(): don't
recognize <string>
then you need to disable your antivirus software temporarily and reenter the
command. After you have successfully run the mex operation, you can
re-enable your antivirus software.
Alternatively, you can open a separate MS-DOS window and enter the mex
command from that window.
General Configuration Problem
Make sure you followed the configuration steps for your platform described in
this chapter. Also, refer to “Custom Building MEX-Files” on page 1-20 for
additional information.
Understanding MEX-File Problems
This section contains information regarding common problems that occur when
creating MEX-files. Use the figure, below, to help isolate these problems.
1-33
1
Calling C and Fortran Programs from MATLAB
Start
Can you compile
and run timestwo.c
or timestwo.f?
no
1
Are you using a
supported compiler
?
Acquire a supported compiler.
See “Supported Compilers”
for details.
no
Stop
yes
Double check your configuration.
yes
See “Testing Your Configuration
on UNIX (or Windows)”
2
Check for:
Can you compile
your program
?
no
ANSI C code
General C syntax errors
yes
3
Can MATLAB
load your MEX-file
?
no
Check:
Spelling of mexFunction
Link against all libraries
you intend to use.
yes
4
Segmentation fault
or bus error
?
yes
no
5
Do you get
the right answer
?
Use:
mexPrintf
matlab -check_malloc1
no
Run in debugger.
yes
Use:
matlab -check_malloc1
mex -argcheck2
Stop
1
2
UNIX only
MEX-files only
Figure 1-1: Troubleshooting MEX-File Creation Problems
Problems 1 through 5 refer to specific sections of the previous flowchart. For
additional suggestions on resolving MEX build problems, access the
1-34
Troubleshooting
MathWorks Technical Support Web site at http://www.mathworks.com/
support.
Problem 1 - Compiling a MathWorks Program Fails
The most common configuration problem in creating C MEX-files on UNIX
involves using a non-ANSI C compiler, or failing to pass to the compiler a flag
that tells it to compile ANSI C code.
A reliable way of knowing if you have this type of configuration problem is if
the header files supplied by The MathWorks generate a string of syntax errors
when you try to compile your code. See “Building MEX-Files” on page 1-11 for
information on selecting the appropriate options file or, if necessary, obtain an
ANSI C compiler.
Problem 2 - Compiling Your Own Program Fails
A second way of generating a string of syntax errors occurs when you attempt
to mix ANSI and non-ANSI C code. The MathWorks provides header and
source files that are ANSI C compliant. Therefore, your C code must also be
ANSI compliant.
Other common problems that can occur in any C program are neglecting to
include all necessary header files, or neglecting to link against all required
libraries.
Problem 3 - MEX-File Load Errors
If you receive an error of the form
Unable to load mex file:
??? Invalid MEX-file
MATLAB is unable to recognize your MEX-file as being valid.
MATLAB loads MEX-files by looking for the gateway routine, mexFunction. If
you misspell the function name, MATLAB is not able to load your MEX-file and
generates an error message. On Windows, check that you are exporting
mexFunction correctly.
On some platforms, if you fail to link against required libraries, you may get an
error when MATLAB loads your MEX-file rather than when you compile your
MEX-file. In such cases, you see a system error message referring to unresolved
1-35
1
Calling C and Fortran Programs from MATLAB
symbols or unresolved references. Be sure to link against the library that
defines the function in question.
On Windows, MATLAB will fail to load MEX-files if it cannot find all DLLs
referenced by the MEX-file; the DLLs must be on the path or in the same
directory as the MEX-file. This is also true for third party DLLs.
Problem 4 - Segmentation Fault or Bus Error
If your MEX-file causes a segmentation violation or bus error, it means that the
MEX-file has attempted to access protected, read-only, or unallocated memory.
Since this is such a general category of programming errors, such problems are
sometimes difficult to track down.
Segmentation violations do not always occur at the same point as the logical
errors that cause them. If a program writes data to an unintended section of
memory, an error may not occur until the program reads and interprets the
corrupted data. Consequently, a segmentation violation or bus error can occur
after the MEX-file finishes executing.
MATLAB provides three features to help you in troubleshooting problems of
this nature. Listed in order of simplicity, they are:
• Recompile your MEX-file with argument checking (C MEX-files only).
You can add a layer of error checking to your MEX-file by recompiling with
the mex script flag -argcheck. This warns you about invalid arguments to
both MATLAB MEX-file (mex) and matrix access (mx) API functions.
Although your MEX-file will not run as efficiently as it can, this switch
detects such errors as passing null pointers to API functions.
• Run MATLAB with the -check_malloc option (UNIX only). The MATLAB
startup flag, -check_malloc, indicates that MATLAB should maintain
additional memory checking information. When memory is freed, MATLAB
checks to make sure that memory just before and just after this memory
remains unwritten and that the memory has not been previously freed.
If an error occurs, MATLAB reports the size of the allocated memory block.
Using this information, you can track down where in your code this memory
was allocated, and proceed accordingly.
Although using this flag prevents MATLAB from running as efficiently as it
can, it detects such errors as writing past the end of a dimensioned array, or
freeing previously freed memory.
1-36
Troubleshooting
• Run MATLAB within a debugging environment. This process is already
described in the chapters on creating C and Fortran MEX-files, respectively.
Problem 5 - Program Generates Incorrect Results
If your program generates the wrong answer(s), there are several possible
causes. First, there could be an error in the computational logic. Second, the
program could be reading from an uninitialized section of memory. For
example, reading the 11th element of a 10-element vector yields unpredictable
results.
Another possibility for generating a wrong answer could be overwriting valid
data due to memory mishandling. For example, writing to the 15th element of
a 10-element vector might overwrite data in the adjacent variable in memory.
This case can be handled in a similar manner as segmentation violations as
described in Problem 4.
In all of these cases, you can use mexPrintf to examine data values at
intermediate stages, or run MATLAB within a debugger to exploit all the tools
the debugger provides.
Compiler and Platform-Specific Issues
This section refers to situations specific to particular compilers and platforms.
MEX-Files Created in Watcom IDE
If you use the Watcom IDE to create MEX-files and get unresolved references
to API functions when linking against our libraries, check the argument
passing convention. The Watcom IDE uses a default switch that passes
parameters in registers. MATLAB requires that you pass parameters on the
stack.
Memory Management Compatibility Issues
To address performance issues, we have made some changes to the internal
MATLAB memory management model. These changes will allow us to provide
future enhancements to the MEX-file API.
As of MATLAB 5.2, MATLAB implicitly calls mxDestroyArray, the mxArray
destructor, at the end of a MEX-file’s execution on any mxArrays that are not
returned in the left-hand side list (plhs[]). You are now warned if MATLAB
detects any misconstructed or improperly destructed mxArrays.
1-37
1
Calling C and Fortran Programs from MATLAB
We highly recommend that you fix code in your MEX-files that produces any of
the warnings discussed in the following sections. For additional information,
see “Memory Management” on page 2-38 in Creating C Language MEX-Files.
Note Currently, the following warnings are enabled by default for backwards
compatibility reasons. In future releases of MATLAB, the warnings will be
disabled by default. The programmer will be responsible for enabling these
warnings during the MEX-file development cycle.
Improperly Destroying an mxArray
You cannot use mxFree to destroy an mxArray.
Warning
Warning: You are attempting to call mxFree on a <class-id> array.
The destructor for mxArrays is mxDestroyArray; please call this
instead. MATLAB will attempt to fix the problem and continue, but
this will result in memory faults in future releases.
Example That Causes Warning
In the following example, mxFree does not destroy the array object. This
operation frees the structure header associated with the array, but MATLAB
will still operate as if the array object needs to be destroyed. Thus MATLAB
will try to destroy the array object, and in the process, attempt to free its
structure header again.
mxArray *temp = mxCreateDoubleMatrix(1,1,mxREAL);
...
mxFree(temp); /* INCORRECT */
1-38
Troubleshooting
Solution
Call mxDestroyArray instead.
mxDestroyArray(temp);
/* CORRECT */
Incorrectly Constructing a Cell or Structure mxArray
You cannot call mxSetCell or mxSetField variants with prhs[] as the member
array.
Warning
Warning: You are attempting to use an array from another scope
(most likely an input argument) as a member of a cell array or
structure. You need to make a copy of the array first. MATLAB will
attempt to fix the problem and continue, but this will result in
memory faults in future releases.
Example That Causes Warning
In the following example, when the MEX-file returns, MATLAB will destroy
the entire cell array. Since this includes the members of the cell, this will
implicitly destroy the MEX-file’s input arguments. This can cause several
strange results, generally having to do with the corruption of the caller’s
workspace, if the right-hand side argument used is a temporary array (i.e., a
literal or the result of an expression).
myfunction('hello')
/* myfunction is the name of your MEX-file and your code */
/* contains the following:
*/
mxArray *temp = mxCreateCellMatrix(1,1);
...
mxSetCell(temp, 0, prhs[0]); /* INCORRECT */
Solution
Make a copy of the right-hand side argument with mxDuplicateArray and use
that copy as the argument to mxSetCell (or mxSetField variants); for example
mxSetCell(temp, 0, mxDuplicateArray(prhs[0]));
/* CORRECT */
1-39
1
Calling C and Fortran Programs from MATLAB
Creating a Temporary mxArray with Improper Data
You cannot call mxDestroyArray on an mxArray whose data was not allocated
by an API routine.
Warning
Warning: You have attempted to point the data of an array to a
block of memory not allocated through the MATLAB API. MATLAB will
attempt to fix the problem and continue, but this will result in
memory faults in future releases.
Example That Causes Warning
If you call mxSetPr, mxSetPi, mxSetData, or mxSetImagData, specifying memory
that was not allocated by mxCalloc, mxMalloc, or mxRealloc as the intended
data block (second argument), then when the MEX-file returns, MATLAB will
attempt to free the pointer to real data and the pointer to imaginary data (if
any). Thus MATLAB will attempt to free memory, in this example, from the
program stack. This will cause the above warning when MATLAB attempts to
reconcile its consistency checking information.
mxArray *temp = mxCreateDoubleMatrix(0,0,mxREAL);
double data[5] = {1,2,3,4,5};
...
mxSetM(temp,1); mxSetN(temp,5); mxSetPr(temp, data);
/* INCORRECT */
Solution
Rather than use mxSetPr to set the data pointer, instead create the mxArray
with the right size and use memcpy to copy the stack data into the buffer
returned by mxGetPr.
mxArray *temp = mxCreateDoubleMatrix(1,5,mxREAL);
double data[5] = {1,2,3,4,5};
...
memcpy(mxGetPr(temp), data, 5*sizeof(double)); /* CORRECT */
Potential Memory Leaks
Prior to Version 5.2, if you created an mxArray using one of the API creation
routines and then you overwrote the pointer to the data using mxSetPr,
MATLAB would still free the original memory. This is no longer the case.
1-40
Troubleshooting
For example,
pr = mxCalloc(5*5, sizeof(double));
... <load data into pr>
plhs[0] = mxCreateDoubleMatrix(5,5,mxREAL);
mxSetPr(plhs[0], pr); /* INCORRECT */
will now leak 5*5*8 bytes of memory, where 8 bytes is the size of a double.
You can avoid that memory leak by changing the code
plhs[0] = mxCreateDoubleMatrix(5,5,mxREAL);
pr = mxGetPr(plhs[0]);
... <load data into pr>
or alternatively
pr = mxCalloc(5*5, sizeof(double));
... <load data into pr>
plhs[0] = mxCreateDoubleMatrix(5,5,mxREAL);
mxFree(mxGetPr(plhs[0]));
mxSetPr(plhs[0], pr);
Note that the first solution is more efficient.
Similar memory leaks can also occur when using mxSetPi, mxSetData,
mxSetImagData, mxSetIr, or mxSetJc. You can address this issue as shown
above to avoid such memory leaks.
MEX-Files Should Destroy Their Own Temporary Arrays
In general, we recommend that MEX-files destroy their own temporary arrays
and clean up their own temporary memory. All mxArrays except those returned
in the left-hand side list and those returned by mexGetArrayPtr may be safely
destroyed. This approach is consistent with other MATLAB API applications
(i.e., MAT-file applications, engine applications, and MATLAB Compiler
generated applications, which do not have any automatic cleanup mechanism.)
1-41
1
Calling C and Fortran Programs from MATLAB
Additional Information
The following sections describe how to find additional information and
assistance in building your applications. It covers the following topics:
• “Files and Directories - UNIX Systems”
• “Files and Directories - Windows Systems”
• “Examples”
• “Technical Support”
Files and Directories - UNIX Systems
This section describes the directory organization and purpose of the files
associated with the MATLAB API on UNIX systems.
The following figure illustrates the directories in which the MATLAB API files
are located. In the illustration, <matlab> symbolizes the top-level directory
where MATLAB is installed on your system.
<matlab>
bin
extern
lib
$ARCH
include
src
eng_mat
mex
examples
mx
refbook
1-42
Additional Information
<matlab>/bin
The <matlab>/bin directory contains two files that are relevant for the
MATLAB API.
mex
UNIX shell script that creates MEX-files from C or
Fortran MEX-file source code. See “Introducing
MEX-Files” on page 1-3 for more details on mex.
matlab
UNIX shell script that initializes your environment
and then invokes the MATLAB interpreter.
This directory also contains the preconfigured options files that the mex script
uses with particular compilers. This table lists the options files.
Table 1-4: Preconfigured Options Files
Options File
Description
cxxopts.sh
Used with the mex script and the system C++ compiler
to compile MEX-files
engopts.sh
Used with the mex script and the system C or Fortran
compiler to compile engine applications
gccopts.sh
Used with the mex script and the GNU C (gcc) compiler
to compile MEX-files
matopts.sh
Used with the mex script and the system C or Fortran
compiler to compile MAT-file applications
mexopts.sh
Used with the mex script and the system ANSI C or
Fortran compiler to compile MEX-files
<matlab>/extern/lib/$ARCH
The <matlab>/extern/lib/$ARCH directory contains libraries, where $ARCH
specifies a particular UNIX platform. On some UNIX platforms, this directory
contains two versions of this library. Library filenames ending with .a are
static libraries and filenames ending with .so or .sl are shared libraries.
1-43
1
Calling C and Fortran Programs from MATLAB
<matlab>/extern/include
The <matlab>/extern/include directory contains the header files for
developing C and C++ applications that interface with MATLAB.
The relevant header files for the MATLAB API are:
engine.h
Header file for MATLAB engine programs. Contains
function prototypes for engine routines.
mat.h
Header file for programs accessing MAT-files.
Contains function prototypes for mat routines.
matrix.h
Header file containing a definition of the mxArray
structure and function prototypes for matrix access
routines.
mex.h
Header file for building MEX-files. Contains function
prototypes for mex routines.
<matlab>/extern/src
The <matlab>/extern/src directory contains those C source files that are
necessary to support certain MEX-file features such as argument checking and
versioning.
Files and Directories - Windows Systems
This section describes the directory organization and purpose of the files
associated with the MATLAB API on Microsoft Windows systems.
The following figure illustrates the directories in which the MATLAB API files
are located. In the illustration, <matlab> symbolizes the top-level directory
where MATLAB is installed on your system.
1-44
Additional Information
<matlab>
bin
win32
mexopts
extern
include
src
examples
eng_mat
mex
mx
refbook
<matlab>\bin\win32
The <matlab>\bin\win32 directory contains the mex.bat batch file that builds
C and Fortran files into MEX-files. Also, this directory contains mex.pl, which
is a Perl script used by mex.bat.
<matlab>\bin\win32\mexopts
The <matlab>\bin\win32\mexopts directory contains the preconfigured
options files that the mex script uses with particular compilers. See Table 1-2,
Options Files, on page 1-17 for a complete list of the options files.
1-45
1
Calling C and Fortran Programs from MATLAB
<matlab>\extern\include
The <matlab>\extern\include directory contains the header files for
developing C and C++ applications that interface with MATLAB.
The relevant header files for the MATLAB API (MEX-files, engine, and
MAT-files) are:
engine.h
Header file for MATLAB engine programs. Contains
function prototypes for engine routines.
mat.h
Header file for programs accessing MAT-files.
Contains function prototypes for mat routines.
matrix.h
Header file containing a definition of the mxArray
structure and function prototypes for matrix access
routines.
mex.h
Header file for building MEX-files. Contains function
prototypes for mex routines.
_*.def
Files used by Borland compiler.
*.def
Files used by MSVC and Microsoft Fortran compilers.
mexversion.rc
Resource file for inserting versioning information into
MEX-files.
<matlab>\extern\src
The <matlab>\extern\src directory contains files that are used for debugging
MEX-files.
Examples
This book uses many examples to show how to write C and Fortran MEX-files.
Examples From the Text
The refbook subdirectory in the extern/examples directory contains the
MEX-file examples (C and Fortran) that are used in this book, External
Interfaces.
You can find the most recent versions of these examples using the anonymous
FTP server locations
ftp://ftp.mathworks.com/pub/tech-support/docexamples/apiguide/
R12/refbook
1-46
Additional Information
MEX Reference Examples
The mex subdirectory of /extern/examples directory contains MEX-file
examples. It includes the examples described in the online External Interfaces/
API reference pages for MEX interface functions (the functions beginning with
the mex prefix).
You can find the most recent versions of these examples using the anonymous
FTP server location
ftp://ftp.mathworks.com/pub/tech-support/docexamples/apiguide/
R12/mex
MX Examples
The mx subdirectory of extern/examples contains examples for using the array
access functions. Although you can use these functions in stand-alone
programs, most of these are MEX-file examples. The exception is
mxSetAllocFcns.c, since this function is available only to stand-alone
programs.
You can find the most recent versions of these examples using the anonymous
FTP server location
ftp://ftp.mathworks.com/pub/tech-support/docexamples/apiguide/
R12/mx
Engine and MAT Examples
The eng_mat subdirectory in the extern/examples directory contains the
MEX-file examples (C and Fortran) for using the MATLAB engine facility, as
well as examples for reading and writing MATLAB data files (MAT-files).
These examples are all stand-alone programs.
You can find the most recent versions of these examples using the anonymous
FTP server locations
ftp://ftp.mathworks.com/pub/tech-support/docexamples/apiguide/
R12/eng_mat
1-47
1
Calling C and Fortran Programs from MATLAB
Technical Support
The MathWorks provides additional Technical Support through its web site. A
few of the services provided are as follows:
• FAQ (Frequently Asked Questions)
This is a list of the most frequently asked questions received by our Technical
Support staff, along with answers from the solutions database.
http://www.mathworks.com/support/index.shtml
• Solution Search Engine
This knowledge base on our web site includes thousands of solutions and links
to Technical Notes and is updated several times each week.
http://www.mathworks.com/support/solutions/index.shtml
• Technical Notes
Technical notes are written by our Technical Support staff to address
commonly asked questions.
http://www.mathworks.com/support/tech-notes/index.shtml
1-48
2
Creating C Language
MEX-Files
C MEX-Files . . . . . . . . . . . . . . . . . . . . 2-3
The Components of a C MEX-File . . . . . . . . . . . . 2-3
Required Arguments to a MEX-File . . . . . . . . . . . 2-5
Examples of C MEX-Files . . . . . . . . .
A First Example — Passing a Scalar . . . . . .
Passing Strings . . . . . . . . . . . . . .
Passing Two or More Inputs or Outputs . . . .
Passing Structures and Cell Arrays . . . . . .
Handling Complex Data . . . . . . . . . . .
Handling 8-,16-, and 32-Bit Data . . . . . . .
Manipulating Multidimensional Numerical Arrays
Handling Sparse Arrays . . . . . . . . . . .
Calling Functions from C MEX-Files . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 2-7
. 2-7
. 2-11
. 2-13
. 2-16
. 2-21
. 2-23
. 2-25
. 2-29
. 2-33
Advanced Topics . . . . . . . .
Help Files . . . . . . . . . . .
Linking Multiple Files . . . . . .
Workspace for MEX-File Functions .
Memory Management . . . . . . .
Using LAPACK and BLAS Functions
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 2-37
. 2-37
. 2-37
. 2-37
. 2-38
. 2-40
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Debugging C Language MEX-Files . . . . . . . . . . 2-46
Debugging on UNIX . . . . . . . . . . . . . . . . . 2-46
Debugging on Windows . . . . . . . . . . . . . . . . 2-47
2
Creating C Language MEX-Files
This chapter describes how to write MEX-files in the C programming language.
It discusses the MEX-file itself, how these C language files interact with
MATLAB, how to pass and manipulate arguments of different data types, how
to debug your MEX-file programs, and several other, more advanced topics.
The following list summarizes the contents of this chapter:
• “C MEX-Files”
• “Examples of C MEX-Files”
• “Advanced Topics”
• “Debugging C Language MEX-Files”
2-2
C MEX-Files
C MEX-Files
C MEX-files are built by using the mex script to compile your C source code with
additional calls to API routines.
The Components of a C MEX-File
The source code for a MEX-file consists of two distinct parts:
• A computational routine that contains the code for performing the
computations that you want implemented in the MEX-file. Computations
can be numerical computations as well as inputting and outputting data.
• A gateway routine that interfaces the computational routine with MATLAB
by the entry point mexFunction and its parameters prhs, nrhs, plhs, nlhs,
where prhs is an array of right-hand input arguments, nrhs is the number
of right-hand input arguments, plhs is an array of left-hand output
arguments, and nlhs is the number of left-hand output arguments. The
gateway calls the computational routine as a subroutine.
In the gateway routine, you can access the data in the mxArray structure and
then manipulate this data in your C computational subroutine. For example,
the expression mxGetPr(prhs[0]) returns a pointer of type double * to the real
data in the mxArray pointed to by prhs[0]. You can then use this pointer like
any other pointer of type double * in C. After calling your C computational
routine from the gateway, you can set a pointer of type mxArray to the data it
returns. MATLAB is then able to recognize the output from your computational
routine as the output from the MEX-file.
The following C MEX Cycle figure shows how inputs enter a MEX-file, what
functions the gateway routine performs, and how outputs return to MATLAB.
2-3
2
Creating C Language MEX-Files
MATLAB
A call to
MEX-file func:
[C,D]=func(A,B)
tells MATLAB to
pass variables A and
B to your MEX-file.
C and D are left
unassigned.
INPUTS
const mxArray *B
B = prhs[1]
const mxArray *A
A = prhs[0]
func.c
void mexFunction(
int nlhs, mxArray *plhs[],
int nrhs, const mxArray *prhs[])
In the gateway routine:
• Use the mxCreate functions to create
the MATLAB arrays for your output
arguments. Set plhs[0], [1], … to the
pointers to the newly created
MATLAB arrays.
• Use the mxGet functions to extract
your data from prhs[0], [1], …
MATLAB
On return from
MEX-file func:
[C,D]=func(A,B)
plhs[0] is assigned
to C and plhs[1] is
assigned to D.
• Call your C subroutine passing the
input and output data pointers as
function parameters.
mxArray *D
D = plhs[1]
mxArray *C
C = plhs[0]
OUTPUTS
Figure 2-1: C MEX Cycle
2-4
C MEX-Files
Required Arguments to a MEX-File
The two components of the MEX-file may be separate or combined. In either
case, the files must contain the #include "mex.h" header so that the entry
point and interface routines are declared properly. The name of the gateway
routine must always be mexFunction and must contain these parameters.
void mexFunction(
int nlhs, mxArray *plhs[],
int nrhs, const mxArray *prhs[])
{
/* more C code ... */
The parameters nlhs and nrhs contain the number of left- and right-hand
arguments with which the MEX-file is invoked. In the syntax of the MATLAB
language, functions have the general form
[a,b,c,…] = fun(d,e,f,…)
where the ellipsis (…) denotes additional terms of the same format. The a,b,c,…
are left-hand arguments and the d,e,f,… are right-hand arguments.
The parameters plhs and prhs are vectors that contain pointers to the left- and
right-hand arguments of the MEX-file. Note that both are declared as
containing type mxArray *, which means that the variables pointed at are
MATLAB arrays. prhs is a length nrhs array of pointers to the right-hand side
inputs to the MEX-file, and plhs is a length nlhs array that will contain
pointers to the left-hand side outputs that your function generates.
2-5
2
Creating C Language MEX-Files
For example, if you invoke a MEX-file from the MATLAB workspace with the
command
x = fun(y,z);
the MATLAB interpreter calls mexFunction with the arguments
nlhs = 1
nrhs = 2
plhs
prhs
•
•
•
Ø
Y
Z
plhs is a 1-element C array where the single element is a null pointer. prhs is
a 2-element C array where the first element is a pointer to an mxArray named
Y and the second element is a pointer to an mxArray named Z.
The parameter plhs points at nothing because the output x is not created until
the subroutine executes. It is the responsibility of the gateway routine to create
an output array and to set a pointer to that array in plhs[0]. If plhs[0] is left
unassigned, MATLAB prints a warning message stating that no output has
been assigned.
Note It is possible to return an output value even if nlhs = 0. This
corresponds to returning the result in the ans variable.
2-6
Examples of C MEX-Files
Examples of C MEX-Files
The following sections include information and examples describing how to
pass and manipulate the different data types when working with MEX-files.
These topics include
• “A First Example — Passing a Scalar”
• “Passing Strings”
• “Passing Two or More Inputs or Outputs”
• “Passing Structures and Cell Arrays”
• “Handling Complex Data”
• “Handling 8-,16-, and 32-Bit Data”
• “Manipulating Multidimensional Numerical Arrays”
• “Handling Sparse Arrays”
• “Calling Functions from C MEX-Files”
The MATLAB API provides a full set of routines that handle the various data
types supported by MATLAB. For each data type there is a specific set of
functions that you can use for data manipulation. The first example discusses
the simple case of doubling a scalar. After that, the examples discuss how to
pass in, manipulate, and pass back various data types, and how to handle
multiple inputs and outputs. Finally, the sections discuss passing and
manipulating various MATLAB data types.
Note You can find the most recent versions of the example programs at the
anonymous FTP server
ftp://ftp.mathworks.com/pub/tech-support/docexamples/apiguide/R12/refbook
A First Example — Passing a Scalar
Let’s look at a simple example of C code and its MEX-file equivalent. Here is a
C computational function that takes a scalar and doubles it.
#include <math.h>
void timestwo(double y[], double x[])
2-7
2
Creating C Language MEX-Files
{
y[0] = 2.0*x[0];
return;
}
Below is the same function written in the MEX-file format.
#include "mex.h"
/*
* timestwo.c - example found in API guide
*
* Computational function that takes a scalar and doubles it.
*
* This is a MEX-file for MATLAB.
* Copyright (c) 1984-2000 The MathWorks, Inc.
*/
/* $Revision: 1.8 $ */
void timestwo(double y[], double x[])
{
y[0] = 2.0*x[0];
}
void mexFunction( int nlhs, mxArray *plhs[],
int nrhs, const mxArray *prhs[] )
{
double *x,*y;
int
mrows,ncols;
/* Check for proper
if(nrhs != 1) {
mexErrMsgTxt("One
} else if(nlhs > 1)
mexErrMsgTxt("Too
}
number of arguments. */
input required.");
{
many output arguments");
/* The input must be a noncomplex scalar double.*/
mrows = mxGetM(prhs[0]);
ncols = mxGetN(prhs[0]);
2-8
Examples of C MEX-Files
if(!mxIsDouble(prhs[0]) || mxIsComplex(prhs[0]) ||
!(mrows == 1 && ncols == 1) ) {
mexErrMsgTxt("Input must be a noncomplex scalar double.");
}
/* Create matrix for the return argument. */
plhs[0] = mxCreateDoubleMatrix(mrows,ncols, mxREAL);
/* Assign pointers to each input and output. */
x = mxGetPr(prhs[0]);
y = mxGetPr(plhs[0]);
/* Call the timestwo subroutine. */
timestwo(y,x);
}
In C, function argument checking is done at compile time. In MATLAB, you can
pass any number or type of arguments to your M-function, which is responsible
for argument checking. This is also true for MEX-files. Your program must
safely handle any number of input or output arguments of any supported type.
To compile and link this example source file at the MATLAB prompt, type
mex timestwo.c
This carries out the necessary steps to create the MEX-file called timestwo
with an extension corresponding to the platform on which you’re running. You
can now call timestwo as if it were an M-function.
x = 2;
y = timestwo(x)
y =
4
You can create and compile MEX-files in MATLAB or at your operating
system’s prompt. MATLAB uses mex.m, an M-file version of the mex script, and
your operating system uses mex.bat on Windows and mex.sh on UNIX. In
either case, typing
mex filename
at the prompt produces a compiled version of your MEX-file.
2-9
2
Creating C Language MEX-Files
In the above example, scalars are viewed as 1-by-1 matrices. Alternatively, you
can use a special API function called mxGetScalar that returns the values of
scalars instead of pointers to copies of scalar variables. This is the alternative
code (error checking has been omitted for brevity).
#include "mex.h"
/*
* timestwoalt.c - example found in API guide
*
* Use mxGetScalar to return the values of scalars instead of
* pointers to copies of scalar variables.
*
* This is a MEX-file for MATLAB.
* Copyright (c) 1984-2000 The MathWorks, Inc.
*/
/* $Revision: 1.5 $ */
void timestwo_alt(double *y, double x)
{
*y = 2.0*x;
}
void mexFunction( int nlhs, mxArray *plhs[],
int nrhs, const mxArray *prhs[] )
{
double *y;
double x;
/* Create a 1-by-1 matrix for the return argument. */
plhs[0] = mxCreateDoubleMatrix(1,1,mxREAL);
/* Get the scalar value of the input x. */
/* Note: mxGetScalar returns a value, not a pointer. */
x = mxGetScalar(prhs[0]);
/* Assign a pointer to the output. */
y = mxGetPr(plhs[0]);
2-10
Examples of C MEX-Files
/* Call the timestwo_alt subroutine. */
timestwo_alt(y,x);
}
This example passes the input scalar x by value into the timestwo_alt
subroutine, but passes the output scalar y by reference.
Passing Strings
Any MATLAB data type can be passed to and from MEX-files. For example,
this C code accepts a string and returns the characters in reverse order.
/* $Revision: 1.10 $ */
/*=============================================================
* revord.c
* Example for illustrating how to copy the string data from
* MATLAB to a C-style string and back again.
*
* Takes a string and returns a string in reverse order.
*
* This is a MEX-file for MATLAB.
* Copyright (c) 1984-2000 The MathWorks, Inc.
*============================================================*/
#include "mex.h"
void revord(char *input_buf, int buflen, char *output_buf)
{
int
i;
/* Reverse the order of the input string. */
for(i = 0; i < buflen-1; i++)
*(output_buf+i) = *(input_buf+buflen-i-2);
}
In this example, the API function mxCalloc replaces calloc, the standard C
function for dynamic memory allocation. mxCalloc allocates dynamic memory
using MATLAB’s memory manager and initializes it to zero. You must use
mxCalloc in any situation where C would require the use of calloc. The same
is true for mxMalloc and mxRealloc; use mxMalloc in any situation where C
would require the use of malloc and use mxRealloc where C would require
realloc.
2-11
2
Creating C Language MEX-Files
Note MATLAB automatically frees up memory allocated with the mx
allocation routines (mxCalloc, mxMalloc, mxRealloc) upon exiting your
MEX-file. If you don’t want this to happen, use the API function
mexMakeMemoryPersistent.
Below is the gateway routine that calls the C computational routine revord.
void mexFunction( int nlhs, mxArray *plhs[],
int nrhs, const mxArray *prhs[])
{
char *input_buf, *output_buf;
int
buflen,status;
/* Check for proper number of arguments. */
if(nrhs != 1)
mexErrMsgTxt("One input required.");
else if(nlhs > 1)
mexErrMsgTxt("Too many output arguments.");
/* Input must be a string. */
if(mxIsChar(prhs[0]) != 1)
mexErrMsgTxt("Input must be a string.");
/* Input must be a row vector. */
if(mxGetM(prhs[0]) != 1)
mexErrMsgTxt("Input must be a row vector.");
/* Get the length of the input string. */
buflen = (mxGetM(prhs[0]) * mxGetN(prhs[0])) + 1;
/* Allocate memory for input and output strings. */
input_buf = mxCalloc(buflen, sizeof(char));
output_buf = mxCalloc(buflen, sizeof(char));
/*
*
*
*
2-12
Copy the string data from prhs[0] into a C string
input_buf. If the string array contains several rows,
they are copied, one column at a time, into one long
string array. */
Examples of C MEX-Files
status = mxGetString(prhs[0], input_buf, buflen);
if(status != 0)
mexWarnMsgTxt("Not enough space. String is truncated.");
/* Call the C subroutine. */
revord(input_buf, buflen, output_buf);
/* Set C-style string output_buf to MATLAB mexFunction output*/
plhs[0] = mxCreateString(output_buf);
return;
}
The gateway routine allocates memory for the input and output strings. Since
these are C strings, they need to be one greater than the number of elements
in the MATLAB string. Next the MATLAB string is copied to the input string.
Both the input and output strings are passed to the computational subroutine
(revord), which loads the output in reverse order. Note that the output buffer
is a valid null-terminated C string because mxCalloc initializes the memory to
0. The API function mxCreateString then creates a MATLAB string from the
C string, output_buf. Finally, plhs[0], the left-hand side return argument to
MATLAB, is set to the MATLAB array you just created.
By isolating variables of type mxArray from the computational subroutine, you
can avoid having to make significant changes to your original C code.
In this example, typing
x = 'hello world';
y = revord(x)
produces
The string to convert is 'hello world'.
y =
dlrow olleh
Passing Two or More Inputs or Outputs
The plhs[] and prhs[] parameters are vectors that contain pointers to each
left-hand side (output) variable and each right-hand side (input) variable,
respectively. Accordingly, plhs[0] contains a pointer to the first left-hand side
argument, plhs[1] contains a pointer to the second left-hand side argument,
2-13
2
Creating C Language MEX-Files
and so on. Likewise, prhs[0] contains a pointer to the first right-hand side
argument, prhs[1] points to the second, and so on.
This example, xtimesy, multiplies an input scalar by an input scalar or matrix
and outputs a matrix. For example, using xtimesy with two scalars gives
x = 7;
y = 7;
z = xtimesy(x,y)
z =
49
Using xtimesy with a scalar and a matrix gives
x = 9;
y = ones(3);
z = xtimesy(x,y)
z =
9
9
9
9
9
9
9
9
9
This is the corresponding MEX-file C code.
#include "mex.h"
/*
* xtimesy.c - example found in API guide
*
* Multiplies an input scalar times an input matrix and outputs a
* matrix.
*
* This is a MEX-file for MATLAB.
* Copyright (c) 1984-2000 The MathWorks, Inc.
*/
/* $Revision: 1.10 $ */
2-14
Examples of C MEX-Files
void xtimesy(double x, double *y, double *z, int m, int n)
{
int i,j,count = 0;
for(i = 0; i < n; i++) {
for(j = 0; j < m; j++) {
*(z+count) = x * *(y+count);
count++;
}
}
}
/* The gateway routine */
void mexFunction( int nlhs, mxArray *plhs[],
int nrhs, const mxArray *prhs[])
{
double *y,*z;
double x;
int
status,mrows,ncols;
/* Check for proper number of arguments. */
/* NOTE: You do not need an else statement when using
mexErrMsgTxt within an if statement. It will never
get to the else statement if mexErrMsgTxt is executed.
(mexErrMsgTxt breaks you out of the MEX-file.)
*/
if(nrhs != 2)
mexErrMsgTxt("Two inputs required.");
if(nlhs != 1)
mexErrMsgTxt("One output required.");
/* Check to make sure the first input argument is a scalar. */
if(!mxIsDouble(prhs[0]) || mxIsComplex(prhs[0]) ||
mxGetN(prhs[0])*mxGetM(prhs[0]) != 1 ) {
mexErrMsgTxt("Input x must be a scalar.");
}
/* Get the scalar input x. */
x = mxGetScalar(prhs[0]);
2-15
2
Creating C Language MEX-Files
/* Create a pointer to the input matrix y. */
y = mxGetPr(prhs[1]);
/* Get the dimensions of the matrix input y. */
mrows = mxGetM(prhs[1]);
ncols = mxGetN(prhs[1]);
/* Set the output pointer to the output matrix. */
plhs[0] = mxCreateDoubleMatrix(mrows,ncols, mxREAL);
/* Create a C pointer to a copy of the output matrix. */
z = mxGetPr(plhs[0]);
/* Call the C subroutine. */
xtimesy(x,y,z,mrows,ncols);
}
As this example shows, creating MEX-file gateways that handle multiple
inputs and outputs is straightforward. All you need to do is keep track of which
indices of the vectors prhs and plhs correspond to the input and output
arguments of your function. In the example above, the input variable x
corresponds to prhs[0] and the input variable y to prhs[1].
Note that mxGetScalar returns the value of x rather than a pointer to x. This
is just an alternative way of handling scalars. You could treat x as a 1-by-1
matrix and use mxGetPr to return a pointer to x.
Passing Structures and Cell Arrays
Structures and cell arrays were new data types introduced in MATLAB 5. For
a discussion of the features of structures and cell arrays and the built-in
functions MATLAB provides for manipulating them, refer to “Structures and
Cell Arrays” in the Using MATLAB manual. Like all other data types in
MATLAB, structures and cell arrays can be passed into and out of C MEX-files.
Passing structures and cell arrays into MEX-files is just like passing any other
data types, except the data itself is of type mxArray. In practice, this means that
mxGetField (for structures) and mxGetCell (for cell arrays) return pointers of
type mxArray. You can then treat the pointers like any other pointers of type
mxArray, but if you want to pass the data contained in the mxArray to a C
routine, you must use an API function such as mxGetData to access it.
2-16
Examples of C MEX-Files
This example takes an m-by-n structure matrix as input and returns a new
1-by-1 structure that contains these fields:
• String input generates an m-by-n cell array
• Numeric input (noncomplex, scalar values) generates an m-by-n vector of
numbers with the same class ID as the input, for example int, double, and
so on.
* =============================================================
* phonebook.c
* Example for illustrating how to manipulate structure and cell
* array
*
* Takes a (MxN) structure matrix and returns a new structure
* (1x1) containing corresponding fields:for string input, it
* will be (MxN) cell array; and for numeric (noncomplex, scalar)
* input, it will be (MxN) vector of numbers with the same
* classID as input, such as int, double etc..
*
* This is a MEX-file for MATLAB.
* Copyright (c) 1984-2000 The MathWorks, Inc.
*=============================================================*/
/* $Revision: 1.6 $ */
#include "mex.h"
#include "string.h"
#define MAXCHARS 80
/* max length of string contained in each
field */
/* the gateway routine. */
void mexFunction( int nlhs, mxArray *plhs[],
int nrhs, const mxArray *prhs[] )
{
const char **fnames;
/* pointers to field names */
const int *dims;
mxArray
*tmp, *fout;
char
*pdata;
int
ifield, jstruct, *classIDflags;
int
NStructElems, nfields, ndim;
2-17
2
Creating C Language MEX-Files
/* Check proper input and output */
if(nrhs != 1)
mexErrMsgTxt("One input required.");
else if(nlhs > 1)
mexErrMsgTxt("Too many output arguments.");
else if(!mxIsStruct(prhs[0]))
mexErrMsgTxt("Input must be a structure.");
/* Get input arguments */
nfields = mxGetNumberOfFields(prhs[0]);
NStructElems = mxGetNumberOfElements(prhs[0]);
/* allocate memory for storing classIDflags */
classIDflags = mxCalloc(nfields, sizeof(int));
/* Check empty field, proper data type, and data type
consistency; get classID for each field. */
for(ifield = 0; ifield < nfields; ifield++) {
for(jstruct = 0; jstruct < NStructElems; jstruct++) {
tmp = mxGetFieldByNumber(prhs[0], jstruct, ifield);
if(tmp == NULL) {
mexPrintf("%s%d\t%s%d\n",
"FIELD:", ifield+1, "STRUCT INDEX :", jstruct+1);
mexErrMsgTxt("Above field is empty!");
}
if(jstruct == 0) {
if((!mxIsChar(tmp) && !mxIsNumeric(tmp)) ||
mxIsSparse(tmp)) {
mexPrintf("%s%d\t%s%d\n",
"FIELD:", ifield+1, "STRUCT INDEX :", jstruct+1);
mexErrMsgTxt("Above field must have either "
"string or numeric non-sparse data.");
}
classIDflags[ifield] = mxGetClassID(tmp);
} else {
if(mxGetClassID(tmp) != classIDflags[ifield]) {
mexPrintf("%s%d\t%s%d\n",
"FIELD:", ifield+1, "STRUCT INDEX :", jstruct+1);
2-18
Examples of C MEX-Files
mexErrMsgTxt("Inconsistent data type in above field!");
}
else if(!mxIsChar(tmp) && ((mxIsComplex(tmp) ||
mxGetNumberOfElements(tmp) != 1))) {
mexPrintf("%s%d\t%s%d\n",
"FIELD:", ifield+1, "STRUCT INDEX :", jstruct+1);
mexErrMsgTxt("Numeric data in above field "
"must be scalar and noncomplex!");
}
}
}
}
/* Allocate memory for storing pointers */
fnames = mxCalloc(nfields, sizeof(*fnames));
/* Get field name pointers */
for(ifield = 0; ifield < nfields; ifield++) {
fnames[ifield] = mxGetFieldNameByNumber(prhs[0],ifield);
}
/* Create a 1x1 struct matrix for output */
plhs[0] = mxCreateStructMatrix(1, 1, nfields, fnames);
mxFree(fnames);
ndim = mxGetNumberOfDimensions(prhs[0]);
dims = mxGetDimensions(prhs[0]);
for(ifield = 0; ifield < nfields; ifield++) {
/* Create cell/numeric array */
if(classIDflags[ifield] == mxCHAR_CLASS) {
fout = mxCreateCellArray(ndim, dims);
} else {
fout = mxCreateNumericArray(ndim, dims,
classIDflags[ifield], mxREAL);
pdata = mxGetData(fout);
}
/* Copy data from input structure array */
for(jstruct = 0; jstruct < NStructElems; jstruct++) {
tmp = mxGetFieldByNumber(prhs[0],jstruct,ifield);
2-19
2
Creating C Language MEX-Files
if(mxIsChar(tmp)) {
mxSetCell(fout, jstruct, mxDuplicateArray(tmp));
} else {
size_t
sizebuf;
sizebuf = mxGetElementSize(tmp);
memcpy(pdata, mxGetData(tmp), sizebuf);
pdata += sizebuf;
}
}
/* Set each field in output structure */
mxSetFieldByNumber(plhs[0], 0, ifield, fout);
}
mxFree(classIDflags);
return;
}
To see how this program works, enter this structure.
friends(1).name = 'Jordan Robert';
friends(1).phone = 3386;
friends(2).name = 'Mary Smith';
friends(2).phone = 3912;
friends(3).name = 'Stacy Flora';
friends(3).phone = 3238;
friends(4).name = 'Harry Alpert';
friends(4).phone = 3077;
The results of this input are
phonebook(friends)
ans =
name: {1x4 cell }
phone: [3386 3912 3238 3077]
2-20
Examples of C MEX-Files
Handling Complex Data
Complex data from MATLAB is separated into real and imaginary parts.
MATLAB’s API provides two functions, mxGetPr and mxGetPi, that return
pointers (of type double *) to the real and imaginary parts of your data.
This example takes two complex row vectors and convolves them.
/* $Revision: 1.8 $ */
/*=========================================================
* convec.c
* Example for illustrating how to pass complex data
* from MATLAB to C and back again
*
* Convolves two complex input vectors.
*
* This is a MEX-file for MATLAB.
* Copyright (c) 1984-2000 The MathWorks, Inc.
*=======================================================*/
#include "mex.h"
/* Computational subroutine */
void convec( double *xr, double *xi, int nx,
double *yr, double *yi, int ny,
double *zr, double *zi)
{
int i,j;
zr[0] = 0.0;
zi[0] = 0.0;
/* Perform the convolution of the complex vectors. */
for(i = 0; i < nx; i++) {
for(j = 0; j < ny; j++) {
*(zr+i+j) = *(zr+i+j) + *(xr+i) * *(yr+j) - *(xi+i)
* *(yi+j);
*(zi+i+j) = *(zi+i+j) + *(xr+i) * *(yi+j) + *(xi+i)
* *(yr+j);
}
}
}
2-21
2
Creating C Language MEX-Files
Below is the gateway routine that calls this complex convolution.
/* The gateway routine. */
void mexFunction( int nlhs, mxArray *plhs[],
int nrhs, const mxArray *prhs[] )
{
double *xr, *xi, *yr, *yi, *zr, *zi;
int
rows, cols, nx, ny;
/* Check for the proper number of arguments. */
if(nrhs != 2)
mexErrMsgTxt("Two inputs required.");
if(nlhs > 1)
mexErrMsgTxt("Too many output arguments.");
/* Check that both inputs are row vectors. */
if(mxGetM(prhs[0]) != 1 || mxGetM(prhs[1]) != 1 )
mexErrMsgTxt("Both inputs must be row vectors.");
rows = 1;
/* Check that both inputs are complex. */
if(!mxIsComplex(prhs[0]) || !mxIsComplex(prhs[1]) )
mexErrMsgTxt("Inputs must be complex.\n");
/* Get the length of each input vector. */
nx = mxGetN(prhs[0]);
ny = mxGetN(prhs[1]);
/*
xr
xi
yr
yi
Get pointers to real and imaginary parts of the inputs. */
= mxGetPr(prhs[0]);
= mxGetPi(prhs[0]);
= mxGetPr(prhs[1]);
= mxGetPi(prhs[1]);
/* Create a new array and set the output pointer to it. */
cols = nx + ny - 1;
plhs[0] = mxCreateDoubleMatrix(rows, cols, mxCOMPLEX);
zr = mxGetPr(plhs[0]);
zi = mxGetPi(plhs[0]);
2-22
Examples of C MEX-Files
/* Call the C subroutine. */
convec(xr, xi, nx, yr, yi, ny, zr, zi);
return;
}
Entering these numbers at the MATLAB prompt
x = [3.000 - 1.000i, 4.000 + 2.000i, 7.000 - 3.000i];
y = [8.000 - 6.000i, 12.000 + 16.000i, 40.000 - 42.000i];
and invoking the new MEX-file
z = convec(x,y)
results in
z =
1.0e+02 *
Columns 1 through 4
0.1800 - 0.2600i 0.9600 + 0.2800i 1.3200 - 1.4400i 3.7600 - 0.1200i
Column 5
1.5400 - 4.1400i
which agrees with the results that the built-in MATLAB function conv.m
produces.
Handling 8-,16-, and 32-Bit Data
You can create and manipulate signed and unsigned 8-, 16-, and 32-bit data
from within your MEX-files. The MATLAB API provides a set of functions that
support these data types. The API function mxCreateNumericArray constructs
an unpopulated N-dimensional numeric array with a specified data size. Refer
to the entry for mxClassID in the online reference pages for a discussion of how
the MATLAB API represents these data types.
Once you have created an unpopulated MATLAB array of a specified data type,
you can access the data using mxGetData and mxGetImagData. These two
functions return pointers to the real and imaginary data. You can perform
2-23
2
Creating C Language MEX-Files
arithmetic on data of 8-, 16- or 32-bit precision in MEX-files and return the
result to MATLAB, which will recognize the correct data class. Although from
within MATLAB it is not currently possible to perform arithmetic or to call
MATLAB functions that perform data manipulation on data of 8-, 16-, or 32-bit
precision, you can display the data at the MATLAB prompt and save it in a
MAT-file.
This example constructs a 2-by-2 matrix with unsigned 16-bit integers, doubles
each element, and returns both matrices to MATLAB.
#include <string.h> /* Needed for memcpy() */
#include "mex.h"
/*
* doubleelement.c - Example found in API Guide
*
* Constructs a 2-by-2 matrix with unsigned 16-bit integers,
* doubles each element, and returns the matrix.
*
* This is a MEX-file for MATLAB.
* Copyright (c) 1984-2000 The MathWorks, Inc.
*/
/* $Revision: 1.8 $ */
#define NDIMS 2
#define TOTAL_ELEMENTS 4
/* The computational subroutine */
void dbl_elem(unsigned short *x)
{
unsigned short scalar = 2;
int i,j;
for(i = 0; i < 2;i++) {
for(j = 0; j < 2;j++) {
*(x+i+j) = scalar * *(x+i+j);
}
}
}
2-24
Examples of C MEX-Files
/* The gataway function */
void mexFunction( int nlhs, mxArray *plhs[],
int nrhs, const mxArray *prhs[] )
{
const int dims[] = {2,2};
unsigned char *start_of_pr;
unsigned short data[] = {1,2,3,4};
int bytes_to_copy;
/* Call the computational subroutine. */
dbl_elem(data);
/* Create a 2-by-2 array of unsigned 16-bit integers. */
plhs[0] = mxCreateNumericArray(NDIMS,dims,
mxUINT16_CLASS,mxREAL);
/* Populate the real part of the created array. */
start_of_pr = (unsigned char *)mxGetPr(plhs[0]);
bytes_to_copy = TOTAL_ELEMENTS * mxGetElementSize(plhs[0]);
memcpy(start_of_pr,data,bytes_to_copy);
}
At the MATLAB prompt, entering
doubleelement
produces
ans =
2
8
6
4
The output of this function is a 2-by-2 matrix populated with unsigned 16-bit
integers. You can view the contents of this matrix in MATLAB, but you cannot
manipulate the data in any fashion.
Manipulating Multidimensional Numerical Arrays
Multidimensional numerical arrays were a new data type introduced in
MATLAB 5. For a discussion of the features of multidimensional numerical
arrays and the built-in functions MATLAB provides to manipulate them, refer
to “Multidimensional Arrays” in the Using MATLAB manual. Like all other
2-25
2
Creating C Language MEX-Files
data types in MATLAB, arrays can be passed into and out of MEX-files written
in C. You can manipulate multidimensional numerical arrays by using
mxGetData and mxGetImagData to return pointers to the real and imaginary
parts of the data stored in the original multidimensional array.
This example takes an N-dimensional array of doubles and returns the indices
for the nonzero elements in the array.
/*=============================================================
* findnz.c
* Example for illustrating how to handle N-dimensional arrays in
* a MEX-file. NOTE: MATLAB uses 1-based indexing, C uses 0-based
* indexing.
*
* Takes an N-dimensional array of doubles and returns the indices
* for the non-zero elements in the array. findnz works
* differently than the FIND command in MATLAB in that it returns
* all the indices in one output variable, where the column
* element contains the index for that dimension.
*
*
* This is a MEX-file for MATLAB.
* Copyright (c) 1984-2000 by The MathWorks, Inc.
*
*============================================================*/
/* $Revision: 1.5 $ */
#include "mex.h"
/*
*
*
*
*
*
*
If you are using a compiler that equates NaN to zero, you must
compile this example using the flag -DNAN_EQUALS_ZERO. For
example:
mex -DNAN_EQUALS_ZERO findnz.c
This will correctly define the IsNonZero macro for your
compiler. */
#if NAN_EQUALS_ZERO
#define IsNonZero(d) ((d) != 0.0 || mxIsNaN(d))
#else
2-26
Examples of C MEX-Files
#define IsNonZero(d) ((d) != 0.0)
#endif
void mexFunction(int nlhs, mxArray *plhs[],
int nrhs, const mxArray *prhs[])
{
/* Declare variables. */
int elements, j, number_of_dims, cmplx;
int nnz = 0, count = 0;
double *pr, *pi, *pind;
const int *dim_array;
/* Check for proper number of input and output arguments. */
if(nrhs != 1) {
mexErrMsgTxt("One input argument required.");
}
if(nlhs > 1) {
mexErrMsgTxt("Too many output arguments.");
}
/* Check data type of input argument. */
if(!(mxIsDouble(prhs[0]))) {
mexErrMsgTxt("Input array must be of type double.");
}
/* Get the number of elements in the input argument. */
elements = mxGetNumberOfElements(prhs[0]);
/* Get the data. */
pr = (double *)mxGetPr(prhs[0]);
pi = (double *)mxGetPi(prhs[0]);
cmplx = ((pi == NULL) ? 0 : 1);
/* Count the number of non-zero elements to be able to allocate
the correct size for output variable. */
for(j = 0; j < elements; j++) {
if(IsNonZero(pr[j]) || (cmplx && IsNonZero(pi[j]))) {
nnz++;
}
}
2-27
2
Creating C Language MEX-Files
/* Get the number of dimensions in the input argument.
Allocate the space for the return argument */
number_of_dims = mxGetNumberOfDimensions(prhs[0]);
plhs[0] = mxCreateDoubleMatrix(nnz, number_of_dims, mxREAL);
pind = mxGetPr(plhs[0]);
/* Get the number of dimensions in the input argument. */
dim_array = mxGetDimensions(prhs[0]);
/* Fill in the indices to return to MATLAB. This loops through
* the elements and checks for non-zero values. If it finds a
* non-zero value, it then calculates the corresponding MATLAB
* indices and assigns them into the output array. The 1 is added
* to the calculated index because MATLAB is 1-based and C is
* 0-based. */
for(j = 0; j < elements; j++) {
if(IsNonZero(pr[j]) || (cmplx && IsNonZero(pi[j]))) {
int temp = j;
int k;
for(k = 0; k < number_of_dims; k++) {
pind[nnz*k+count] = ((temp % (dim_array[k])) + 1);
temp /= dim_array[k];
}
count++;
}
}
}
Entering a sample matrix at the MATLAB prompt gives
matrix = [ 3 0 9 0; 0 8 2 4; 0 9 2 4; 3 0 9 3; 9 9 2 0]
matrix =
3
0
9
0
0
8
2
4
0
9
2
4
3
0
9
3
9
9
2
0
2-28
Examples of C MEX-Files
This example determines the position of all nonzero elements in the matrix.
Running the MEX-file on this matrix produces
nz = findnz(matrix)
nz =
1
1
4
1
5
1
2
2
3
2
5
2
1
3
2
3
3
3
4
3
5
3
2
4
3
4
4
4
Handling Sparse Arrays
The MATLAB API provides a set of functions that allow you to create and
manipulate sparse arrays from within your MEX-files. These API routines
access and manipulate ir and jc, two of the parameters associated with sparse
arrays. For more information on how MATLAB stores sparse arrays, refer to
the section, “The MATLAB Array” on page 1-6.
This example illustrates how to populate a sparse matrix.
/*==============================================================
* fulltosparse.c
* This example demonstrates how to populate a sparse
* matrix. For the purpose of this example, you must pass in a
* non-sparse 2-dimensional argument of type double.
*
*
*
*
*
Comment: You might want to modify this MEX-file so that you can
use it to read large sparse data sets into MATLAB.
This is a MEX-file for MATLAB.
Copyright (c) 1984-2000 The MathWorks, Inc.
2-29
2
Creating C Language MEX-Files
*
*=============================================================*/
/* $Revision: 1.5 $ */
#include <math.h> /* Needed for the ceil() prototype. */
#include "mex.h"
/* If you are using a compiler that equates NaN to be zero, you
* must compile this example using the flag -DNAN_EQUALS_ZERO.
* For example:
*
*
mex -DNAN_EQUALS_ZERO fulltosparse.c
*
* This will correctly define the IsNonZero macro for your C
* compiler.
*/
#if defined(NAN_EQUALS_ZERO)
#define IsNonZero(d) ((d) != 0.0 || mxIsNaN(d))
#else
#define IsNonZero(d) ((d) != 0.0)
#endif
void mexFunction(
int nlhs,
mxArray *plhs[],
int nrhs, const mxArray *prhs[]
)
{
/* Declare variables. */
int j,k,m,n,nzmax,*irs,*jcs,cmplx,isfull;
double *pr,*pi,*si,*sr;
double percent_sparse;
/* Check for proper number of input and output arguments. */
if(nrhs != 1) {
mexErrMsgTxt("One input argument required.");
}
if(nlhs > 1) {
mexErrMsgTxt("Too many output arguments.");
2-30
Examples of C MEX-Files
}
/* Check data type of input argument. */
if(!(mxIsDouble(prhs[0]))) {
mexErrMsgTxt("Input argument must be of type double.");
}
if(mxGetNumberOfDimensions(prhs[0]) != 2) {
mexErrMsgTxt("Input argument must be two dimensional\n");
}
/* Get the size and pointers to input data. */
m = mxGetM(prhs[0]);
n = mxGetN(prhs[0]);
pr = mxGetPr(prhs[0]);
pi = mxGetPi(prhs[0]);
cmplx = (pi == NULL ? 0 : 1);
/* Allocate space for sparse matrix.
* NOTE: Assume at most 20% of the data is sparse.
* to cause it to round up.
*/
Use ceil
percent_sparse = 0.2;
nzmax = (int)ceil((double)m*(double)n*percent_sparse);
plhs[0] = mxCreateSparse(m,n,nzmax,cmplx);
sr = mxGetPr(plhs[0]);
si = mxGetPi(plhs[0]);
irs = mxGetIr(plhs[0]);
jcs = mxGetJc(plhs[0]);
/* Copy nonzeros. */
k = 0;
isfull = 0;
for(j = 0; (j < n ); j++) {
int i;
jcs[j] = k;
for(i = 0; (i < m ); i++) {
2-31
2
Creating C Language MEX-Files
if(IsNonZero(pr[i]) || (cmplx && IsNonZero(pi[i]))) {
/* Check to see if non-zero element will fit in
* allocated output array. If not, increase
* percent_sparse by 10%, recalculate nzmax, and augment
* the sparse array.
*/
if(k >= nzmax) {
int oldnzmax = nzmax;
percent_sparse += 0.1;
nzmax = (int)ceil((double)m*(double)n*percent_sparse);
/* Make sure nzmax increases atleast by 1. */
if(oldnzmax == nzmax)
nzmax++;
mxSetNzmax(plhs[0], nzmax);
mxSetPr(plhs[0], mxRealloc(sr, nzmax*sizeof(double)));
if(si != NULL)
mxSetPi(plhs[0], mxRealloc(si, nzmax*sizeof(double)));
mxSetIr(plhs[0], mxRealloc(irs, nzmax*sizeof(int)));
sr = mxGetPr(plhs[0]);
si = mxGetPi(plhs[0]);
irs = mxGetIr(plhs[0]);
}
sr[k] = pr[i];
if(cmplx) {
si[k] = pi[i];
}
irs[k] = i;
k++;
}
}
pr += m;
pi += m;
}
jcs[n] = k;
}
2-32
Examples of C MEX-Files
At the MATLAB prompt, entering
full = eye(5)
full =
1
0
0
1
0
0
0
0
0
0
0
0
1
0
0
0
0
0
1
0
0
0
0
0
1
creates a full, 5-by-5 identity matrix. Using fulltosparse on the full matrix
produces the corresponding sparse matrix.
spar = fulltosparse(full)
spar =
(1,1)
1
(2,2)
1
(3,3)
1
(4,4)
1
(5,5)
1
Calling Functions from C MEX-Files
It is possible to call MATLAB functions, operators, M-files, and other MEX-files
from within your C source code by using the API function mexCallMATLAB. This
example creates an mxArray, passes various pointers to a subfunction to
acquire data, and calls mexCallMATLAB to calculate the sine function and plot
the results.
/* $Revision: 1.4 $ */
/*=============================================================
* sincall.c
*
* Example for illustrating how to use mexCallMATLAB
*
* Creates an mxArray and passes its associated pointers (in
* this demo, only pointer to its real part, pointer to number of
* rows, pointer to number of columns) to subfunction fill() to
* get data filled up, then calls mexCallMATLAB to calculate sin
* function and plot the result.
*
2-33
2
Creating C Language MEX-Files
* This is a MEX-file for MATLAB.
* Copyright (c) 1984-2000 The MathWorks, Inc.
*============================================================*/
#include "mex.h"
#define MAX 1000
/* Subroutine for filling up data */
void fill( double *pr, int *pm, int *pn, int max )
{
int i;
/* You can fill up to max elements, so (*pr) <= max. */
*pm = max/2;
*pn = 1;
for(i = 0; i < (*pm); i++)
pr[i] = i*(4*3.14159/max);
}
/* Gateway function */
void mexFunction( int nlhs, mxArray *plhs[],
int nrhs, const mxArray *prhs[] )
{
int
m, n, max = MAX;
mxArray *rhs[1], *lhs[1];
rhs[0] = mxCreateDoubleMatrix(max, 1, mxREAL);
/* Pass the pointers and let fill() fill up data. */
fill(mxGetPr(rhs[0]), &m, &n, MAX);
mxSetM(rhs[0], m);
mxSetN(rhs[0], n);
/* Get the sin wave and plot it. */
mexCallMATLAB(1, lhs, 1, rhs, "sin");
mexCallMATLAB(0, NULL, 1, lhs, "plot");
/* Clean up allocated memory. */
mxDestroyArray(rhs[0]);
2-34
Examples of C MEX-Files
mxDestroyArray(lhs[0]);
return;
}
Running this example
sincall
displays the results
Note It is possible to generate an object of type mxUNKNOWN_CLASS using
mexCallMATLAB. See the example below.
The following example creates an M-file that returns two variables but only
assigns one of them a value.
function [a,b] = foo[c]
a = 2*c;
2-35
2
Creating C Language MEX-Files
MATLAB displays the following warning message.
Warning: One or more output arguments not assigned during call to
'foo'.
If you then call foo using mexCallMATLAB, the unassigned output variable will
now be of type mxUNKNOWN_CLASS.
2-36
Advanced Topics
Advanced Topics
These sections cover advanced features of MEX-files that you can use when
your applications require sophisticated MEX-files.
Help Files
• Because the MATLAB interpreter chooses the MEX-file when both an M-file
and a MEX-file with the same name are encountered in the same directory,
it is possible to use M-files for documenting the behavior of your MEX-files.
The MATLAB help command will automatically find and display the
appropriate M-file when help is requested and the interpreter will find and
execute the corresponding MEX-file when the function is invoked.
Linking Multiple Files
It is possible to combine several object files and to use object file libraries when
building MEX-files. To do so, simply list the additional files with their full
extension, separated by spaces. For example, on the PC
mex circle.c square.obj rectangle.c shapes.lib
is a legal command that operates on the .c, .obj, and .lib files to create a
MEX-file called circle.dll, where dll is the extension corresponding to the
MEX-file type on the PC. The name of the resulting MEX-file is taken from the
first file in the list.
You may find it useful to use a software development tool like MAKE to manage
MEX-file projects involving multiple source files. Simply create a MAKEFILE
that contains a rule for producing object files from each of your source files and
then invoke mex to combine your object files into a MEX-file. This way you can
ensure that your source files are recompiled only when necessary.
Workspace for MEX-File Functions
Unlike M-file functions, MEX-file functions do not have their own variable
workspace. MEX-file functions operate in the caller’s workspace.
mexEvalString evaluates the string in the caller’s workspace. In addition, you
can use the mexGetArray and mexPutArray routines to get and put variables
into the caller’s workspace.
2-37
2
Creating C Language MEX-Files
Memory Management
Memory management within MEX-files is not unlike memory management for
regular C or Fortran applications. However, there are special considerations
because the MEX-file must exist within the context of a larger application, i.e.,
MATLAB itself.
Automatic Cleanup of Temporary Arrays
When a MEX-file returns to MATLAB, it gives to MATLAB the results of its
computations in the form of the left-hand side arguments – the mxArrays
contained within the plhs[] list. Any mxArrays created by the MEX-file that
are not in this list are automatically destroyed. In addition, any memory
allocated with mxCalloc, mxMalloc, or mxRealloc during the MEX-file’s
execution is automatically freed.
In general, we recommend that MEX-files destroy their own temporary arrays
and free their own dynamically allocated memory. It is more efficient for the
MEX-file to perform this cleanup than to rely on the automatic mechanism.
However, there are several circumstances in which the MEX-file will not reach
its normal return statement. The normal return will not be reached if:
• A call to mexErrMsgTxt occurs.
• A call to mexCallMATLAB occurs and the function being called creates an
error. (A MEX-file can trap such errors by using mexSetTrapFlag, but not all
MEX-files would necessarily need to trap errors.)
• The user interrupts the MEX-file’s execution using Ctrl-C.
• The MEX-file runs out of memory. When this happens, MATLAB’s
out-of-memory handler will immediately terminate the MEX-file.
A careful MEX-file programmer can ensure safe cleanup of all temporary
arrays and memory before returning in the first two cases, but not in the last
two cases. In the last two cases, the automatic cleanup mechanism is necessary
to prevent memory leaks.
Persistent Arrays
You can exempt an array, or a piece of memory, from MATLAB’s automatic
cleanup by calling mexMakeArrayPersistent or mexMakeMemoryPersistent.
However, if a MEX-file creates such persistent objects, there is a danger that a
memory leak could occur if the MEX-file is cleared before the persistent object
is properly destroyed. In order to prevent this from happening, a MEX-file that
2-38
Advanced Topics
creates persistent objects should register a function, using mexAtExit, which
will dispose of the objects. (You can use a mexAtExit function to dispose of other
resources as well; for example, you can use mexAtExit to close an open file.)
For example, here is a simple MEX-file that creates a persistent array and
properly disposes of it.
#include "mex.h"
static int initialized = 0;
static mxArray *persistent_array_ptr = NULL;
void cleanup(void) {
mexPrintf("MEX-file is terminating, destroying array\n");
mxDestroyArray(persistent_array_ptr);
}
void mexFunction(int nlhs,
mxArray *plhs[],
int nrhs,
const mxArray *prhs[])
{
if(!initialized) {
mexPrintf("MEX-file initializing, creating array\n");
/* Create persistent array and register its cleanup. */
persistent_array_ptr = mxCreateDoubleMatrix(1, 1, mxREAL);
mexMakeArrayPersistent(persistent_array_ptr);
mexAtExit(cleanup);
initialized = 1;
/* Set the data of the array to some interesting value. */
*mxGetPr(persistent_array_ptr) = 1.0;
} else {
mexPrintf("MEX-file executing; value of first array
element is %g\n",
*mxGetPr(persistent_array_ptr));
}
}
2-39
2
Creating C Language MEX-Files
Hybrid Arrays
Functions such as mxSetPr, mxSetData, and mxSetCell allow the direct
placement of memory pieces into an mxArray. mxDestroyArray will destroy
these pieces along with the entire array. Because of this, it is possible to create
an array that cannot be destroyed, i.e., an array on which it is not safe to call
mxDestroyArray. Such an array is called a hybrid array, because it contains
both destroyable and nondestroyable components.
For example, it is not legal to call mxFree (or the ANSI free() function, for that
matter) on automatic variables. Therefore, in the following code fragment,
pArray is a hybrid array.
mxArray *pArray = mxCreateDoubleMatrix(0, 0, mxREAL);
double data[10];
mxSetPr(pArray, data);
mxSetM(pArray, 1);
mxSetN(pArray, 10);
Another example of a hybrid array is a cell array or structure, one of whose
children is a read-only array (an array with the const qualifier, such as one of
the inputs to the MEX-file). The array cannot be destroyed because the input
to the MEX-file would also be destroyed.
Because hybrid arrays cannot be destroyed, they cannot be cleaned up by the
automatic mechanism outlined in “Automatic Cleanup of Temporary Arrays”
on page 2-38. As described in that section, the automatic cleanup mechanism
is the only way to destroy temporary arrays in case of a user interrupt.
Therefore, temporary hybrid arrays are illegal and may cause your MEX-file to
crash.
Although persistent hybrid arrays are viable, we recommend avoiding their
use wherever possible.
Using LAPACK and BLAS Functions
LAPACK is a large, multiauthor Fortran subroutine library that MATLAB
uses for numerical linear algebra. BLAS, which stands for Basic Linear
Algebra Subroutines, is used by MATLAB to speed up matrix multiplication
and the LAPACK routines themselves. The functions provided by LAPACK
and BLAS can also be called directly from within your C MEX-files.
2-40
Advanced Topics
This section explains how to write and build MEX-files that call LAPACK and
BLAS functions. It provides information on:
• Specifying the Function Name
• Passing Arguments to Fortran from C
• Handling Complex Numbers
• Building the MEX-File
It also provides a symmetric indefinite factorization example that uses
functions from LAPACK.
Specifying the Function Name
When calling an LAPACK or BLAS function, some platforms require an
underscore character following the function name in the call statement.
On the PC, IBM_RS, and HP platforms, use the function name alone, with no
trailing underscore. For example, to call the LAPACK dgemm function, use
dgemm (arg1, arg2, ..., argn);
On the SGI, LINUX, Solaris, and Alpha platforms, add the underscore after the
function name. For example, to call dgemm on any of these platforms, use
dgemm_ (arg1, arg2, ..., argn);
Passing Arguments to Fortran from C
Since the LAPACK and BLAS functions are written in Fortran, arguments
passed to and from these functions must be passed by reference. The following
example calls dgemm, passing all arguments by reference. An ampersand (&)
precedes each argument unless that argument is already a reference.
#include "mex.h"
void mexFunction( int nlhs, mxArray *plhs[], int nrhs, mxArray
*prhs[] )
{
double *A, *B, *C, one = 1.0, zero = 0.0;
int m,n,p;
char *chn = "N";
A = mxGetPr(prhs[0]);
2-41
2
Creating C Language MEX-Files
B
m
p
n
=
=
=
=
mxGetPr(prhs[1]);
mxGetM(prhs[0]);
mxGetN(prhs[0]);
mxGetN(prhs[1]);
if(p != mxGetM( prhs[1])) {
mexErrMsgTxt("Inner dimensions of matrix multiply do not
match");
}
plhs[0] = mxCreateDoubleMatrix( m, n, mxREAL );
C = mxGetPr( plhs[0] );
/* Pass all arguments to Fortran by reference */
dgemm (chn, chn, &m, &n, &p, &one, A, &m, B, &p, &zero, C, &m);
}
Handling Complex Numbers
MATLAB stores complex numbers differently than Fortran. MATLAB stores
the real and imaginary parts of a complex number in separate, equal length
vectors, pr and pi. Fortran stores the same number in one location with the
real and imaginary parts interleaved.
As a result, complex variables exchanged between MATLAB and the Fortran
functions in LAPACK and BLAS are incompatible. MATLAB provides
conversion routines that change the storage format of complex numbers to
address this incompatibility.
Input Arguments. For all complex variables passed as input arguments to a
Fortran function, you need to convert the storage of the MATLAB variable to
be compatible with the Fortran function. Use the mat2fort function for this.
See the example that follows.
Output Arguments. For all complex variables passed as output arguments to a
Fortran function, you need to do the following:
1 When allocating storage for the complex variable, allocate a real variable
with twice as much space as you would for a MATLAB variable of the same
size. You need to do this because the returned variable uses the Fortran
2-42
Advanced Topics
format, which takes twice the space. See the allocation of zout in the
example that follows.
2 Once the variable is returned to MATLAB, convert its storage so that it is
compatible with MATLAB. Use the fort2mat function for this.
Example – Passing Complex Variables. The example below shows how to call an
LAPACK function from MATLAB, passing complex prhs[0] as input and
receiving complex plhs[0] as output. Temporary variables zin and zout are
used to hold prhs[0] and plhs[0] in Fortran format.
#include "mex.h"
#include "fort.h"
/* defines mat2fort and fort2mat */
void mexFunction( int nlhs, mxArray *plhs[], int nrhs, mxArray
*prhs[] )
{
int lda, n;
double *zin, *zout;
lda = mxGetM( prhs[0] );
n = mxGetN( prhs[0] );
/* Convert input to Fortran format */
zin = mat2fort( prhs[0], lda, n );
/* Allocate a real, complex, lda-by-n variable to store output */
zout = mxCalloc( 2*lda*n, sizeof(double) );
/* Call complex LAPACK function */
zlapack_function( zin, &lda, &n, zout );
/* Convert output to MATLAB format */
plhs[0] = fort2mat( zout, lda, lda, n );
/* Free intermediate Fortran format arrays */
mxFree( zin );
mxFree( zout );
}
2-43
2
Creating C Language MEX-Files
Preserving Input Values from Modification
Many LAPACK and BLAS functions modify the values of arguments passed in
to them. It is advisable to make a copy of arguments that can be modified prior
to passing them to the function. For complex inputs, this point is moot since the
mat2fort version of the input is a new piece of memory, but for real data this
is not the case.
The following example calls an LAPACK function that modifies the first input
argument. The code in this example makes a copy of prhs[0], and then passes
the copy to the LAPACK function to preserve the contents of prhs[0].
/* lapack_function modifies A so make a copy of the input */
m = mxGetM( prhs[0] );
n = mxGetN( prhs[0] );
A = mxCalloc( m*n, sizeof(double) );
/* Copy mxGetPr( prhs[0] ) into A */
temp = mxGetPr( prhs[0] );
for ( k = 0; k < m*n; k++ ) {
A[k] = temp[k];
}
/* lapack_function does not modify B so it is OK to use the input
directly */
B = mxGetPr( prhs[1] );
lapack_function( A,B );
/* modifies A but not B */
/* Free A when you are done with it */
mxFree(A);
Building the MEX-File
The examples in this section show how to compile and link a C MEX file,
myCmexFile.c, on the platforms supported by MATLAB. In each example, the
term <matlab> stands for the MATLAB root directory.
If you build your C MEX-file on a PC or IBM_RS platform, you need to
explicitly specify a library file to link with.
On the PC, use either one of the following.
mex myCmexFile.c <matlab>/extern/lib/win32/digital/df60/
libmwlapack.lib
2-44
Advanced Topics
mex myCmexFile.c <matlab>/extern/lib/win32/microsoft/msvc60/
libmwlapack.lib
For IBM_RS, use the following syntax.
mex myCmexFile.c -L<matlab>/bin/ibm_rs -lmwlapack
On all other platforms, you can build your MEX-file as you would any other C
MEX-file. For example,
mex myCmexFile.c
MEX-Files That Perform Complex Number Conversion. MATLAB supplies the files
fort.c and fort.h, which provide routines for conversion between MATLAB
and FORTRAN complex data structures. These files define the mat2fort and
fort2mat routines mentioned previously under “Handling Complex Numbers”
on page 2-42.
If your program uses these routines, then you need to:
1 Include the fort.h file in your program, using, #include "fort.h". See the
example above.
2 Build the fort.c file with your program. Specify the pathname, <matlab>/
extern/examples/refbook for both fort.c and fort.h in the build
command, (where <matlab> stands for the MATLAB root directory).
On the PC, use either one of the following.
mex myCmexFile.c <matlab>/extern/examples/refbook/fort.c
-I<matlab>/extern/examples/refbook <matlab>/extern/lib/win32/
digital/df60/libmwlapack.lib
mex myCmexFile.c <matlab>/extern/examples/refbook/fort.c
-I<matlab>/extern/examples/refbook <matlab>/extern/lib/win32/
microsoft/msvc60/libmwlapack.lib
For IBM_RS, use
mex myCmexFile.c <matlab>/extern/examples/refbook/fort.c
-I<matlab>/extern/examples/refbook -L<matlab>/bin/ibm_rs
-lmwlapack
For all other platforms, use
2-45
2
Creating C Language MEX-Files
mex myCmexFile.c <matlab>/extern/examples/refbook/fort.c
-I<matlab>/extern/examples/refbook
Example – Symmetric Indefinite Factorization Using LAPACK
You will find an example C MEX-file that calls two LAPACK functions in the
directory <matlab>/extern/examples/refbook, where <matlab> stands for the
MATLAB root directory. There are two versions of this file:
• utdu_slv.c – calls functions zhesvx and dsysvx, and thus is compatible with
the PC, HP, and IBM platforms.
• utdu_slv_.c – calls functions zhesvx_ and dsysvx_, and thus is compatible
with the SGI, LINUX, Solaris, and Alpha platforms.
2-46
Debugging C Language MEX-Files
Debugging C Language MEX-Files
On most platforms, it is now possible to debug MEX-files while they are
running within MATLAB. Complete source code debugging, including setting
breakpoints, examining variables, and stepping through the source code
line-by-line, is now available.
Note The section entitled, “Troubleshooting” on page 1-32, provides
additional information on isolating problems with MEX-files.
To debug a MEX-file from within MATLAB, you must first compile the
MEX-file with the -g option to mex.
mex -g filename.c
Debugging on UNIX
You need to start MATLAB from within a debugger. To do this, specify the
name of the debugger you want to use with the -D option when starting
MATLAB.
This example shows how to debug yprime.c on Solaris using dbx, the UNIX
debugger.
unix> mex -g yprime.c
unix> matlab -Ddbx
<dbx> stop dlopen <matlab>/extern/examples/mex/yprime.mexsol
Once the debugger loads MATLAB into memory, you can start it by issuing a
run command.
<dbx> run
Now, run the MEX-file that you want to debug as you would ordinarily do
(either directly or by means of some other function or script). Before executing
the MEX-file, you will be returned to the debugger.
>> yprime(1,1:4)
<dbx> stop in ‘yprime.mexsol‘mexFunction
2-47
2
Creating C Language MEX-Files
Note The tick marks used are back ticks (‘), not single quotes (').
You may need to tell the debugger where the MEX-file was loaded or the name
of the MEX-file, in which case MATLAB will display the appropriate command
for you to use. At this point, you are ready to start debugging. You can list the
source code for your MEX-file and set breakpoints in it. It is often convenient
to set one at mexFunction so that you stop at the beginning of the gateway
routine. To proceed from the breakpoint, issue a continue command to the
debugger.
<dbx> cont
Once you hit one of your breakpoints, you can make full use of any facilities
that your debugger provides to examine variables, display memory, or inspect
registers. Refer to the documentation provided with your debugger for
information on its use.
Note For information on debugging on other UNIX platforms, access the
MathWorks Technical Support Web site at http://www.mathworks.com/
support.
Debugging on Windows
The following sections provide instructions on how to debug on Microsoft
Windows systems using various compilers.
Microsoft Compiler
If you are using the Microsoft compiler:
1 Start the Microsoft Visual Studio (Version 5 or 6) by typing at the DOS
prompt
msdev filename.dll
2 In the Microsoft environment, from the Project menu, select Settings. In
the window that opens, select the Debug tab. This options window contains
2-48
Debugging C Language MEX-Files
edit boxes. In the edit box labeled Executable for debug session, enter the
full path to where MATLAB resides. All other edit boxes should be empty.
3 Open the source files and set a break point on the desired line of code by
right-clicking with your mouse on the line of code.
4 From the Build menu, select Debug, and click Go.
5 You will now be able to run your MEX-file in MATLAB and use the Microsoft
debugging environment. For more information on how to debug in the
Microsoft environment, see the Microsoft Development Studio or Microsoft
Visual Studio documentation.
Watcom Compiler
If you are using the Watcom compiler:
1 Start the debugger by typing on the DOS command line
WDW
2 The Watcom Debugger starts and a New Program window opens. In this
window type the full path to MATLAB. For example,
c:\matlab\bin\matlab.exe
Then click OK.
3 From the Break menu, select On Image Load and type the name of your
MEX-file DLL in capital letters. For example,
YPRIME
Then select ADD and click OK to close the window.
4 From the Run menu, select GO. This should start MATLAB.
2-49
2
Creating C Language MEX-Files
5 When MATLAB starts, in the command window change directories to where
your MEX-file resides and run your MEX-file. If a message similar to the
following appears, ignore the message and click OK.
LDR: Automatic DLL Relocation in matlab.exe
LDR: DLL filename.dll base <number> relocated due to collision
with matlab.exe
6 Open the file you want to debug and set breakpoints in the source code.
2-50
3
Creating Fortran
MEX-Files
Fortran MEX-Files . . . . . . . . . . . . . . . . . 3-3
The Components of a Fortran MEX-File . . . . . . . . . 3-3
The %val Construct . . . . . . . . . . . . . . . . . . 3-8
Examples of Fortran MEX-Files . . .
A First Example — Passing a Scalar . . .
Passing Strings . . . . . . . . . . .
Passing Arrays of Strings . . . . . . .
Passing Matrices . . . . . . . . . . .
Passing Two or More Inputs or Outputs .
Handling Complex Data . . . . . . . .
Dynamically Allocating Memory . . . .
Handling Sparse Matrices . . . . . . .
Calling Functions from Fortran MEX-Files
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 3-9
. 3-10
. 3-12
. 3-14
. 3-17
. 3-19
. 3-22
. 3-25
. 3-28
. 3-32
Advanced Topics . . . . . . .
Help Files . . . . . . . . . .
Linking Multiple Files . . . . .
Workspace for MEX-File Functions
Memory Management . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 3-36
. 3-36
. 3-36
. 3-36
. 3-37
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Debugging Fortran Language MEX-Files . . . . . . . 3-38
Debugging on UNIX . . . . . . . . . . . . . . . . . 3-38
Debugging on Windows . . . . . . . . . . . . . . . . 3-39
3
Creating Fortran MEX-Files
This chapter describes how to write MEX-files in the Fortran programming
language. It discusses the MEX-file itself, how these Fortran language files
interact with MATLAB, how to pass and manipulate arguments of different
data types, how to debug your MEX-file programs, and several other, more
advanced topics.
The following list summarizes the contents of this chapter:
• “Fortran MEX-Files”
• “Examples of Fortran MEX-Files”
• “Advanced Topics”
• “Debugging Fortran Language MEX-Files”
3-2
Fortran MEX-Files
Fortran MEX-Files
Fortran MEX-files are built by using the mex script to compile your Fortran
source code with additional calls to API routines.
MEX-files in Fortran can only create double-precision data and strings, unlike
their C counterparts, which can create any data type supported by MATLAB.
You can treat Fortran MEX-files, once compiled, exactly like M-functions.
The Components of a Fortran MEX-File
This section discusses the specific elements needed in a Fortran MEX-file. The
source code for a Fortran MEX-file, like the C MEX-file, consists of two distinct
parts:
• A computational routine that contains the code for performing the
computations that you want implemented in the MEX-file. Computations
can be numerical computations as well as inputting and outputting data.
• A gateway routine that interfaces the computational routine with MATLAB
by the entry point mexFunction and its parameters prhs, nrhs, plhs, nlhs,
where prhs is an array of right-hand input arguments, nrhs is the number
of right-hand input arguments, plhs is an array of left-hand output
arguments, and nlhs is the number of left-hand output arguments. The
gateway calls the computational routine as a subroutine.
The computational and gateway routines may be separate or combined. The
following figure, Fortran MEX Cycle, shows how inputs enter an API function,
what functions the gateway routine performs, and how output returns to
MATLAB.
3-3
3
Creating Fortran MEX-Files
MATLAB
A call to
MEX-file func:
[C,D]=func(A,B)
tells MATLAB to
pass variables A and B
to your MEX-file. C
and D are left
unassigned.
INPUTS
integer B
B = prhs(2)
integer A
A = prhs(1)
func.f
subroutine mexFunction(
nlhs, plhs, nrhs, prhs)
integer plhs(*), prhs(*), nlhs, nrhs
In the gateway routine:
• Use the mxCreate functions to create the
MATLAB arrays for your output
arguments. Set plhs(1), (2), … to the
pointers to the newly created MATLAB
arrays.
• Use the mxGet functions to extract your
data from prhs(1), (2), …
MATLAB
On return from
MEX-file func:
• Call your Fortran subroutine passing
the input and output data pointers as
function parameters using %val.
integer D
D = plhs(2)
[C,D]=func(A,B)
plhs(1) is assigned
to C and plhs(2) is
assigned to D.
integer C
C = plhs(1)
OUTPUTS
Figure 3-1: Fortran MEX Cycle
3-4
Fortran MEX-Files
The Pointer Concept
The MATLAB API works with a unique data type, the mxArray. Because there
is no way to create a new data type in Fortran, MATLAB passes a special
identifier, called a pointer, to a Fortran program. You can get information
about an mxArray by passing this pointer to various API functions called
Access Routines. These access routines allow you to get a native Fortran data
type containing exactly the information you want, i.e., the size of the mxArray,
whether or not it is a string, or its data contents.
There are several implications when using pointers in Fortran:
• The %val construct.
If your Fortran compiler supports the %val construct, then there is one type
of pointer you can use without requiring an access routine, namely a pointer
to data (i.e., the pointer returned by mxGetPr or mxGetPi). You can use %val
to pass this pointer’s contents to a subroutine, where it is declared as a
Fortran double-precision matrix.
If your Fortran compiler does not support the %val construct, you must use
the mxCopy__ routines (e.g., mxCopyPtrToReal8) to access the contents of the
pointer. For more information about the %val construct and an example, see
the section, “The %val Construct” on page 3-8.
• Variable declarations.
To use pointers properly, you must declare them to be the correct size. On
DEC Alpha machines, all pointers should be declared as integer*8. On all
other platforms, pointers should be declared as integer*4.
If your Fortran compiler supports preprocessing with the C preprocessor, you
can use the preprocessing stage to map pointers to the appropriate
declaration. In UNIX, see the examples ending with .F in the examples
directory for a possible approach.
Caution Declaring a pointer to be the incorrect size may cause your program
to crash.
3-5
3
Creating Fortran MEX-Files
The Gateway Routine
The entry point to the gateway subroutine must be named mexFunction and
must contain these parameters.
subroutine mexFunction(nlhs, plhs, nrhs, prhs)
integer plhs(*), prhs(*)
integer nlhs, nrhs
Note Fortran is case-insensitive. This document uses mixed-case function
names for ease of reading.
In a Fortran MEX-file, the parameters nlhs and nrhs contain the number of
left- and right-hand arguments with which the MEX-file is invoked. prhs is a
length nrhs array that contains pointers to the right-hand side inputs to the
MEX-file, and plhs is a length nlhs array that contains pointers to the
left-hand side outputs that your Fortran function generates.
In the syntax of the MATLAB language, functions have the general form
[a,b,c,…] = fun(d,e,f,…)
where the ellipsis (…) denotes additional terms of the same format. The a,b,c,…
are left-hand arguments and the d,e,f,… are right-hand arguments.
As an example of the gateway routine, consider invoking a MEX-file from the
MATLAB workspace with the command
x = fun(y,z);
the MATLAB interpreter calls mexFunction with the arguments
nlhs = 1
nrhs = 2
plhs
prhs
3-6
•
•
•
Ø
Y
Z
Fortran MEX-Files
plhs is a 1-element C array where the single element is a null pointer. prhs is
a 2-element C array where the first element is a pointer to an mxArray named
Y and the second element is a pointer to an mxArray named Z.
The parameter plhs points at nothing because the output x is not created until
the subroutine executes. It is the responsibility of the gateway routine to create
an output array and to set a pointer to that array in plhs(1). If plhs(1) is left
unassigned, MATLAB prints a warning message stating that no output has
been assigned.
Note It is possible to return an output value even if nlhs = 0. This
corresponds to returning the result in the ans variable.
The gateway routine should validate the input arguments and call
mexErrMsgTxt if anything is amiss. This step includes checking the number,
type, and size of the input arrays as well as examining the number of output
arrays. The examples included later in this section illustrate this technique.
The mx functions provide a set of access methods (subroutines) for
manipulating MATLAB arrays. These functions are fully documented in the
online API reference pages. The mx prefix is shorthand for mxArray and it
means that the function enables you to access and/or manipulate some of the
information in the MATLAB array. For example, mxGetPr gets the real data
from the MATLAB array. Additional routines are provided for transferring
data between MATLAB arrays and Fortran arrays.
The gateway routine must call mxCreateFull, mxCreateSparse, or
mxCreateString to create MATLAB arrays of the required sizes in which to
return the results. The return values from these calls should be assigned to the
appropriate elements of plhs.
The gateway routine may call mxCalloc to allocate temporary work arrays for
the computational routine if it needs them.
The gateway routine should call the computational routine to perform the
desired calculations or operations. There are a number of additional routines
that MEX-files can use. These routines are distinguished by the initial
characters mex, as in mexCallMATLAB and mexErrMsgTxt.
3-7
3
Creating Fortran MEX-Files
When a MEX-file completes its task, it returns control to MATLAB. Any
MATLAB arrays that are created by the MEX-file that are not returned to
MATLAB through the left-hand side arguments are automatically destroyed.
The %val Construct
The %val construct is supported by most, but not all, Fortran compilers.
DIGITAL Visual Fortran does support the construct. %val causes the value of
the variable, rather than the address of the variable, to be passed to the
subroutine. If you are using a Fortran compiler that does not support the %val
construct, you must copy the array values into a temporary true Fortran array
using special routines. For example, consider a gateway routine that calls its
computational routine, yprime, by
call yprime(%val(yp), %val(t), %val(y))
If your Fortran compiler does not support the %val construct, you would replace
the call to the computational subroutine with
C Copy array pointers to local arrays.
call mxCopyPtrToReal8(t, tr, 1)
call mxCopyPtrToReal8(y, yr, 4)
C
C Call the computational subroutine.
call yprime(ypr, tr, yr)
C
C Copy local array to output array pointer.
call mxCopyReal8ToPtr(ypr, yp, 4)
You must also add the following declaration line to the top of the gateway
routine.
real*8 ypr(4), tr, yr(4)
Note that if you use mxCopyPtrToReal8 or any of the other mxCopy__ routines,
the size of the arrays declared in the Fortran gateway routine must be greater
than or equal to the size of the inputs to the MEX-file coming in from MATLAB.
Otherwise mxCopyPtrToReal8 will not work correctly.
3-8
Examples of Fortran MEX-Files
Examples of Fortran MEX-Files
The following sections include information and examples describing how to
pass and manipulate the different data types when working with MEX-files.
These topics include:
• “A First Example — Passing a Scalar”
• “Passing Strings”
• “Passing Arrays of Strings”
• “Passing Matrices”
• “Passing Two or More Inputs or Outputs”
• “Handling Complex Data”
• “Dynamically Allocating Memory”
• “Handling Sparse Matrices”
• “Calling Functions from Fortran MEX-Files”
The MATLAB API provides a set of routines for Fortran that handle
double-precision data and strings in MATLAB. For each data type, there is a
specific set of functions that you can use for data manipulation.
Note to UNIX Users The example Fortran files in the directory <matlab>/
extern/examples/refbook have extensions .F and .f. The distinction
between these extensions is that the .F files need to be preprocessed.
Note You can find the most recent versions of the example programs from
this chapter at the anonymous FTP server,
ftp://ftp.mathworks.com/pub/tech-support/docexamples/apiguide/R12/refbook
3-9
3
Creating Fortran MEX-Files
A First Example — Passing a Scalar
Let’s look at a simple example of Fortran code and its MEX-file equivalent.
Here is a Fortran computational routine that takes a scalar and doubles it.
subroutine timestwo(y, x)
real*8 x, y
C
y = 2.0 * x
return
end
Below is the same function written in the MEX-file format.
C-------------------------------------------------------------C
timestwo.f
C
C
Multiply the input argument by 2.
C
C
C
This is a MEX-file for MATLAB.
Copyright (c) 1984-2000 The MathWorks, Inc.
$Revision: 1.11 $
subroutine mexFunction(nlhs, plhs, nrhs, prhs)
C-------------------------------------------------------------C
(pointer) Replace integer by integer*8 on the DEC Alpha
C
platform.
C
integer plhs(*), prhs(*)
integer mxGetPr, mxCreateFull
integer x_pr, y_pr
C-------------------------------------------------------------C
integer
integer
integer
real*8
3-10
nlhs, nrhs
mxGetM, mxGetN, mxIsNumeric
m, n, size
x, y
Examples of Fortran MEX-Files
C
Check for proper number of arguments.
if(nrhs .ne. 1) then
call mexErrMsgTxt('One input required.')
elseif(nlhs .ne. 1) then
call mexErrMsgTxt('One output required.')
endif
C
Get the size of the input array.
m = mxGetM(prhs(1))
n = mxGetN(prhs(1))
size = m*n
C
Check to ensure the input is a number.
if(mxIsNumeric(prhs(1)) .eq. 0) then
call mexErrMsgTxt('Input must be a number.')
endif
C
Create matrix for the return argument.
plhs(1) = mxCreateFull(m, n, 0)
x_pr = mxGetPr(prhs(1))
y_pr = mxGetPr(plhs(1))
call mxCopyPtrToReal8(x_pr, x, size)
C
Call the computational subroutine.
call timestwo(y, x)
C
Load the data into y_pr, which is the output to MATLAB.
call mxCopyReal8ToPtr(y, y_pr, size)
return
end
subroutine timestwo(y, x)
real*8 x, y
C
y = 2.0 * x
return
end
3-11
3
Creating Fortran MEX-Files
To compile and link this example source file, at the MATLAB prompt type
mex timestwo.f
This carries out the necessary steps to create the MEX-file called timestwo
with an extension corresponding to the machine type on which you’re running.
You can now call timestwo as if it were an M-function.
x = 2;
y = timestwo(x)
y =
4
Passing Strings
Passing strings from MATLAB to a Fortran MEX-file is straightforward. This
program accepts a string and returns the characters in reverse order.
C
$Revision: 1.14 $
C==============================================================
C
C
revord.f
C
Example for illustrating how to copy string data from
C
MATLAB to a Fortran-style string and back again.
C
C
Takes a string and returns a string in reverse order.
C
C
This is a MEX-file for MATLAB.
C
Copyright (c) 1984-2000 The MathWorks, Inc.
C==============================================================
10
3-12
subroutine revord(input_buf, strlen, output_buf)
character input_buf(*), output_buf(*)
integer
i, strlen
do 10 i=1,strlen
output_buf(i) = input_buf(strlen-i+1)
continue
return
end
Examples of Fortran MEX-Files
Below is the gateway routine that calls the computational routine.
C
The gateway routine
subroutine mexFunction(nlhs, plhs, nrhs, prhs)
integer nlhs, nrhs
C-------------------------------------------------------------C
(pointer) Replace integer by integer*8 on the DEC Alpha
C
platform.
C
integer plhs(*), prhs(*)
integer mxCreateString, mxGetString
C-------------------------------------------------------------C
integer mxGetM, mxGetN, mxIsString
integer status, strlen
character*100 input_buf, output_buf
C
C
C
Check for proper number of arguments.
if (nrhs .ne. 1) then
call mexErrMsgTxt('One input required.')
elseif (nlhs .gt. 1) then
call mexErrMsgTxt('Too many output arguments.')
The input must be a string.
elseif(mxIsString(prhs(1)) .ne. 1) then
call mexErrMsgTxt('Input must be a string.')
The input must be a row vector.
elseif (mxGetM(prhs(1)) .ne. 1) then
call mexErrMsgTxt('Input must be a row vector.')
endif
C
Get the length of the input string.
strlen = mxGetM(prhs(1))*mxGetN(prhs(1))
C
Get the string contents (dereference the input integer).
status = mxGetString(prhs(1), input_buf, 100)
C
Check if mxGetString is successful.
if (status .ne. 0) then
call mexErrMsgTxt('String length must be less than 100.')
endif
3-13
3
Creating Fortran MEX-Files
C
C
Initialize outbuf_buf to blanks. This is necessary on some
compilers.
output_buf = ' '
C
Call the computational subroutine.
call revord(input_buf, strlen, output_buf)
C
Set output_buf to MATLAB mexFunction output.
plhs(1) = mxCreateString(output_buf)
return
end
After checking for the correct number of inputs, this MEX-file gateway routine
verifies that the input was either a row or column vector string. It then finds
the size of the string and places the string into a Fortran character array. Note
that in the case of character strings, it is not necessary to copy the data into a
Fortran character array by using mxCopyPtrToCharacter. In fact,
mxCopyPtrToCharacter works only with MAT-files. For more information
about MAT-files, see Chapter 6, “Importing and Exporting Data.”
For an input string
x = 'hello world';
typing
y = revord(x)
produces
y =
dlrow olleh
Passing Arrays of Strings
Passing arrays of strings adds a slight complication to the example in the
previous section, “Passing Strings”. Because MATLAB stores elements of a
matrix by column instead of by row, it is essential that the size of the string
array be correctly defined in the Fortran MEX-file. The key point is that the
row and column sizes as defined in MATLAB must be reversed in the Fortran
3-14
Examples of Fortran MEX-Files
MEX-file. Consequently, when returning to MATLAB, the output matrix must
be transposed.
This example places a string array/character matrix into MATLAB as output
arguments rather than placing it directly into the workspace. Inside MATLAB,
call this function by typing
passstr;
You will get the matrix mystring of size 5-by-15. There are some manipulations
that need to be done here. The original string matrix is of the size 5-by-15.
Because of the way MATLAB reads and orients elements in matrices, the size
of the matrix must be defined as M=15 and N=5 from the MEX-file. After the
matrix is put into MATLAB, the matrix must be transposed.
C
$Revision: 1.11 $
C==============================================================
C
C
passstr.f
C
Example for illustrating how to pass a character matrix
C
from Fortran to MATLAB.
C
C
Passes a string array/character matrix into MATLAB as
C
output arguments rather than placing it directly into the
C
workspace.
C
C
This is a MEX-file for MATLAB.
C
Copyright (c) 1984-2000 The MathWorks, Inc.
C==============================================================
subroutine mexFunction(nlhs, plhs, nrhs, prhs)
C-------------------------------------------------------------C
(pointer) Replace integer by integer*8 on the DEC Alpha
C
platform.
C
integer plhs(*), prhs(*)
integer p_str, mxCreateString
C-------------------------------------------------------------C
integer nlhs, nrhs
integer i
character*75 thestring
3-15
3
Creating Fortran MEX-Files
character*15 string(5)
C
Create the string to passed into MATLAB.
string(1) = 'MATLAB
'
string(2) = 'The Scientific '
string(3) = 'Computing
'
string(4) = 'Environment
'
string(5) = '
by TMW, Inc.'
C
Concatenate the set of 5 strings into a long string.
thestring = string(1)
do 10 i = 2, 6
thestring = thestring(:((i-1)*15)) // string(i)
continue
10
C
C
Create the string matrix to be passed into MATLAB.
Set the matrix size to be M=15 and N=5.
p_str = mxcreatestring(thestring)
call mxSetM(p_str, 15)
call mxSetN(p_str, 5)
C
Transpose the resulting matrix in MATLAB.
call mexCallMATLAB(1, plhs, 1, p_str, 'transpose')
return
end
Typing
passstr
at the MATLAB prompt produces this result
ans =
MATLAB
The Scientific
Computing
Environment
by TMW, Inc.
3-16
Examples of Fortran MEX-Files
Passing Matrices
In MATLAB, you can pass matrices into and out of MEX-files written in
Fortran. You can manipulate the MATLAB arrays by using mxGetPr and
mxGetPi to assign pointers to the real and imaginary parts of the data stored
in the MATLAB arrays. You can create new MATLAB arrays from within your
MEX-file by using mxCreateFull.
This example takes a real 2-by-3 matrix and squares each element.
C-------------------------------------------------------------C
C
matsq.f
C
C
Squares the input matrix
C
This is a MEX-file for MATLAB.
C
Copyright (c) 1984-2000 The MathWorks, Inc.
C
$Revision: 1.12 $
C-------------------------------------------------------------subroutine matsq(y, x, m, n)
real*8 x(m,n), y(m,n)
integer m, n
C
10
20
do 20 i=1,m
do 10 j=1,n
y(i,j)= x(i,j)**2
continue
continue
return
end
This is the gateway routine that calls the computational subroutine.
subroutine mexFunction(nlhs, plhs, nrhs, prhs)
C-------------------------------------------------------------C
(pointer) Replace integer by integer*8 on the DEC Alpha
C
platform.
C
integer plhs(*), prhs(*)
integer mxCreateFull, mxGetPr
3-17
3
Creating Fortran MEX-Files
integer x_pr, y_pr
C-------------------------------------------------------------C
integer nlhs, nrhs
integer mxGetM, mxGetN, mxIsNumeric
integer m, n, size
real*8 x(1000), y(1000)
3-18
C
Check for proper number of arguments.
if(nrhs .ne. 1) then
call mexErrMsgTxt('One input required.')
elseif(nlhs .ne. 1) then
call mexErrMsgTxt('One output required.')
endif
C
Get the size of the input array.
m = mxGetM(prhs(1))
n = mxGetN(prhs(1))
size = m*n
C
Column * row should be smaller than 1000.
if(size.gt.1000) then
call mexErrMsgTxt('Row * column must be <= 1000.')
endif
C
Check to ensure the array is numeric (not strings).
if(mxIsNumeric(prhs(1)) .eq. 0) then
call mexErrMsgTxt('Input must be a numeric array.')
endif
C
Create matrix for the return argument.
plhs(1) = mxCreateFull(m, n, 0)
x_pr = mxGetPr(prhs(1))
y_pr = mxGetPr(plhs(1))
call mxCopyPtrToReal8(x_pr, x, size)
C
Call the computational subroutine.
call matsq(y, x, m, n)
Examples of Fortran MEX-Files
C
Load the data into y_pr, which is the output to MATLAB.
call mxCopyReal8ToPtr(y, y_pr, size)
return
end
After performing error checking to ensure that the correct number of inputs
and outputs was assigned to the gateway subroutine and to verify the input
was in fact a numeric matrix, matsq.f creates a matrix for the argument
returned from the computational subroutine. The input matrix data is then
copied to a Fortran matrix by using mxCopyPtrToReal8. Now the computational
subroutine can be called, and the return argument can then be placed into
y_pr, the pointer to the output, using mxCopyReal8ToPtr.
For a 2-by-3 real matrix
x = [1 2 3; 4 5 6];
typing
y = matsq(x)
produces this result
y =
1
16
4
25
9
36
Passing Two or More Inputs or Outputs
The plhs and prhs parameters are vectors that contain pointers to each
left-hand side (output) variable and right-hand side (input) variable.
Accordingly, plhs(1) contains a pointer to the first left-hand side argument,
plhs(2) contains a pointer to the second left-hand side argument, and so on.
Likewise, prhs(1) contains a pointer to the first right-hand side argument,
prhs(2) points to the second, and so on.
For example, this routine multiplies an input scalar times an input scalar or
matrix. This is the Fortran code for the computational subroutine.
subroutine xtimesy(x, y, z, m, n)
real*8 x, y(3,3), z(3,3)
integer m, n
3-19
3
Creating Fortran MEX-Files
10
20
do 20 i=1,m
do 10 j=1,n
z(i,j)=x*y(i,j)
continue
continue
return
end
Below is the gateway routine that calls xtimesy, the computation subroutine
that multiplies a scalar by a scalar or matrix.
C-------------------------------------------------------------C
C
xtimesy.f
C
C
Multiply the first input by the second input.
C
C
C
This is a MEX file for MATLAB.
Copyright (c) 1984-2000 The MathWorks, Inc.
$Revision: 1.11 $
subroutine mexFunction(nlhs, plhs, nrhs, prhs)
C-------------------------------------------------------------C
(pointer) Replace integer by integer*8 on the DEC Alpha
C
platform.
C
integer plhs(*), prhs(*)
integer mxCreateFull
integer x_pr, y_pr, z_pr
C-------------------------------------------------------------C
integer nlhs, nrhs
integer m, n, size
integer mxGetM, mxGetN, mxIsNumeric
real*8 x, y(3,3), z(3,3)
C
3-20
Check for proper number of arguments.
if (nrhs .ne. 2) then
call mexErrMsgTxt('Two inputs required.')
elseif (nlhs .ne. 1) then
call mexErrMsgTxt('One output required.')
Examples of Fortran MEX-Files
endif
C
Check to see both inputs are numeric.
if (mxIsNumeric(prhs(1)) .ne. 1) then
call mexErrMsgTxt('Input # 1 is not a numeric.')
elseif (mxIsNumeric(prhs(2)) .ne. 1) then
call mexErrMsgTxt('Input #2 is not a numeric array.')
endif
C
Check that input #1 is a scalar.
m = mxGetM(prhs(1))
n = mxGetN(prhs(1))
if(n .ne. 1 .or. m .ne. 1) then
call mexErrMsgTxt('Input #1 is not a scalar.')
endif
C
Get the size of the input matrix.
m = mxGetM(prhs(2))
n = mxGetN(prhs(2))
size = m*n
C
Create matrix for the return argument.
plhs(1) = mxCreateFull(m, n, 0)
x_pr = mxGetPr(prhs(1))
y_pr = mxGetPr(prhs(2))
z_pr = mxGetPr(plhs(1))
C
Load the data into Fortran arrays.
call mxCopyPtrToReal8(x_pr, x, 1)
call mxCopyPtrToReal8(y_pr, y, size)
C
Call the computational subroutine.
call xtimesy(x, y, z, m, n)
C
Load the output into a MATLAB array.
call mxCopyReal8ToPtr(z, z_pr, size)
return
end
3-21
3
Creating Fortran MEX-Files
As this example shows, creating MEX-file gateways that handle multiple
inputs and outputs is straightforward. All you need to do is keep track of which
indices of the vectors prhs and plhs correspond to which input and output
arguments of your function. In this example, the input variable x corresponds
to prhs(1) and the input variable y to prhs(2).
For an input scalar x and a real 3-by-3 matrix,
x = 3; y = ones(3);
typing
z = xtimesy(x, y)
yields this result
z =
3
3
3
3
3
3
3
3
3
Handling Complex Data
MATLAB stores complex double-precision data as two vectors of numbers —
one contains the real data and one contains the imaginary data. The API
provides two functions, mxCopyPtrToComplex16 and mxCopyComplex16ToPtr,
which allow you to copy the MATLAB data to a native complex*16 Fortran
array.
This example takes two complex vectors (of length 3) and convolves them.
C
$Revision: 1.14 $
C==============================================================
C
C
convec.f
C
Example for illustrating how to pass complex data from
C
MATLAB to FORTRAN (using COMPLEX data type) and back
C
again.
C
C
Convolves two complex input vectors.
C
C
This is a MEX-file for MATLAB.
C
Copyright (c) 1984-2000 The MathWorks, Inc.
3-22
Examples of Fortran MEX-Files
C==============================================================
C
C
Computational subroutine
subroutine convec(x, y, z, nx, ny)
complex*16 x(*), y(*), z(*)
integer nx, ny
C
10
20
30
Initialize the output array.
do 10 i=1,nx+ny-1
z(i) = (0.0,0.0)
continue
do 30 i=1,nx
do 20 j=1,ny
z(i+j-1) = z(i+j-1) + x(i) * y(j)
continue
continue
return
end
C
The gateway routine.
subroutine mexFunction(nlhs, plhs, nrhs, prhs)
integer nlhs, nrhs
C-------------------------------------------------------------C
(pointer) Replace integer by integer*8 on the DEC Alpha
C
platform.
C
integer plhs(*), prhs(*)
integer mxGetPr, mxGetPi, mxCreateFull
C-------------------------------------------------------------C
integer mx, nx, my, ny, nz
integer mxGetM, mxGetN, mxIsComplex
complex*16 x(100), y(100), z(199)
C
Check for proper number of arguments.
if (nrhs .ne. 2) then
call mexErrMsgTxt('Two inputs required.')
elseif (nlhs .gt. 1) then
call mexErrMsgTxt('Too many output arguments.')
3-23
3
Creating Fortran MEX-Files
endif
C
Check that inputs are both row vectors.
mx = mxGetM(prhs(1))
nx = mxGetN(prhs(1))
my = mxGetM(prhs(2))
ny = mxGetN(prhs(2))
nz = nx+ny-1
C
Only handle row vector input.
if(mx .ne. 1 .or. my .ne. 1) then
call mexErrMsgTxt('Both inputs must be row vector.')
Check sizes of the two input.
elseif(nx .gt. 100 .or. ny .gt. 100) then
call mexErrMsgTxt('Inputs must have less than 100
elements.')
Check to see both inputs are complex.
elseif ((mxIsComplex(prhs(1)) .ne. 1) .or. +
(mxIsComplex(prhs(2)) .ne. 1)) then
call mexErrMsgTxt('Inputs must be complex.')
endif
C
C
C
Create the output array.
plhs(1) = mxCreateFull(1, nz, 1)
C
Load the data into Fortran arrays(native COMPLEX data).
call mxCopyPtrToComplex16(mxGetPr(prhs(1)),
mxGetPi(prhs(1)), x, nx)
call mxCopyPtrToComplex16(mxGetPr(prhs(2)),
mxGetPi(prhs(2)), y, ny)
C
Call the computational subroutine.
call convec(x, y, z, nx, ny)
C
Load the output into a MATLAB array.
call mxCopyComplex16ToPtr(z,mxGetPr(plhs(1)),
mxGetPi(plhs(1)), nz)
return
end
3-24
Examples of Fortran MEX-Files
Entering these numbers at the MATLAB prompt
x = [3 - 1i, 4 + 2i, 7 - 3i]
x =
3.0000 - 1.0000i
4.0000 + 2.0000i
7.0000 - 3.0000i
y = [8 - 6i, 12 + 16i, 40 - 42i]
y =
8.0000 - 6.0000i
12.0000 +16.0000i
40.0000 -42.0000i
and invoking the new MEX-file
z = convec(x, y)
results in
z =
1.0e+02 *
Columns 1 through 4
0.1800 - 0.2600i
3.7600 - 0.1200i
0.9600 + 0.2800i
1.3200 - 1.4400i
Column 5
1.5400 - 4.1400i
which agrees with the results the built-in MATLAB function conv.m produces.
Dynamically Allocating Memory
It is possible to allocate memory dynamically in a Fortran MEX-file, but you
must use %val to do it. This example takes an input matrix of real data and
doubles each of its elements.
3-25
3
Creating Fortran MEX-Files
C
$Revision: 1.12 $
C===============================================================
C
C
dblmat.f
C
Example for illustrating how to use %val.
C
Doubles the input matrix. The demo only handles real part
C
of input.
C
NOTE: If your Fortran compiler does not support %val,
C
use mxCopy_routine.
C
C
NOTE: The subroutine compute() is in the file called
C
compute.f.
C
C
This is a MEX-file for MATLAB.
C
Copyright (c) 1984-2000 The MathWorks, Inc.
C
C===============================================================
C
Gateway subroutine
subroutine mexfunction(nlhs, plhs, nrhs, prhs)
C--------------------------------------------------------------C
(pointer) Replace integer by integer*8 on the DEC Alpha
C
platform.
C
integer plhs(*), prhs(*)
integer pr_in, pr_out
integer mxGetPr, mxCreateFull
C--------------------------------------------------------------C
integer nlhs, nrhs, mxGetM, mxGetN
integer m_in, n_in, size
if(nrhs .ne. 1) then
call mexErrMsgTxt('One input required.')
endif
if(nlhs .gt. 1) then
call mexErrMsgTxt('Cannot return more than one output.')
endif
3-26
Examples of Fortran MEX-Files
m_in = mxGetM(prhs(1))
n_in = mxGetN(prhs(1))
size = m_in * n_in
pr_in = mxGetPr(prhs(1))
mxCreateFull dynamically allocates memory.
plhs(1) = mxCreateFull(m_in, n_in, 0)
pr_out = mxGetPr(plhs(1))
C
C
Call the computational routine.
call compute(%val(pr_out), %val(pr_in), size)
return
end
C
$Revision: 1.3 $
C===============================================================
C
C
compute.f
C
C
This subroutine doubles the input matrix. Your version of
C
compute() may do whaveter you would like it to do.
C
C
This is a MEX-file for MATLAB.
C
Copyright (c) 1984-2000 The MathWorks, Inc.
C
C===============================================================
C
Computational subroutine
subroutine compute(out_mat, in_mat, size)
integer size, i
real*8 out_mat(*), in_mat(*)
10
do 10 i=1,size
out_mat(i) = 2*in_mat(i)
continue
return
end
3-27
3
Creating Fortran MEX-Files
For an input 2-by-3 matrix
x = [1 2 3; 4 5 6];
typing
y = dblmat(x)
yields
y =
2
8
4
10
6
12
Note The dblmat.f example, as well as fulltosparse.f and sincall.f, are
split into two parts, the gateway and the computational subroutine, because of
restrictions in some compilers.
Handling Sparse Matrices
The MATLAB API provides a set of functions that allow you to create and
manipulate sparse matrices from within your MEX-files. There are special
parameters associated with sparse matrices, namely ir, jc, and nzmax. For
information on how to use these parameters and how MATLAB stores sparse
matrices in general, refer to the section on “The MATLAB Array” on page 1-6.
Note Sparse array indexing is zero-based, not one-based.
This example illustrates how to populate a sparse matrix.
C
$Revision: 1.6 $
C===============================================================
C
C
fulltosparse.f
C
Example for illustrating how to populate a sparse matrix.
C
For the purpose of this example, you must pass in a
C
non-sparse 2-dimensional argument of type real double.
3-28
Examples of Fortran MEX-Files
C
C
NOTE: The subroutine loadsparse() is in the file called
C
loadsparse.f.
C
This is a MEX-file for MATLAB.
C
Copyright (c) 1984-2000 The MathWorks, Inc.
C
C===============================================================
C
The gateway routine.
subroutine mexFunction(nlhs, plhs, nrhs, prhs)
integer nlhs, nrhs
C--------------------------------------------------------------C
(pointer) Replace integer by integer*8 on the DEC Alpha
C
64-bit platform
C
integer plhs(*), prhs(*)
integer mxGetPr, mxCreateSparse, mxGetIr, mxGetJc
integer pr, sr, irs, jcs
C--------------------------------------------------------------C
integer m, n, nzmax
integer mxGetM, mxGetN, mxIsComplex, mxIsDouble
integer loadsparse
C
Check for proper number of arguments.
if (nrhs .ne. 1) then
call mexErrMsgTxt('One input argument required.')
endif
if (nlhs .gt. 1) then
call mexErrMsgTxt('Too many output arguments.')
endif
C
Check data type of input argument.
if (mxIsDouble(prhs(1)) .eq. 0) then
call mexErrMsgTxt('Input argument must be of type double.')
endif
if (mxIsComplex(prhs(1)) .eq. 1) then
call mexErrMsgTxt('Input argument must be real only')
endif
3-29
3
Creating Fortran MEX-Files
C
C
C
C
C
C
Get the size and pointers to input data.
m = mxGetM(prhs(1))
n = mxGetN(prhs(1))
pr = mxGetPr(prhs(1))
Allocate space.
NOTE: Assume at most 20% of the data is sparse.
nzmax = dble(m*n) *.20 + .5
NOTE: The maximum number of non-zero elements cannot be less
than the number of columns in the matrix.
if (n .gt. nzmax) then
nzmax = n
endif
plhs(1) = mxCreateSparse(m,n,nzmax,0)
sr = mxGetPr(plhs(1))
irs = mxGetIr(plhs(1))
jcs = mxGetJc(plhs(1))
Load the sparse data.
if (loadsparse(%val(pr),%val(sr),%val(irs),%val(jcs),
+m,n,nzmax) .eq. 1) then
call mexPrintf('Truncating output, input is > 20%% sparse')
endif
return
end
This is the subroutine that fulltosparse calls to fill the mxArray with the
sparse data.
C
$Revision: 1.4 $
C===============================================================
C
C
loadsparse.f
C
This is the subfunction called by fulltosparse that fills the
C
mxArray with the sparse data. Your version of
C
loadsparse can operate however you would like it to on the
C
data.
3-30
Examples of Fortran MEX-Files
C
C
This is a MEX-file for MATLAB.
C
Copyright (c) 1984-2000 The MathWorks, Inc.
C
C===============================================================
C
Load sparse data subroutine.
function loadsparse(a,b,ir,jc,m,n,nzmax)
integer nzmax, m, n
integer ir(*), jc(*)
real*8 a(*), b(*)
integer i, j, k
C
C
C
200
100
C
300
Copy nonzeros.
k = 1
do 100 j=1,n
NOTE: Sparse indexing is zero based.
jc(j) = k-1
do 200 i=1,m
if (a((j-1)*m+i).ne. 0.0) then
if (k .gt. nzmax) then
jc(n+1) = nzmax
loadsparse = 1
goto 300
endif
b(k) = a((j-1)*m+i)
NOTE: Sparse indexing is zero based.
ir(k) = i-1
k = k+1
endif
continue
continue
NOTE: Sparse indexing is zero based.
jc(n+1) = k-1
loadsparse = 0
return
end
3-31
3
Creating Fortran MEX-Files
At the MATLAB prompt, entering
full = eye(5)
full =
1
0
0
1
0
0
0
0
0
0
0
0
1
0
0
0
0
0
1
0
0
0
0
0
1
creates a full, 5-by-5 identity matrix. Using fulltosparse on the full matrix
produces the corresponding sparse matrix.
spar = fulltosparse(full)
spar =
(1,1)
1
(2,2)
1
(3,3)
1
(4,4)
1
(5,5)
1
Calling Functions from Fortran MEX-Files
It’s possible to call MATLAB functions, operators, M-files, and even other
MEX-files from within your Fortran source code by using the API function
mexCallMATLAB. This example creates an mxArray, passes various pointers to a
subfunction to acquire data, and calls mexCallMATLAB to calculate the sine
function and plot the results.
C
$Revision: 1.8 $
C
================================================================
C
C
sincall.f
C
C
Example for illustrating how to use mexCallMATLAB.
C
C
Creates an mxArray and passes its associated pointers (in
C
this demo, only pointer to its real part, pointer to number
C
of rows, pointer to number of columns) to subfunction fill()
C
to get data filled up, then calls mexCallMATLAB to calculate
C
sin function and plot the result.
3-32
Examples of Fortran MEX-Files
C
C
NOTE: The subfunction fill() is in the file called fill.f.
C
C
This is a MEX-file for MATLAB.
C
Copyright (c) 1984-2000 The MathWorks, Inc.
C
C ==============================================================
C
Gateway subroutine
subroutine mexFunction(nlhs, plhs, nrhs, prhs)
integer nlhs, nrhs
C--------------------------------------------------------------C
(pointer) Replace integer by integer*8 on the DEC Alpha
C
platform.
C
integer plhs(*), prhs(*)
integer rhs(1), lhs(1)
integer mxGetPr, mxCreateFull
C--------------------------------------------------------------C
integer m, n, max
C
initializition
m=1
n=1
max=1000
rhs(1) = mxCreateFull(max, 1, 0)
C
Pass
call
call
call
the pointer and variable and let fill() fill up data.
fill(%val(mxGetPr(rhs(1))), m, n, max)
mxSetM(rhs(1), m)
mxSetN(rhs(1), n)
call mexCallMATLAB(1, lhs, 1, rhs, 'sin')
call mexCallMATLAB(0, NULL, 1, lhs, 'plot')
C
Clean up the unfreed memory after calling mexCallMATLAB.
call mxFreeMatrix(rhs(1))
3-33
3
Creating Fortran MEX-Files
call mxFreeMatrix(lhs(1))
return
end
C
$Revision: 1.3 $
C
================================================================
C
C
fill.f
C
This is the subfunction called by sincall that fills the
C
mxArray with data. Your version of fill can load your data
C
however you would like.
C
C
This is a MEX-file for MATLAB.
C
Copyright (c) 1984-2000 The MathWorks, Inc.
C
C
================================================================
C
Subroutine for filling up data.
subroutine fill(pr, m, n, max)
real*8 pr(*)
integer i, m, n, max
10
m=max/2
n=1
do 10 i=1,m
pr(i)=i*(4*3.1415926/max)
return
end
It is possible to use mexCallMATLAB (or any other API routine) from within your
computational Fortran subroutine. Note that you can only call most MATLAB
functions with double-precision data. M-functions that perform computations,
like eig, will not work correctly with data that is not double precision.
3-34
Examples of Fortran MEX-Files
Running this example
sincall
displays the results
Note It is possible to generate an object of type mxUNKNOWN_CLASS using
mexCallMATLAB. See the example below.
The following example creates an M-file that returns two variables but only
assigns one of them a value.
function [a,b]=foo[c]
a=2*c;
MATLAB displays the following warning message.
Warning: One or more output arguments not assigned during call to
'foo'.
If you then call foo using mexCallMATLAB, the unassigned output variable will
now be of type mxUNKNOWN_CLASS.
3-35
3
Creating Fortran MEX-Files
Advanced Topics
These sections cover advanced features of MEX-files that you can use when
your applications require sophisticated MEX-files.
Help Files
Because the MATLAB interpreter chooses the MEX-file when both an M-file
and a MEX-file with the same name are encountered in the same directory, it
is possible to use M-files for documenting the behavior of your MEX-files. The
MATLAB help command will automatically find and display the appropriate
M-file when help is requested and the interpreter will find and execute the
corresponding MEX-file when the function is actually invoked.
Linking Multiple Files
You can combine several source files when building MEX-files. For example,
mex circle.f square.o rectangle.f shapes.o
is a legal command that operates on the .f and .o files to create a MEX-file
called circle.ext, where ext is the extension corresponding to the MEX-file
type. The name of the resulting MEX-file is taken from the first file in the list.
You may find it useful to use a software development tool like MAKE to manage
MEX-file projects involving multiple source files. Simply create a MAKEFILE
that contains a rule for producing object files from each of your source files and
then invoke mex to combine your object files into a MEX-file. This way you can
ensure that your source files are recompiled only when necessary.
Note On UNIX, you must use the -fortran switch to the mex script if you are
linking Fortran objects.
Workspace for MEX-File Functions
Unlike M-file functions, MEX-file functions do not have their own variable
workspace. mexEvalString evaluates the string in the caller’s workspace. In
addition, you can use the mexGetMatrix and mexPutMatrix routines to get and
put variables into the caller’s workspace.
3-36
Advanced Topics
Memory Management
As of Version 5.2, MATLAB now implicitly destroys (by calling
mxDestroyArray) any arrays created by a MEX-file that are not returned in the
left-hand side list (plhs()). Consequently, any misconstructed arrays left over
at the end of a MEX-file’s execution have the potential to cause memory errors.
In general, we recommend that MEX-files destroy their own temporary arrays
and clean up their own temporary memory. For additional information on
memory management techniques, see the sections “Memory Management” on
page 2-38 and “Memory Management Compatibility Issues” on page 1-37.
3-37
3
Creating Fortran MEX-Files
Debugging Fortran Language MEX-Files
On most platforms, it is now possible to debug MEX-files while they are
running within MATLAB. Complete source code debugging, including setting
breakpoints, examining variables, and stepping through the source code
line-by-line, is now available.
Note The section on “Troubleshooting” on page 1-32 provides additional
information on isolating problems with MEX-files.
To debug a MEX-file from within MATLAB, you must first compile the
MEX-file with the -g option to mex.
mex -g filename.f
Debugging on UNIX
You must start MATLAB from within a debugger. To do this, specify the name
of the debugger you want to use with the -D option when starting MATLAB.
For example, to use dbx, the UNIX debugger, type
matlab -Ddbx
Once the debugger loads MATLAB into memory, you can start it by issuing a
run command. Now, from within MATLAB, enable MEX-file debugging by
typing
dbmex on
at the MATLAB prompt. Then run the MEX-file you want to debug as you
would ordinarily (either directly or by means of some other function or script).
Before executing the MEX-file, you will be returned to the debugger.
You may need to tell the debugger where the MEX-file was loaded or the name
of the MEX-file, in which case MATLAB will display the appropriate command
for you to use. At this point, you are ready to start debugging. You can list the
source code for your MEX-file and set break points in it. It is often convenient
to set one at mexFunction so that you stop at the beginning of the gateway
routine.
3-38
Debugging Fortran Language MEX-Files
Note The name mexFunction may be slightly altered by the compiler (i.e., it
may have an underscore appended). To determine how this symbol appears in
a given MEX-file, use the UNIX command
nm <MEX-file> | grep -i mexfunction
To proceed from the breakpoint, issue a continue command to the debugger.
Once you hit one of your breakpoints, you can make full use of any facilities
your debugger provides to examine variables, display memory, or inspect
registers. Refer to the documentation provided with your debugger for
information on its use.
If you are at the MATLAB prompt and want to return control to the debugger,
you can issue the command
dbmex stop
which allows you to gain access to the debugger so you can set additional
breakpoints or examine source code. To resume execution, issue a continue
command to the debugger.
Debugging on Windows
Compaq Visual Fortran
If you are using the Compaq (or DIGITAL) Visual Fortran compiler, you use
the Microsoft debugging environment to debug your program.
1 Start the Microsoft Visual Studio by typing at the DOS prompt
msdev filename.dll
2 In the Microsoft environment, from the Project menu, select Settings. In
the window that opens, select the Debug tab. This options window contains
edit boxes. In the edit box labeled Executable for debug session, enter the
full path where MATLAB resides. All other edit boxes should be empty.
3 Open the source files and set a break point on the desired line of code by
right-clicking with your mouse on the line of code.
3-39
3
Creating Fortran MEX-Files
4 From the Build menu, select Debug, and click Go.
5 You will now be able to run your MEX-file in MATLAB and use the Microsoft
debugging environment. For more information on how to debug in the
Microsoft environment, see the Microsoft Development Studio
documentation.
3-40
4
Calling MATLAB from C
and Fortran Programs
Using the MATLAB Engine . . . . . . . . . . . . . 4-3
The Engine Library . . . . . . . . . . . . . . . . . . 4-3
GUI-Intensive Applications . . . . . . . . . . . . . . 4-5
Examples of Calling Engine Functions .
Calling MATLAB From a C Application . .
Calling MATLAB From a Fortran Application
Attaching to an Existing MATLAB Session .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4-6
4-6
4-11
4-15
Compiling and Linking Engine Programs .
Masking Floating-Point Exceptions . . . . .
Compiling and Linking on UNIX . . . . . .
Compiling and Linking on Windows . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4-17
4-17
4-18
4-19
4
Calling MATLAB from C and Fortran Programs
The MATLAB engine library is a set of routines that allows you to call
MATLAB from your own programs, thereby employing MATLAB as a
computation engine. MATLAB engine programs are C or Fortran programs
that communicate with a separate MATLAB process via pipes (in UNIX) and
through ActiveX on Windows. There is a library of functions provided with
MATLAB that allows you to start and end the MATLAB process, send data to
and from MATLAB, and send commands to be processed in MATLAB.
The following list summarizes the contents of this chapter:
• “Using the MATLAB Engine”
• “Examples of Calling Engine Functions”
• “Compiling and Linking Engine Programs”
For additional information and support in building your applications, see the
section entitled “Additional Information” on page 1-42.
4-2
Using the MATLAB Engine
Using the MATLAB Engine
Some of the things you can do with the MATLAB engine are:
• Call a math routine, for example, to invert an array or to compute an FFT
from your own program. When employed in this manner, MATLAB is a
powerful and programmable mathematical subroutine library.
• Build an entire system for a specific task, for example, radar signature
analysis or gas chromatography, where the front end (GUI) is programmed
in C and the back end (analysis) is programmed in MATLAB, thereby
shortening development time.
The MATLAB engine operates by running in the background as a separate
process from your own program. This offers several advantages:
• On UNIX, the MATLAB engine can run on your machine, or on any other
UNIX machine on your network, including machines of a different
architecture. Thus you could implement a user interface on your workstation
and perform the computations on a faster machine located elsewhere on your
network. The description of the engOpen function offers further information.
• Instead of requiring that all of MATLAB be linked to your program (a
substantial amount of code), only a small engine communication library is
needed.
The Engine Library
The engine library contains the following routines for controlling the MATLAB
computation engine. Their names all begin with the three-letter prefix eng.
These tables list all the available engine functions and their purposes.
Table 4-1: C Engine Routines
Function
Purpose
engOpen
Start up MATLAB engine
engClose
Shut down MATLAB engine
engGetArray
Get a MATLAB array from the MATLAB engine
engPutArray
Send a MATLAB array to the MATLAB engine
4-3
4
Calling MATLAB from C and Fortran Programs
Table 4-1: C Engine Routines (Continued)
Function
Purpose
engEvalString
Execute a MATLAB command
engOutputBuffer
Create a buffer to store MATLAB text output
engOpenSingleUse
Start a MATLAB engine session for single,
nonshared use
engGetVisible
Determine visibility of MATLAB engine session
engSetVisible
Show or hide MATLAB engine session
Table 4-2: Fortran Engine Routines
Function
Purpose
engOpen
Start up MATLAB engine
engClose
Shut down MATLAB engine
engGetMatrix
Get a MATLAB array from the MATLAB engine
engPutMatrix
Send a MATLAB array to the MATLAB engine
engEvalString
Execute a MATLAB command
engOutputBuffer
Create a buffer to store MATLAB text output
The MATLAB engine also uses the mx prefixed API routines discussed in
Chapter 2, “Creating C Language MEX-Files” and Chapter 3, “Creating
Fortran MEX-Files.”
Communicating with MATLAB
On UNIX, the engine library communicates with the MATLAB engine using
pipes, and, if needed, rsh for remote execution. On Microsoft Windows, the
engine library communicates with MATLAB using ActiveX. Chapter 7,
“ActiveX and DDE Support” contains a detailed description of ActiveX.
4-4
Using the MATLAB Engine
GUI-Intensive Applications
If you have graphical user interface (GUI) intensive applications that execute
a lot of callbacks through the MATLAB engine, you should force these callbacks
to be evaluated in the context of the base workspace. Use evalin to specify that
the base workspace is to be used in evaluating the callback expression, as
follows.
engEvalString(ep, "evalin('base', expression)")
Specifying the base workspace in this manner ensures that MATLAB will
process the callback correctly and return results for that call.
This does not apply to computational applications that do not execute
callbacks.
4-5
4
Calling MATLAB from C and Fortran Programs
Examples of Calling Engine Functions
This section contains examples that illustrate how to call engine functions from
C and Fortran programs. The examples cover the following topics:
• “Calling MATLAB From a C Application”
• “Calling MATLAB From a Fortran Application”
• “Attaching to an Existing MATLAB Session”
It is important to understand the sequence of steps you must follow when using
the engine functions. For example, before using engPutArray, you must create
the matrix, assign a name to it, and populate it.
After reviewing these examples, follow the instructions in “Compiling and
Linking Engine Programs” on page 4-17 to build the application and test it. By
building and running the application, you will ensure that your system is
properly configured for engine applications.
Calling MATLAB From a C Application
This program, engdemo.c, illustrates how to call the engine functions from a
stand-alone C program. For the Windows version of this program, see
engwindemo.c in the <matlab>\extern\examples\eng_mat directory. Engine
examples, like the MAT-file examples, are located in the eng_mat directory.
/* $Revision: 1.6 $ */
/*
* engdemo.c
*
* This is a simple program that illustrates how to call the
* MATLAB engine functions from a C program.
*
*
Copyright (c) 1996-2000 The MathWorks, Inc.
*
*/
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include "engine.h"
#define BUFSIZE 256
4-6
Examples of Calling Engine Functions
int main()
{
Engine *ep;
mxArray *T = NULL, *result = NULL;
char buffer[BUFSIZE];
double time[10] = { 0.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0,
8.0, 9.0 };
/*
* Start the MATLAB engine locally by executing the string
* "matlab".
*
* To start the session on a remote host, use the name of
* the host as the string rather than \0.
*
* For more complicated cases, use any string with whitespace,
* and that string will be executed literally to start MATLAB.
*/
if (!(ep = engOpen("\0"))) {
fprintf(stderr, "\nCan't start MATLAB engine\n");
return EXIT_FAILURE;
}
/*
* PART I
*
* For the first half of this demonstration, we will send data
* to MATLAB, analyze the data, and plot the result.
*/
/*
* Create a variable for our data.
*/
T = mxCreateDoubleMatrix(1, 10, mxREAL);
mxSetName(T, "T");
memcpy((void *)mxGetPr(T), (void *)time, sizeof(time));
/*
* Place the variable T into the MATLAB workspace.
*/
engPutArray(ep, T);
4-7
4
Calling MATLAB from C and Fortran Programs
/*
* Evaluate a function of time, distance = (1/2)g.*t.^2
* (g is the acceleration due to gravity).
*/
engEvalString(ep, "D = .5.*(-9.8).*T.^2;");
/*
* Plot the result.
*/
engEvalString(ep, "plot(T,D);");
engEvalString(ep, "title('Position vs. Time for a falling
object');");
engEvalString(ep, "xlabel('Time (seconds)');");
engEvalString(ep, "ylabel('Position (meters)');");
/*
* Use fgetc() to make sure that we pause long enough to be
* able to see the plot.
*/
printf("Press Return to continue\n\n");
fgetc(stdin);
/*
* We're done for Part I! Free memory, close MATLAB engine.
*/
printf("Done for Part I.\n");
mxDestroyArray(T);
engEvalString(ep, "close;");
/*
* PART II
*
* For the second half of this demonstration, we will request
* a MATLAB string, which should define a variable X. MATLAB
* will evaluate the string and create the variable. We
* will then recover the variable, and determine its type.
*/
/*
* Use engOutputBuffer to capture MATLAB output, so we can
* echo it back.
4-8
Examples of Calling Engine Functions
*/
engOutputBuffer(ep, buffer, BUFSIZE);
while (result == NULL) {
char str[BUFSIZE];
/*
* Get a string input from the user.
*/
printf("Enter a MATLAB command to evaluate. This
command should\n");
printf("create a variable X. This program will then
determine\n");
printf("what kind of variable you created.\n");
printf("For example: X = 1:5\n");
printf(">> ");
fgets(str, BUFSIZE-1, stdin);
/*
* Evaluate input with engEvalString.
*/
engEvalString(ep, str);
/*
* Echo the output from the command. First two characters
* are always the double prompt (>>).
*/
printf("%s", buffer+2);
/*
* Get result of computation.
*/
printf("\nRetrieving X...\n");
if ((result = engGetArray(ep,"X")) == NULL)
printf("Oops! You didn't create a variable X.\n\n");
else {
printf("X is class %s\t\n", mxGetClassName(result));
}
}
4-9
4
Calling MATLAB from C and Fortran Programs
/*
* We're done! Free memory, close MATLAB engine and exit.
*/
printf("Done!\n");
mxDestroyArray(result);
engClose(ep);
return EXIT_SUCCESS;
}
The first part of this program launches MATLAB and sends it data. MATLAB
then analyzes the data and plots the results.
The program then continues with
Press Return to continue
Pressing Return continues the program.
Done for Part I.
Enter a MATLAB command to evaluate. This command should
create a variable X. This program will then determine
what kind of variable you created.
For example: X = 1:5
4-10
Examples of Calling Engine Functions
Entering X = 17.5 continues the program execution.
X = 17.5
X =
17.5000
Retrieving X...
X is class double
Done!
Finally, the program frees memory, closes the MATLAB engine, and exits.
Calling MATLAB From a Fortran Application
This program, fengdemo.f, illustrates how to call the engine functions from a
stand-alone Fortran program.
C
C
C
fengdemo.f
C
C
This program illustrates how to call the MATLAB
C
Engine functions from a Fortran program.
C
C Copyright (c) 1996-2000 by The MathWorks, Inc.
C
C===============================================================
C $Revision: 1.7 $
program main
C--------------------------------------------------------------C
(pointer) Replace integer by integer*8 on the DEC Alpha
C
platform.
C
integer engOpen, engGetMatrix, mxCreateFull, mxGetPr
integer ep, T, D
C--------------------------------------------------------------C
4-11
4
Calling MATLAB from C and Fortran Programs
C
Other variable declarations here
double precision time(10), dist(10)
integer engPutMatrix, engEvalString, engClose
integer temp, status
data time / 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0 /
C
ep = engOpen('matlab ')
C
if (ep .eq. 0) then
write(6,*) 'Can''t start MATLAB engine'
stop
endif
C
T = mxCreateFull(1, 10, 0)
call mxSetName(T, 'T')
call mxCopyReal8ToPtr(time, mxGetPr(T), 10)
C
C
C
C
Place the variable T into the MATLAB workspace.
status = engPutMatrix(ep, T)
C
if (status .ne. 0) then
write(6,*) 'engPutMatrix failed'
stop
endif
C
C
C
C
C
Evaluate a function of time, distance = (1/2)g.*t.^2
(g is the acceleration due to gravity)
if (engEvalString(ep, 'D = .5.*(-9.8).*T.^2;') .ne. 0) then
write(6,*) 'engEvalString failed'
stop
endif
C
4-12
Examples of Calling Engine Functions
C
C
C
Plot the result.
if (engEvalString(ep, 'plot(T,D);') .ne. 0) then
write(6,*) 'engEvalString failed'
stop
endif
if (engEvalString(ep, 'title(''Position vs. Time'')') .ne. 0)
then
write(6,*) 'engEvalString failed'
stop
endif
if (engEvalString(ep, 'xlabel(''Time (seconds)'')') .ne. 0)
then
write(6,*) 'engEvalString failed'
stop
endif
if (engEvalString(ep, 'ylabel(''Position (meters)'')') .ne. 0)
then
write(6,*) 'engEvalString failed'
stop
endif
C
C
C
C
C
Read from console to make sure that we pause long enough to be
able to see the plot.
print *, 'Type 0 <return> to Exit'
print *, 'Type 1 <return> to continue'
read(*,*) temp
C
if (temp.eq.0) then
print *, 'EXIT!'
stop
end if
C
if (engEvalString(ep, 'close;') .ne. 0) then
write(6,*) 'engEvalString failed'
stop
endif
4-13
4
Calling MATLAB from C and Fortran Programs
C
20
10
D = engGetMatrix(ep, 'D')
call mxCopyPtrToReal8(mxGetPr(D), dist, 10)
print *, 'MATLAB computed the following distances:'
print *, ' time(s) distance(m)'
do 10 i=1,10
print 20, time(i), dist(i)
format(' ', G10.3, G10.3)
continue
C
C
call mxFreeMatrix(T)
call mxFreeMatrix(D)
status = engClose(ep)
C
if (status .ne. 0) then
write(6,*) 'engClose failed'
stop
endif
C
stop
end
Executing this program launches MATLAB, sends it data, and plots the
results.
4-14
Examples of Calling Engine Functions
The program continues with
Type 0 <return> to Exit
Type 1 <return> to continue
Entering 1 at the prompt continues the program execution.
1
MATLAB computed the following distances:
time(s) distance(m)
1.00
-4.90
2.00
-19.6
3.00
-44.1
4.00
-78.4
5.00
-123.
6.00
-176.
7.00
-240.
8.00
-314.
9.00
-397.
10.0
-490.
Finally, the program frees memory, closes the MATLAB engine, and exits.
Attaching to an Existing MATLAB Session
You can make a MATLAB engine program attach to a MATLAB session that is
already running by starting the MATLAB session with /Automation in the
command line. When you make a call to engOpen, it will then connect to this
existing session. You should only call engOpen once, as any engOpen calls will
now connect to this one MATLAB session.
The /Automation option also causes the command window to be minimized.
You must open it manually.
Note For more information on the /Automation command line argument and
ActiveX in general, see “Introducing MATLAB ActiveX Integration” on
page 7-3.
4-15
4
Calling MATLAB from C and Fortran Programs
For example:
1 Shut down any MATLAB sessions.
2 From the Start button on the Windows menu bar, click Run.
3 In the Open field, type
d:\matlab\bin\win32\matlab.exe /Automation
where d:\matlab\bin\win32 represents the path to the MATLAB
executable.
4 Click OK. This starts MATLAB.
5 In MATLAB, change directories to $MATLAB/extern/examples/eng_mat,
where $MATLAB is the MATLAB root directory.
6 Compile the engwindemo.c example.
7 Run the engwindemo program by typing at the MATLAB prompt
!engwindemo
This does not start another MATLAB session, but rather uses the MATLAB
session that is already open.
Note On the UNIX platform, you cannot make a MATLAB engine program
use a MATLAB session that is already running.
4-16
Compiling and Linking Engine Programs
Compiling and Linking Engine Programs
To produce an executable version of an engine program, you must compile it
and link it with the appropriate library. This section describes the steps
required to compile and link engine programs on UNIX and Windows systems.
It begins by looking at a special consideration for compilers that do not mask
floating-point exceptions. Topics covered are:
• “Masking Floating-Point Exceptions”
• “Compiling and Linking on UNIX”
• “Compiling and Linking on Windows”
Masking Floating-Point Exceptions
Certain mathematical operations can result in nonfinite values. For example,
division by zero results in the nonfinite IEEE value, inf. A floating-point
exception occurs when such an operation is performed. Because MATLAB uses
an IEEE model that supports nonfinite values such as inf and NaN, MATLAB
disables, or masks, floating-point exceptions.
Some compilers do not mask floating-point exceptions by default. This causes
engine programs built with such compilers to terminate when a floating-point
exception occurs. Consequently, you need to take special precautions when
using these compilers to mask floating-point exceptions so that your engine
application will perform properly.
Note MATLAB based applications should never get floating-point
exceptions. If you do get a floating-point exception, verify that any third party
libraries that you link against do not enable floating-point exception handling.
The only compiler and platform on which you need to mask floating-point
exceptions is the Borland C++ compiler on Windows.
Borland C++ Compiler on Windows
To mask floating-point exceptions when using the Borland C++ compiler on the
Windows platform, you must add some code to your program. Include the
following at the beginning of your main() or WinMain() function, before any
calls to MATLAB API functions.
4-17
4
Calling MATLAB from C and Fortran Programs
#include <float.h>
.
.
.
_control87(MCW_EM,MCW_EM);
.
.
.
Compiling and Linking on UNIX
Under UNIX at runtime, you must tell the system where the API shared
libraries reside. These sections provide the necessary UNIX commands
depending on your shell and system architecture.
Setting Runtime Library Path
In C shell, the command to set the library path is
setenv LD_LIBRARY_PATH <matlab>/extern/lib/$Arch:$LD_LIBRARY_PATH
In Bourne shell, the commands to set the library path are
LD_LIBRARY_PATH=<matlab>/extern/lib/$Arch:$LD_LIBRARY_PATH
export LD_LIBRARY_PATH
where <matlab> is the MATLAB root directory and $Arch is your system
architecture (alpha, glnx86, sgi, sol2, hp700, or ibm_rs).
Note that the environment variable (LD_LIBRARY_PATH in this example) varies
on several platforms. This table lists the different environment variable names
you should use on these systems.
Table 4-3: Environment Variable Names
Architecture
Environment Variable
HP700
SHLIB_PATH
IBM RS/6000
LIBPATH
It is convenient to place these commands in a startup script such as ~/.cshrc
for C shell or ~/.profile for Bourne shell.
4-18
Compiling and Linking Engine Programs
Compiling and Linking
MATLAB provides an options file, engopts.sh, that lets you use the mex script
to easily compile and link engine applications. For example, to compile and link
the engdemo.c example, you can use
mex -f <matlab>/bin/engopts.sh <pathname>/engdemo.c
where <pathname> specifies the complete path to the specified file.
If you need to modify the options file for your particular compiler or platform,
use the -v switch to view the current compiler and linker settings and then
make the appropriate changes in a local copy of the engopts.sh file.
Compiling and Linking on Windows
To compile and link engine programs, use the mex script with an engine options
file. Table 1-2, Options Files, on page 1-17 lists the engine options files
included in this release. All of these options files are located in
<matlab>\bin\win32\mexopts.
As an example, to compile and link the stand-alone engine application
engwindemo.c on Windows using MSVC (Version 5.0), use
mex -f <matlab>\bin\win32\mexopts\msvc50engmatopts.bat ...
<pathname>\engwindemo.c
where <pathname> specifies the complete path to the specified file. If you need
to modify the options file for your particular compiler, use the -v switch to view
the current compiler and linker settings and then make the appropriate
changes in the options file.
4-19
4
Calling MATLAB from C and Fortran Programs
4-20
5
Calling Java from
MATLAB
Using Java from MATLAB: An Overview . . . . . . . 5-3
Bringing Java Classes into MATLAB . . . . . . . . . 5-5
Creating and Using Java Objects
. . . . . . . . . . 5-10
Invoking Methods on Java Objects . . . . . . . . . . 5-19
Working with Java Arrays
. . . . . . . . . . . . . 5-29
Passing Data to a Java Method
. . . . . . . . . . . 5-46
Handling Data Returned from a Java Method . . . . . 5-56
Introduction to Programming Examples . . . . . . . 5-62
Example – Reading a URL . . . . . . . . . . . . . . 5-63
Example – Finding an Internet Protocol Address
. . . 5-66
Example – Communicating Through a Serial Port . . . 5-68
Example – Creating and Using a Phone Book . . . . . 5-73
5
Calling Java from MATLAB
This chapter describes how to use the MATLAB interface to Java classes and
objects. This MATLAB capability enables you to conveniently bring Java
classes into the MATLAB environment, to construct objects from those classes,
to call methods on the Java objects, and to save Java objects for later reloading
— all accomplished with MATLAB functions and commands.
The following list summarizes the contents of this chapter:
• “Using Java from MATLAB: An Overview”
• “Bringing Java Classes into MATLAB”
• “Creating and Using Java Objects”
• “Invoking Methods on Java Objects”
• “Working with Java Arrays”
• “Passing Data to a Java Method”
• “Handling Data Returned from a Java Method”
• “Introduction to Programming Examples”
5-2
Using Java from MATLAB: An Overview
Using Java from MATLAB: An Overview
Java Interface Is Integral to MATLAB
Every installation of MATLAB includes a Java Virtual Machine (JVM), so that
you can use the Java interpreter via MATLAB commands, and you can create
and run programs that create and access Java objects. For information on
MATLAB installation, see the MATLAB installation documentation for your
platform.
Benefits of the MATLAB Java Interface
The MATLAB Java interface enables you to:
• Access Java API (application programming interface) class packages that
support essential activities such as I/O and networking. For example, the URL
class provides convenient access to resources on the internet
• Access third-party Java classes
• Easily construct Java objects in MATLAB
• Call Java object methods, using either Java or MATLAB syntax
• Pass data between MATLAB variables and Java objects
Who Should Use the MATLAB Java Interface
The MATLAB Java interface is intended for all MATLAB users who want to
take advantage of the special capabilities of the Java programming language.
For example:
• You need to access, from MATLAB, the capabilities of available Java classes.
• You are familiar with object-oriented programming in Java or in another
language, such as C++.
• You are familiar with MATLAB object-oriented classes, or with MATLAB
MEX-files.
To Learn More About Java Programming
For a complete description of the Java language and for guidance in
object-oriented software design and programming, you’ll need to consult
outside resources. For example, these recently published books may be helpful:
5-3
5
Calling Java from MATLAB
• Java in a Nutshell (Second Edition), by David Flanagan
• Teach Yourself Java in 21 Days, by Lemay and Perkins
Another place to find information is the JavaSoft Web site.
http://www.javasoft.com
For other suggestions on object-oriented programming resources, see:
• Object Oriented Software Construction, by Bertrand Meyer
• Object Oriented Analysis and Design with Applications, by Grady Booch
Platform Support for the Java Virtual Machine
To find out which version of the Java Virtual Machine (JVM) is being used by
MATLAB on your platform, type the following at the MATLAB prompt.
version -java
5-4
Bringing Java Classes into MATLAB
Bringing Java Classes into MATLAB
You can draw from an extensive collection of existing Java classes or create
your own class definitions to use with MATLAB. This section explains how to
go about finding the class definitions that you need or how to create classes of
your own design. Once you have the classes you need, defined in either
individual .class files, packages, or Java Archive files, you will see how to
make them available within the MATLAB environment.
This section addresses the following topics:
• “Sources of Java Classes”
• “Defining New Java Classes”
• “Making Java Classes Available to MATLAB”
• “Loading Java Class Definitions”
• “Simplifying Java Class Names”
Sources of Java Classes
There are three main sources of Java classes that you can use in MATLAB:
• Java built-in classes
The Java language includes general-purpose class packages, such as
java.awt. See your Java language documentation for descriptions of these
packages.
• Third-party classes
There are a number of packages of special-purpose Java classes that you can
use.
• User-defined classes
You can define new Java classes or subclasses of existing classes. You need
to use a Java development environment to do this, as explained in the
following section.
Defining New Java Classes
To define new Java classes and subclasses of existing classes, you must use a
Java development environment, external to MATLAB, that supports Java
version 1.1. You can download the development kit that supports the version
5-5
5
Calling Java from MATLAB
1.1.8 Java Virtual Machine from the web site at Sun Microsystems, (http://
www.java.sun.com/jdk/). The Sun site also provides documentation for the
Java language and classes that you will need for development.
After you create class definitions in .java files, use your Java compiler to
produce .class files from them. The next step is to make the class definitions
in those .class files available for you to use in MATLAB.
Making Java Classes Available to MATLAB
To make your third-party and user-defined Java classes available in MATLAB,
place them on your MATLAB path by adding their locations to the file
classpath.txt. For more information on classpath.txt, see “Finding and
Editing classpath.txt” on page 5-7.
Making Individual (Unpackaged) Classes Available
To make individual classes (classes not part of a package) available in
MATLAB, add to classpath.txt an entry containing the full path to the
directory you want to use for the .class file(s).
For example, to make available your compiled Java classes in the file
d:\work\javaclasses\test.class, add the following entry to your
classpath.txt file.
d:\work\javaclasses
Making Entire Packages Available
To access one or more classes belonging to a package, you need to make the
entire package available to MATLAB. Do this by adding to classpath.txt the
full path to the parent directory of the highest-level directory of the package
path. This directory is the first component in the package name.
For example, if your Java class package com.mw.tbx.ini has its classes in
directory d:\work\com\mw\tbx\ini, add the following entry to your
classpath.txt.
d:\work
Making Classes in a JAR File Available
You can use the jar (Java Archive) tool to create a JAR file, containing multiple
Java classes and packages in a compressed ZIP format. For information on jar
5-6
Bringing Java Classes into MATLAB
and JAR files, consult your Java development documentation or the JavaSoft
web site. See also “To Learn More About Java Programming” on page 5-3.
To make the contents of a JAR file available for use in MATLAB, add to
classpath.txt the full path, including full filename, for the JAR file.
Note Note that the classpath.txt requirement for JAR files is different
than that for .class files and packages, for which you do not specify any
filename.
For example, to make available the JAR file e:\java\classes\utilpkg.jar,
add the following to your classpath.txt.
e:\java\classes\utilpkg.jar
Finding and Editing classpath.txt
To make your Java classes available, you can edit the default classpath.txt,
which is in the directory toolbox/local. Or, you can copy the default
classpath.txt from this directory to your own startup directory where the
changes you make to it do not affect anyone else.
To find the classpath.txt that is used by your MATLAB environment, use the
MATLAB which function.
which classpath.txt
To edit either the default file or the copy you have made in your own directory,
enter the following command in MATLAB.
edit classpath.txt
MATLAB reads classpath.txt only upon startup. So, if you edit
classpath.txt or change your .class files while MATLAB is running, you
must restart MATLAB to put those changes into effect.
Loading Java Class Definitions
Normally, MATLAB loads a Java class automatically when your code first uses
it, (for example, when you call its constructor). However, there is one exception
that you should be aware of.
5-7
5
Calling Java from MATLAB
When you use the which function on methods defined by Java classes, the
function only acts on the classes currently loaded into the MATLAB working
environment. In contrast, which always operates on MATLAB classes, whether
or not they are loaded.
Determining Which Classes Are Loaded
At any time during a MATLAB session, you can obtain a listing of all the Java
classes that are currently loaded. To do so, you use the inmem function, in the
following form.
[M,X,J] = inmem
This function returns the list of Java classes in output argument J. (It also
returns in M the names of all currently loaded M-files, and in X the names of all
currently loaded MEX-files.)
Here’s a sample of output from the inmem function.
[m,x,j] = inmem
m =
'isequal'
'isunix'
'fullfile'
'filesep'
.
.
.
'matlabrc'
x =
'getprofl'
j =
'java.awt.Frame'
'com.mathworks.ide.desktop.MLDesktop'
Simplifying Java Class Names
Your MATLAB commands can refer to any Java class by its fully qualified
name, which includes its package name. For example, the following are fully
qualified names:
5-8
Bringing Java Classes into MATLAB
• java.lang.String
• java.util.Enumeration
A fully qualified name can be rather long, making commands and functions,
such as constructors, cumbersome to edit and to read. You can refer to classes
by the class name alone (without a package name) if you, first, import the fully
qualified name into MATLAB.
The import command has the following forms.
import pkg_name.*
import pkg_name1.* pkg_name2.*
import class_name
import
L = import
%
%
%
%
%
Import all classes in package
Import multiple packages
Import one class
Display current import list
Return current import list
MATLAB adds all classes that you import to a list called the import list. You
can see what classes are on that list by typing import, without any arguments.
Your code can refer to any class on the list by class name alone.
When called from a function, import adds the specified classes to the import
list in effect for that function. When invoked at the command prompt, import
uses the base import list for your MATLAB environment.
For example, suppose a function contains the following statements.
import java.lang.String
import java.util.* java.awt.*
import java.util.Enumeration
Code that follows the import statements above can now refer to the String,
Frame, and Enumeration classes without using the package names.
str = String('hello');
frm = Frame;
methods Enumeration
% Create java.lang.String object
% Create java.awt.Frame object
% List java.util.Enumeration methods
To clear the list of imported Java classes, invoke the command
clear import
5-9
5
Calling Java from MATLAB
Creating and Using Java Objects
In MATLAB, you create a Java object by calling one of the constructors of that
class. You then use commands and programming statements to perform
operations on these objects. You can also save your Java objects to a MAT-file
and, in subsequent sessions, reload them into MATLAB.
This section addresses the following topics:
• “Constructing Java Objects”
• “Concatenating Java Objects”
• “Saving and Loading Java Objects to MAT-Files”
• “Finding the Public Data Fields of an Object”
• “Accessing Private and Public Data”
• “Determining the Class of an Object”
Constructing Java Objects
You construct Java objects in MATLAB by calling the Java class constructor,
which has the same name as the class. For example, the following constructor
creates a Frame object with the title 'Frame A' and the other properties with
their default values.
frame = java.awt.Frame('Frame A');
Displaying the new object frame shows the following.
frame =
java.awt.Frame[frame0,0,0,0x0,invalid,hidden,layout=
java.awt.BorderLayout,resizable,title=Frame A]
All of the programming examples in this chapter contain Java object
constructors. For example, the sample code for Reading a URL creates a
java.net.URL object with the constructor
url = java.net.URL(...
'http://www.ncsa.uiuc.edu/demoweb/url-primer.html')
Using the javaObject Function
Under certain circumstances, you may need to use the javaObject function to
construct a Java object. The following syntax invokes the Java constructor for
5-10
Creating and Using Java Objects
class, class_name, with the argument list that matches x1,...,xn, and
returns a new object, J.
J = javaObject('class_name',x1,...,xn);
For example, to construct and return a Java object of class java.lang.String,
you use
strObj = javaObject('java.lang.String','hello');
Using the javaObject function enables you to:
• Use classes having names with more than 31 consecutive characters
• Specify the class for an object at run-time, for example, as input from an
application user
The default MATLAB constructor syntax requires that no segment of the input
class name be longer than 31 characters. (A class name segment is any portion
of the class name before, between, or after a dot. For example, there are three
segments in class, java.lang.String.) Any class name segment that exceeds
31 characters is truncated by MATLAB. In the rare case where you need to use
a class name of this length, you must use javaObject to instantiate the class.
The javaObject function also allows you to specify the Java class for the object
being constructed at run-time. In this situation, you call javaObject with a
string variable in place of the class name argument.
class = 'java.lang.String';
text = 'hello';
strObj = javaObject(class, text);
In the usual case, when the class to instantiate is known at development time,
it is more convenient to use the MATLAB constructor syntax. For example, to
create a java.lang.String object, you would use
strObj = java.lang.String('hello');
5-11
5
Calling Java from MATLAB
Note Typically, you will not need to use javaObject. The default MATLAB
syntax for instantiating a Java class is somewhat simpler and is preferable for
most applications. Use javaObject primarily for the two cases described
above.
Java Objects Are References in MATLAB
In MATLAB, Java objects are references and do not adhere to MATLAB
copy-on-assignment and pass-by-value rules. For example,
origFrame = java.awt.Frame;
setSize(origFrame, 800, 400);
newFrameRef = origFrame;
In the second statement above, the variable newFrameRef is a second reference
to origFrame, not a copy of the object. In any code following the example above,
any change to the object at newFrameRef also changes the object at origFrame.
This effect occurs whether the object is changed by MATLAB code, or by Java
code.
The following example shows that origFrame and newFrameRef are both
references to the same entity. When the size of the frame is changed via one
reference (newFrameRef), the change is reflected through the other reference
(origFrame), as well.
setSize(newFrameRef, 1000, 800);
getSize(origFrame)
ans =
java.awt.Dimension[width=1000,height=800]
Concatenating Java Objects
You can concatenate Java objects in the same way that you concatenate native
MATLAB data types. You use either the cat function or the square bracket
operators to tell MATLAB to assemble the enclosed objects into a single object.
5-12
Creating and Using Java Objects
Concatenating Objects of the Same Class
If all of the objects being operated on are of the same Java class, then the
concatenation of those objects produces an array of objects from the same class.
In the following example, the cat function concatenates two objects of the class
java.awt.Point. The class of the result is also java.awt.Point.
point1 = java.awt.Point(24,127);
point2 = java.awt.Point(114,29);
cat(1, point1, point2)
ans =
java.awt.Point[]:
[1x1 java.awt.Point]
[1x1 java.awt.Point]
Concatenating Objects of Unlike Classes
When you concatenate objects of unlike classes, MATLAB finds one class from
which all of the input objects inherit, and makes the output an instance of this
class. MATLAB selects the lowest common parent in the Java class hierarchy
as the output class.
For example, concatenating objects of java.lang.Byte, java.lang.Integer,
and java.lang.Double yields an object of java.lang.Number, since this is the
common parent to the three input classes.
byte = java.lang.Byte(127);
integer = java.lang.Integer(52);
double = java.lang.Double(7.8);
[byte; integer; double]
ans =
java.lang.Number[]:
[
127]
[
52]
[7.8000]
5-13
5
Calling Java from MATLAB
If there is no common, lower level parent, then the resultant class is
java.lang.Object, which is the root of the entire Java class hierarchy.
byte = java.lang.Byte(127);
point = java.awt.Point(24,127);
[byte; point]
ans =
java.lang.Object[]:
[
127]
[1x1 java.awt.Point]
Saving and Loading Java Objects to MAT-Files
Use the MATLAB save function to save a Java object to a MAT-file. Use the
load function to load it back into MATLAB from that MAT-file. To save a Java
object to a MAT-file, and to load the object from the MAT-file, make sure that
the object and its class meet all of the following criteria:
• The class implements the Serializable interface (part of the Java API), either
directly or by inheriting it from a parent class. Any embedded or otherwise
referenced objects must also implement Serializable.
• The definition of the class is not changed between saving and loading the
object. Any change to the data fields or methods of a class prevents the
loading (deserialization) of an object that was constructed with the old class
definition.
• Either the class does not have any transient data fields, or the values in
transient data fields of the object to be saved are not significant. Values in
transient data fields are never saved with the object.
If you define your own Java classes, or subclasses of existing classes, you can
follow the criteria above to enable objects of the class to be saved and loaded in
MATLAB. For details on defining classes to support serialization, consult your
Java development documentation. (See also “To Learn More About Java
Programming” on page 5-3).
5-14
Creating and Using Java Objects
Finding the Public Data Fields of an Object
To list the public fields that belong to a Java object, use the fieldnames
function, which takes either of these forms.
names = fieldnames(obj)
names = fieldnames(obj,'-full')
Calling fieldnames without '-full' returns the names of all the data fields
(including inherited) on the object. With the '-full' qualifier, fieldnames
returns the full description of the data fields defined for the object, including
type, attributes, and inheritance information.
Suppose, for example, that you constructed a Frame object with
frame = java.awt.Frame;
To obtain the full description of the data fields on frame, you could use the
command
fieldnames(frame,'-full')
Sample output from this command follows.
ans =
'static final int WIDTH
% Inherited from java.awt.image.ImageObserver'
'static final int HEIGHT
% Inherited from java.awt.image.ImageObserver'
[1x74 char]
'static final int SOMEBITS
% Inherited from java.awt.image.ImageObserver'
'static final int FRAMEBITS
% Inherited from java.awt.image.ImageObserver'
'static final int ALLBITS
% Inherited from java.awt.image.ImageObserver'
'static final int ERROR
% Inherited from java.awt.image.ImageObserver'
'static final int ABORT
% Inherited from java.awt.image.ImageObserver'
'static final float TOP_ALIGNMENT
% Inherited from java.awt.Component'
'static final float CENTER_ALIGNMENT
% Inherited from java.awt.Component'
5-15
5
Calling Java from MATLAB
'static final float BOTTOM_ALIGNMENT
% Inherited from java.awt.Component'
'static final float LEFT_ALIGNMENT
% Inherited from java.awt.Component'
'static final float RIGHT_ALIGNMENT
% Inherited from java.awt.Component'
.
.
.
Accessing Private and Public Data
Java API classes provide accessor methods you can use to read from and, where
allowed, to modify private data fields. These are sometimes referred to as get
and set methods, respectively.
Some Java classes have public data fields, which your code can read or modify
directly. To access these fields, use the syntax object.field.
Examples
The java.awt.Frame class provides an example of access to both private and
public data fields. This class has the read accessor method getSize, which
returns a java.awt.Dimension object. The Dimension object has data fields
height and width, which are public and therefore directly accessible. The
following example shows MATLAB commands accessing this data.
frame = java.awt.Frame;
frameDim = getSize(frame);
height = frameDim.height;
frameDim.width = 42;
The programming examples in this chapter also contain calls to data field
accessors. For instance, the sample code for “Example – Finding an Internet
Protocol Address” on page 5-66 uses calls to accessors on a
java.net.InetAddress object.
hostname = address.getHostName;
ipaddress = address.getHostAddress;
5-16
Creating and Using Java Objects
Accessing Data from a Static Field
In Java, a static data field is a field that applies to an entire class of objects.
Static fields are most commonly accessed in relation to the class name itself in
Java. For example, the code below accesses the WIDTH field of the Frame class
by referring to it in relation to the package and class names, java.awt.Frame,
rather than an object instance.
width = java.awt.Frame.WIDTH;
In MATLAB, you can use that same syntax. Or you can refer to the WIDTH field
in relation to an instance of the class. The example shown here creates an
instance of java.awt.Frame called frameObj, and then accesses the WIDTH field
using the name frameObj rather than the package and class names.
frame = java.awt.Frame('Frame A');
width = frame.WIDTH
width =
1
Assigning to a Static Field
You can assign values to static Java fields by using a static set method of the
class, or by making the assignment in reference to an instance of the class. For
more information, see the previous section, “Accessing Data from a Static
Field”. You can assign value to the field staticFieldName in the example
below by referring to this field in reference to an instance of the class.
objectName = java.className;
objectName.staticFieldName = value;
Note MATLAB does not allow assignment to static fields using the class
name itself.
Determining the Class of an Object
To find the class of a Java object, use the query form of the MATLAB function,
class. After execution of the following example, frameClass contains the name
of the package and class that Java object frame instantiates.
5-17
5
Calling Java from MATLAB
frameClass = class(frame)
frameClass =
java.awt.Frame
Because this form of class also works on MATLAB objects, it does not, in itself,
tell you whether it is a Java class. To determine the type of class, use the
isjava function, which has the form
x = isjava(obj)
isjava returns 1 if obj is Java, and 0 if it is not.
isjava(frame)
ans =
1
To find out whether or not an object is an instance of a specified class, use the
isa function, which has the form
x = isa(obj, 'class_name')
isa returns 1 if obj is an instance of the class named ’class_name’, and 0 if it
is not. Note that ’class_name’ can be a MATLAB built-in or user-defined class,
as well as a Java class.
isa(frame, 'java.awt.Frame')
ans =
1
5-18
Invoking Methods on Java Objects
Invoking Methods on Java Objects
This section explains how to invoke an object’s methods in MATLAB. It also
covers how to obtain information related to the methods that you’re using and
how MATLAB handles certain types of nonstandard situations.
This section addresses the following topics:
• “Using Java and MATLAB Calling Syntax”
• “Invoking Static Methods on Java Classes”
• “Obtaining Information About Methods”
• “Java Methods That Affect MATLAB Commands”
• “How MATLAB Handles Undefined Methods”
• “How MATLAB Handles Java Exceptions”
Using Java and MATLAB Calling Syntax
To call methods on Java objects, you can use the Java syntax
object.method(arg1,...,argn)
In the following example, frame is the java.awt.Frame object created above,
and getTitle and setTitle are methods of that object.
frame.setTitle('Sample Frame')
title = frame.getTitle
title =
Sample Frame
Alternatively, you can call Java object (nonstatic) methods with the MATLAB
syntax
method(object, arg1,...,argn)
With MATLAB syntax, the java.awt.Frame example above becomes
setTitle(frame, 'Sample Frame')
5-19
5
Calling Java from MATLAB
title = getTitle(frame)
title =
Sample Frame
All of the programming examples in this chapter contain invocations of Java
object methods. For example, the code for Reading a URL contains a call, using
MATLAB syntax, to the openStream method on a java.net.URL object, url.
is = openStream(url)
In another example, the code for “Example – Creating and Using a Phone
Book” on page 5-73 contains a call, using Java syntax, to the load method on a
java.utils.Properties object, pb_htable.
pb_htable.load(FIS);
Using the javaMethod Function on Nonstatic Methods
Under certain circumstances, you may need to use the javaMethod function to
call a Java method. The following syntax invokes the method, method_name, on
Java object J with the argument list that matches x1,...,xn. This returns the
value X.
X = javaMethod('method_name',J,x1,...,xn);
For example, to call the startsWith method on a java.lang.String object
passing one argument, use
gAddress = java.lang.String('Four score and seven years ago');
str = java.lang.String('Four score');
javaMethod('startsWith', gAddress, str)
ans =
1
Using the javaMethod function enables you to:
• Use methods having names longer than 31 characters
• Specify the method you want to invoke at run-time, for example, as input
from an application user
The only way to invoke a method whose name is longer than 31 characters is
to use javaMethod. The Java and MATLAB calling syntax does not accept
method names of this length.
5-20
Invoking Methods on Java Objects
With javaMethod, you can also specify the method to be invoked at run-time.
In this situation, your code calls javaMethod with a string variable in place of
the method name argument. When you use javaMethod to invoke a static
method, you can also use a string variable in place of the class name argument.
Note Typically, you will not need to use javaMethod. The default MATLAB
syntax for invoking a Java method is somewhat simpler and is preferable for
most applications. Use javaMethod primarily for the two cases described
above.
Invoking Static Methods on Java Classes
To invoke a static method on a Java class, use the Java invocation syntax
class.method(arg1,...,argn)
For example, call the isNaN static method on the java.lang.Double class.
java.lang.Double.isNaN(2.2)
Alternatively, you can apply static method names to instances of a class. In this
example, the isNaN static method is referenced in relation to the dblObject
instance of the java.lang.Double class.
dblObject = java.lang.Double(2.2);
dblObject.isNaN
ans =
0
Several of the programming examples in this chapter contain examples of
static method invocation. For example, the code for Communicating Through a
Serial Port contains a call to static method getPortIdentifier on Java class
javax.comm.CommPortIdentifier.
commPort =
javax.comm.CommPortIdentifier.getPortIdentifier('COM1');
5-21
5
Calling Java from MATLAB
Using the javaMethod Function on Static Methods
The javaMethod function was introduced in section “Using the javaMethod
Function on Nonstatic Methods” on page 5-20. You can also use this function to
call static methods.
The following syntax invokes the static method, method_name, in class,
class_name, with the argument list that matches x1,...,xn. This returns the
value X.
X = javaMethod('method_name','class_name',x1,...,xn);
For example, to call the static isNaN method of the java.lang.Double class on
a double value of 2.2, you use
javaMethod('isNaN','java.lang.Double',2.2);
Using the javaMethod function to call static methods enables you to:
• Use methods having names longer than 31 characters
• Specify method and class names at run-time, for example, as input from an
application user
Obtaining Information About Methods
MATLAB offers several functions to help obtain information related to the
Java methods you are working with. You can request a list of all of the methods
that are implemented by any class. The list may be accompanied by other
method information such as argument types and exceptions. You can also
request a listing of every Java class that you loaded into MATLAB that
implements a specified method.
Methodsview: Displaying a Listing of Java Methods
If you want to know what methods are implemented by a particular Java (or
MATLAB) class, use the methodsview function in MATLAB. Specify the class
name (along with its package name, for Java classes) in the command line. If
you have imported the package that defines this class, then the class name
alone will suffice.
5-22
Invoking Methods on Java Objects
The following command lists information on all methods in the
java.awt.MenuItem class.
methodsview java.awt.MenuItem
A new window appears, listing one row of information for each method in the
class. This is what the methodsview display looks like. The fieldnames shown
at the top of the window are described following the figure.
5-23
5
Calling Java from MATLAB
Each row in the window displays up to six fields of information describing the
method. The table below lists the fields displayed in the methodsview window
along with a description and examples of each field type.
Table 5-1: Fields Displayed in the Methodsview Window
Field Name
Description
Examples
Qualifiers
Method type qualifiers
abstract, synchronized
Return Type
Data type returned by the method
void, java.lang.String
Name
Method name
addActionListener, dispatchEvent
Arguments
Arguments passed to method
boolean, java.lang.Object
Other
Other relevant information
throws java.io.IOException
Parent
Parent of the specified class
java.awt.MenuComponent
Using the Methods Function on Java Classes
In addition to methodsview, the MATLAB methods function, that returns
information on methods of MATLAB classes, will also work on Java classes.
You can use any of the following forms of this command.
methods class_name
methods class_name -full
n = methods('class_name')
n = methods('class_name','-full')
Use methods without the '-full' qualifier to return the names of all the
methods (including inherited methods) of the class. Names of overloaded
methods are listed only once.
With the '-full' qualifier, methods returns a listing of the method names
(including inherited methods) along with attributes, argument lists, and
inheritance information on each. Each overloaded method is listed separately.
5-24
Invoking Methods on Java Objects
For example, display a full description of all methods of the
java.awt.Dimension object.
methods java.awt.Dimension -full
Methods for class java.awt.Dimension:
Dimension()
Dimension(java.awt.Dimension)
Dimension(int,int)
java.lang.Class getClass() % Inherited from java.lang.Object
int hashCode() % Inherited from java.lang.Object
boolean equals(java.lang.Object)
java.lang.String toString()
void notify() % Inherited from java.lang.Object
void notifyAll() % Inherited from java.lang.Object
void wait(long) throws java.lang.InterruptedException
% Inherited from java.lang.Object
void wait(long,int) throws java.lang.InterruptedException
% Inherited from java.lang.Object
void wait() throws java.lang.InterruptedException
% Inherited from java.lang.Object
java.awt.Dimension getSize()
void setSize(java.awt.Dimension)
void setSize(int,int)
Determining What Classes Define a Method
You can use the which function to display the fully qualified name (package
and class name) of a method implemented by a loaded Java class. With the
-all qualifier, the which function finds all classes with a method of the name
specified.
Suppose, for example, that you want to find the package and class name for the
concat method, with the String class currently loaded. Use the command
which concat
java.lang.String.concat
% String method
If the java.lang.String class has not been loaded, the same which command
would give the output
which concat
concat not found.
5-25
5
Calling Java from MATLAB
If you use which -all for the method equals, with the String and
java.awt.Frame classes loaded, you see the following display.
which -all equals
java.lang.String.equals
java.awt.Frame.equals
com.mathworks.ide.desktop.MLDesktop.equals
% String method
% Frame method
% MLDesktop method
The which function operates differently on Java classes than it does on
MATLAB classes. MATLAB classes are always displayed by which, whether or
not they are loaded. This is not true for Java classes. You can find out which
Java classes are currently loaded by using the command [m,x,j]=inmem,
described in “Determining Which Classes Are Loaded” on page 5-8.
For a description of how Java classes are loaded, see “Making Java Classes
Available to MATLAB” on page 5-6.
Java Methods That Affect MATLAB Commands
MATLAB commands that operate on Java objects and arrays make use of the
methods that are implemented within, or inherited by, these objects’ classes.
There are some MATLAB commands that you can alter somewhat in behavior
by changing the Java methods that they rely on.
Changing the Effect of disp and display
You can use the disp function to display the value of a variable or an
expression in MATLAB. Terminating a command line without a semicolon also
calls the disp function. You can also use disp to display a Java object in
MATLAB.
When disp operates on a Java object, MATLAB formats the output using the
toString method of the class to which the object belongs. If the class does not
implement this method, then an inherited toString method is used. If no
intermediate ancestor classes define this method, it uses the toString method
defined by the java.lang.Object class. You can override inherited toString
methods in classes that you create by implementing such a method within your
class definition. In this way, you can change the way MATLAB displays
information regarding the objects of the class.
5-26
Invoking Methods on Java Objects
Changing the Effect of isequal
The MATLAB isequal function compares two or more arrays for equality in
type, size, and contents. This function can also be used to test Java objects for
equality.
When you compare two Java objects using isequal, MATLAB performs the
comparison using the Java method, equals. MATLAB first determines the
class of the objects specified in the command, and then uses the equals method
implemented by that class. If it is not implemented in this class, then an
inherited equals method is used. This will be the equals method defined by the
java.lang.Object class if no intermediate ancestor classes define this method.
You can override inherited equals methods in classes that you create by
implementing such a method within your class definition. In this way, you can
change the way MATLAB performs comparison of the members of this class.
Changing the Effect of double and char
You can also define your own Java methods toDouble and toChar to change the
output of the MATLAB double and char functions. For more information, see
the sections entitled “Converting to the MATLAB double Data Type” and
“Converting to the MATLAB char Data Type” on page 5-58.
How MATLAB Handles Undefined Methods
If your MATLAB command invokes a nonexistent method on a Java object,
MATLAB looks for a built-in function with the same name. If MATLAB finds a
built-in function of that name, it attempts to invoke it. If MATLAB does not
find a function with that name, it displays a message stating that it cannot find
a method by that name for the class.
For example, MATLAB has a built-in method named size, and the Java API
java.awt.Frame class also has a size method. If you call size on a Frame
object, the size method defined by java.awt.Frame is executed. However, if
you call size on an object of java.lang.String, MATLAB does not find a size
method for this class. It executes the MATLAB size built-in instead.
string = java.lang.String('hello');
size(string)
ans =
1
1
5-27
5
Calling Java from MATLAB
Note When you define a Java class for use in MATLAB, avoid giving any of
its methods the same name as a MATLAB built-in function.
How MATLAB Handles Java Exceptions
If invoking a Java method or constructor throws an exception, MATLAB
catches the exception and transforms it into a MATLAB error. MATLAB puts
the text of the Java error message into its own error message. Receiving an
error from a Java method or constructor has the same appearance as receiving
an error from an M-file.
5-28
Working with Java Arrays
Working with Java Arrays
You can pass singular Java objects to and from methods or you may pass them
in an array, providing the method expects them in that form. This array must
either be a Java array (returned from another method call or created within
MATLAB) or, under certain circumstances, a MATLAB cell array. This section
describes how to create and manipulate Java arrays in MATLAB. Later
sections will describe how to use MATLAB cell arrays in calls to Java methods.
Note The term dimension here refers more to the number of subscripts
required to address the elements of an array than to its length, width, and
height characteristics. For example, a 5-by-1 array is referred to as being
one-dimensional, as its individual elements can be indexed into using only one
array subscript.
This section addresses the following topics:
• “How MATLAB Represents the Java Array”
• “Creating an Array of Objects Within MATLAB”
• “Accessing Elements of a Java Array”
• “Assigning to a Java Array”
• “Concatenating Java Arrays”
• “Creating a New Array Reference”
• “Creating a Copy of a Java Array”
5-29
5
Calling Java from MATLAB
How MATLAB Represents the Java Array
The term java array refers to any array of Java objects returned from a call to
a Java class constructor or method. You may also construct a Java array within
MATLAB using the javaArray function. The structure of a Java array is
significantly different from that of a MATLAB matrix or array. MATLAB hides
these differences whenever possible, allowing you to operate on the arrays
using the usual MATLAB command syntax. Just the same, it may be helpful to
keep the following differences in mind as you work with Java arrays.
Representing More Than One Dimension
An array in the Java language is strictly a one-dimensional structure because
it is measured only in length. If you want to work with a two-dimensional
array, you can create an equivalent structure using an array of arrays. To add
further dimensions, you add more levels to the array, making it an array of
arrays of arrays, and so on. You may want to use such multilevel arrays when
working in MATLAB as it is a matrix and array-based programming language.
MATLAB makes it easy for you to work with multilevel Java arrays by treating
them like the matrices and multidimensional arrays that are a part of the
language itself. You access elements of an array of arrays using the same
MATLAB syntax that you would use if you were handling a matrix. If you were
to add more levels to the array, MATLAB would be able to access and operate
on the structure as if it were a multidimensional MATLAB array.
The left side of the following figure shows Java arrays of one, two, and three
dimensions. To the right of each is the way the same array is represented to
you in MATLAB. Note that single-dimension arrays are represented as a
column vector.
5-30
Working with Java Arrays
Array Access from Java
Array Access from MATLAB
jArray[0]
jArray(1)
jArray[1]
jArray(2)
jArray[2]
jArray(3)
Simple Array
jArray[0][3]
Array of Arrays
jArray[0][4][2]
Array of Arrays of Arrays
One-dimensional Array
jArray(1,4)
Two-Dimensional Array
jArray(1,5,3)
Three-Dimensional Array
5-31
5
Calling Java from MATLAB
Array Indexing
Java array indexing is different than MATLAB array indexing. Java array
indices are zero-based, MATLAB array indices are one-based. In Java
programming, you access the elements of array y of length N using y[0]
through y[N-1]. When working with this array in MATLAB, you access these
same elements using the MATLAB indexing style of y(1) through y(N). Thus,
if you have a Java array of 10 elements, the seventh element is obtained using
y(7), and not y[6] as you would have used when writing a program in Java.
The Shape of the Java Array
A Java array can be different from a MATLAB array in its overall shape. A
two-dimensional MATLAB array maintains a rectangular shape, as each row
is of equal length and each column of equal height. The Java counterpart of
this, an array of arrays, does not necessarily hold to this rectangular form.
Each individual lower level array may have a different length.
Such an array structure is pictured below. This is an array of three underlying
arrays of different lengths. The term ragged is commonly used to describe this
arrangement of array elements as the array ends do not match up evenly.
When a Java method returns an array with this type of structure, it is stored
in a cell array by MATLAB.
jArray[0]
length = 5
jArray[1]
length = 2
jArray[2]
length = 3
Interpreting the Size of a Java Array
When the MATLAB size function is applied to a simple Java array, the
number of rows returned is the length of the Java array and the number of
columns is always 1.
Determining the size of a Java array of arrays is not so simple. The potentially
ragged shape of an array returned from Java makes it impossible to size the
array in the same way as for a rectangular matrix. In a ragged Java array,
there is no one value that represents the size of the lower level arrays.
5-32
Working with Java Arrays
When the size function is applied to a Java array of arrays, the resulting value
describes the top level of the specified array. For the Java array shown here
size(A) = 3x1
size(A(3)) = 5x1
size(A) returns the dimensions of the highest array level of A. The highest
level of the array has a size of 3-by-1.
size(A)
ans =
3
1
To find the size of a lower level array, say the five-element array in row 3, refer
to the row explicitly.
size(A(3))
ans =
5
1
You can specify a dimension in the size command using the following syntax.
However, you will probably find this useful only for sizing the first dimension,
dim=1, as this will be the only non-unary dimension.
m = size(X,dim)
size(A, 1)
ans =
3
Interpreting the Number of Dimensions of a Java Arrays
For Java arrays, whether they are simple one-level arrays or multilevel, the
MATLAB ndims function always returns a value of 2 to indicate the number of
dimensions in the array. This is a measure of the number of dimensions in the
top-level array which will always be equal to 2.
5-33
5
Calling Java from MATLAB
Creating an Array of Objects Within MATLAB
To call a Java method that has one or more arguments defined as an array of
Java objects, you must, under most circumstances, pass your objects in a Java
array. You can construct an array of objects in a call to a Java method or
constructor. Or you can create the array within MATLAB.
The MATLAB javaArray function lets you create a Java array structure that
can be handled in MATLAB as a single multidimensional array. You specify
the number and size of the array dimensions along with the class of objects you
intend to store in it. Using the one-dimensional Java array as its primary
building block, MATLAB then builds an array structure that satisfies the
dimensions requested in the javaArray command.
Using the javaArray Function
To create a Java object array, use the MATLAB javaArray function, which has
the following syntax.
A = javaArray('element_class', m, n, p, ...)
The first argument is the ’element_class’ string, which names the class of the
elements in the array. You must specify the fully qualified name (package and
class name). The remaining arguments (m, n, p, ...) are the number of
elements in each dimension of the array.
An array that you create with javaArray is equivalent to the array that you
would create with the Java code.
A = new element_class[m][n][p]...;
The following command builds a Java array of four lower level arrays, each
capable of holding five objects of the java.lang.Double class. (You will
probably be more likely to use primitive types of double than instances of the
java.lang.Double class, but in this context, it affords us a simple example.)
dblArray = javaArray('java.lang.Double', 4, 5);
The javaArray function does not deposit any values into the array elements
that it creates. You must do this separately. The following MATLAB code
stores objects of the java.lang.Double type in the Java array dblArray that
was just created.
5-34
Working with Java Arrays
for m = 1:4
for n = 1:5
dblArray(m,n) = java.lang.Double((m*10) + n);
end
end
dblArray
dblArray =
java.lang.Double[][]:
[11]
[12]
[13]
[21]
[22]
[23]
[31]
[32]
[33]
[41]
[42]
[43]
[14]
[24]
[34]
[44]
[15]
[25]
[35]
[45]
Another Way to Create a Java Array
You can also create an array of Java objects using syntax that is more typical
to MATLAB. For example, the following syntax creates a 4-by-5 MATLAB
array of type double and assigns zero to each element of the array.
matlabArray(4,5) = 0;
You use similar syntax to create a Java array in MATLAB, except that you
must specify the Java class name. The value being assigned, 0 in this example,
is stored in the final element of the array, javaArray(4,5). All other elements
of the array receive the empty matrix.
javaArray(4,5) = java.lang.Double(0)
javaArray =
java.lang.Double[][]:
[]
[]
[]
[]
[]
[]
[]
[]
[]
[]
[]
[]
[]
[]
[]
[]
[]
[]
[]
[0]
Note You cannot change the dimensions of an existing Java array as you can
with a MATLAB array. The same restriction exists when working with Java
arrays in the Java language. See the example below.
5-35
5
Calling Java from MATLAB
This example first creates a scalar MATLAB array, and then successfully
modifies it to be two-dimensional.
matlabArray = 0;
matlabArray(4,5) = 0
matlabArray =
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
When you try this with a Java array, you get an error. Similarly, you cannot
create an array of Java arrays from a Java array, and so forth.
javaArray = java.lang.Double(0);
javaArray(4,5) = java.lang.Double(0);
??? Index exceeds Java array dimensions.
Accessing Elements of a Java Array
You can access elements of a Java object array by using the MATLAB array
indexing syntax, A(row,col). For example, to access the element of array
dblArray located at row 3, column 4, use
row3_col4 = dblArray(3,4)
row3_col4 =
34.0
In Java, this would be dblArray[2][3].
You can also use MATLAB array indexing syntax to access an element in an
object’s data field. Suppose that myMenuObj is an instance of a window menu
class. This user-supplied class has a data field, menuItemArray, which is a Java
array of java.awt.menuItem. To get element 3 of this array, use the following
command.
currentItem = myMenuObj.menuItemArray(3)
Using Single Subscript Indexing to Access Arrays
Elements of a MATLAB matrix are most commonly referenced using both row
and column subscripts. For example, you use x(3,4) to reference the array
5-36
Working with Java Arrays
element at the intersection of row 3 and column 4. Sometimes it is more
advantageous to use just a single subscript. MATLAB provides this capability
(see the section on “Advanced Indexing” in the Using MATLAB manual).
Indexing into a MATLAB matrix using a single subscript references one
element of the matrix. Using the MATLAB matrix shown here, matlabArray(3)
returns a single element of the matrix.
matlabArray = [11 12 13 14 15; 21 22 23 24 25; ...
31 32 33 34 35; 41 42 43 44 45]
matlabArray =
11
12
13
14
15
21
22
23
24
25
31
32
33
34
35
41
42
43
44
45
matlabArray(3)
ans =
31
Indexing this way into a Java array of arrays references an entire subarray of
the overall structure. Using the dblArray Java array, that looks the same as
matlabArray shown above, dblArray(3) returns the 5-by-1 array that makes
up the entire third row.
row3 = dblArray(3)
row3 =
java.lang.Double[]:
[31]
[32]
[33]
[34]
[35]
This is a useful feature of MATLAB as it allows you to specify an entire array
from a larger array structure, and then manipulate it as an object.
Using the Colon Operator
Use of the MATLAB colon operator (:) is supported in subscripting Java array
references. This operator works just the same as when referencing the contents
of a MATLAB array. Using the Java array of java.lang.Double objects shown
5-37
5
Calling Java from MATLAB
here, the statement dblArray(2,2:4) refers to a portion of the lower level
array, dblArray(2). A new array, row2Array, is created from the elements in
columns 2 through 4.
dblArray
dblArray =
java.lang.Double[][]:
[11]
[12]
[13]
[21]
[22]
[23]
[31]
[32]
[33]
[41]
[42]
[43]
[14]
[24]
[34]
[44]
[15]
[25]
[35]
[45]
row2Array = dblArray(2,2:4)
row2Array =
java.lang.Double[]:
[22]
[23]
[24]
You also can use the colon operator in single-subscript indexing, as covered in
“Using Single Subscript Indexing to Access Arrays” on page 5-36. By making
your subscript a colon rather than a number, you can convert an array of arrays
into one linear array. The following example converts the 4-by-5 array
dblArray into a 20-by-1 linear array.
linearArray = dblArray(:)
linearArray =
java.lang.Double[]:
[11]
[12]
[13]
[14]
[15]
[21]
[22]
.
.
.
5-38
Working with Java Arrays
This works the same way on an N-dimensional Java array structure. Using the
colon operator as a single subscripted index into the array produces a linear
array composed of all of the elements of the original array.
Note Java and MATLAB arrays are stored differently in memory. This is
reflected in the order they are given in a linear array. Java array elements are
stored in an order that matches the rows of the matrix, (11, 12, 13, ... in the
array shown above). MATLAB array elements are stored in an order that
matches the columns, (11, 21, 31, ...).
Using END in a Subscript
You can use the end keyword in the first subscript of an access statement. The
first subscript references the top-level array in a multilevel Java array
structure.
Note Using end on lower level arrays is not valid due to the potentially
ragged nature of these arrays (see “The Shape of the Java Array” on
page 5-32). In this case, there is no consistent end value to be derived.
The following example displays data from the third to the last row of Java
array dblArray.
last2rows = dblArray(3:end, :)
last2rows =
java.lang.Double[][]:
[31]
[32]
[33]
[34]
[41]
[42]
[43]
[44]
[35]
[45]
Assigning to a Java Array
You assign values to objects in a Java array in essentially the same way as you
do in a MATLAB array. Although Java and MATLAB arrays are structured
quite differently, you use the same command syntax to specify which elements
you want to assign to. See “How MATLAB Represents the Java Array” on
page 5-30 for more information on Java and MATLAB array differences.
5-39
5
Calling Java from MATLAB
The following example deposits the value 300 in the dblArray element at row
3, column 2. In Java, this would be dblArray[2][1].
dblArray(3,2) = java.lang.Double(300)
dblArray =
java.lang.Double[][]:
[11]
[ 12]
[13]
[14]
[15]
[21]
[ 22]
[23]
[24]
[25]
[31]
[300]
[33]
[34]
[35]
[41]
[ 42]
[43]
[44]
[45]
You use the same syntax to assign to an element in an object’s data field.
Continuing with the myMenuObj example shown in “Accessing Elements of a
Java Array” on page 5-36, you assign to the third menu item in menuItemArray
as follows.
myMenuObj.menuItemArray(3) = java.lang.String('Save As...');
Using Single Subscript Indexing for Array Assignment
You can use a single-array subscript to index into a Java array structure that
has more than one dimension. Refer to “Using Single Subscript Indexing to
Access Arrays” on page 5-36 for a description of this feature as used with Java
arrays.
You can use single-subscript indexing to assign values to an array as well. The
example below assigns a one-dimensional Java array, onedimArray, to a row of
a two-dimensional Java array, dblArray. Start out by creating the
one-dimensional array.
onedimArray = javaArray('java.lang.Double', 5);
for k = 1:5
onedimArray(k) = java.lang.Double(100 * k);
end
Since dblArray(3) refers to the 5-by-1 array displayed in the third row of
dblArray, you can assign the entire, similarly dimensioned, 5-by-1
onedimArray to it.
dblArray(3) = onedimArray
dblArray =
java.lang.Double[][]:
[ 11]
[ 12]
[ 13]
5-40
[ 14]
[ 15]
Working with Java Arrays
[ 21]
[100]
[ 41]
[ 22]
[200]
[ 42]
[ 23]
[300]
[ 43]
[ 24]
[400]
[ 44]
[ 25]
[500]
[ 45]
Assigning to a Linear Array
You can assign a value to every element of a multidimensional Java array by
treating the array structure as if it were a single linear array. This entails
replacing the single, numerical subscript with the MATLAB colon operator. If
you start with the dblArray array, you can initialize the contents of every
object in the two-dimensional array with the following statement.
dblArray(:) = java.lang.Double(0)
dblArray =
java.lang.Double[][]:
[0]
[0]
[0]
[0]
[0]
[0]
[0]
[0]
[0]
[0]
[0]
[0]
[0]
[0]
[0]
[0]
[0]
[0]
[0]
[0]
Using the Colon Operator
You can use the MATLAB colon operator as you would when working with
MATLAB arrays. The statements below assign given values to each of the four
rows in the Java array, dblArray. Remember that each row actually represents
a separate Java array in itself.
dblArray(1,:) = java.lang.Double(125);
dblArray(2,:) = java.lang.Double(250);
dblArray(3,:) = java.lang.Double(375);
dblArray(4,:) = java.lang.Double(500)
dblArray =
java.lang.Double[][]:
[125]
[125]
[125]
[125]
[250]
[250]
[250]
[250]
[375]
[375]
[375]
[375]
[500]
[500]
[500]
[500]
[125]
[250]
[375]
[500]
Assigning the Empty Matrix
When working with MATLAB arrays, you can assign the empty matrix, (i.e.,
the 0-by-0 array denoted by []) to an element of the array. For Java arrays, you
5-41
5
Calling Java from MATLAB
can also assign [] to array elements. This stores the NULL value, rather than a
0-by-0 array, in the Java array element.
Subscripted Deletion
When you assign the empty matrix value to an entire row or column of a
MATLAB array, you find that MATLAB actually removes the affected row or
column from the array. In the example below, the empty matrix is assigned to
all elements of the fourth column in the MATLAB matrix, matlabArray. Thus,
the fourth column is completely eliminated from the matrix. This changes its
dimensions from 4-by-5 to 4-by-4.
matlabArray = [11 12 13 14 15; 21 22 23 24 25; ...
31 32 33 34 35; 41 42 43 44 45]
matlabArray =
11
12
13
14
15
21
22
23
24
25
31
32
33
34
35
41
42
43
44
45
matlabArray(:,4) = []
matlabArray =
11
12
13
15
21
22
23
25
31
32
33
35
41
42
43
45
You can assign the empty matrix to a Java array, but the effect will be
different. The next example shows that, when the same operation is performed
on a Java array, the structure is not collapsed; it maintains its 4-by-5
dimensions.
dblArray(:,4) = []
dblArray =
java.lang.Double[][]:
[125]
[125]
[250]
[250]
[375]
[375]
[500]
[500]
5-42
[125]
[250]
[375]
[500]
[]
[]
[]
[]
[125]
[250]
[375]
[500]
Working with Java Arrays
The dblArray data structure s actually an array of five-element arrays of
java.lang.Double objects. The empty array assignment placed the NULL value
in the fourth element of each of the lower level arrays.
Concatenating Java Arrays
You can concatenate arrays of Java objects in the same way as arrays of other
data types. (To understand how scalar Java objects are concatenated by
MATLAB see “Concatenating Java Objects” on page 5-12.)
Use either the cat function or the square bracket ([]) operators. This example
horizontally concatenates two Java arrays: d1, and d2.
% Construct a 2-by-3 array of java.lang.Double.
d1 = javaArray('java.lang.Double',2,3);
for m = 1:3
for n = 1:3
d1(m,n) = java.lang.Double(n*2 + m-1);
end;
end;
d1
d1 =
java.lang.Double[][]:
[2]
[4]
[6]
[3]
[5]
[7]
[4]
[6]
[8]
% Construct a 2-by-2 array of java.lang.Double.
d2 = javaArray('java.lang.Double',2,2);
for m = 1:3
for n = 1:2
d2(m,n) = java.lang.Double((n+3)*2 + m-1);
end;
end;
d2
d2 =
java.lang.Double[][]:
[ 8]
[10]
[ 9]
[11]
[10]
[12]
5-43
5
Calling Java from MATLAB
% Concatenate the two along the second dimension.
d3 = cat(2,d1,d2)
d3 =
java.lang.Double[][]:
[2]
[4]
[6]
[3]
[5]
[7]
[4]
[6]
[8]
[ 8]
[ 9]
[10]
[10]
[11]
[12]
Creating a New Array Reference
Because Java arrays in MATLAB are references, assigning an array variable to
another variable results in a second reference to the array.
Consider the following example where two separate array variables reference
a common array. The original array, origArray, is created and initialized. The
statement newArrayRef = origArray creates a copy of this array variable.
Changes made to the array referred to by newArrayRef also show up in the
original array.
origArray = javaArray('java.lang.Double', 3, 4);
for m = 1:3
for n = 1:4
origArray(m,n) = java.lang.Double((m * 10) + n);
end
end
origArray
origArray =
java.lang.Double[][]:
[11]
[12]
[13]
[21]
[22]
[23]
[31]
[32]
[33]
[14]
[24]
[34]
% ----- Make a copy of the array reference ----newArrayRef = origArray;
newArrayRef(3,:) = java.lang.Double(0);
origArray
origArray =
5-44
Working with Java Arrays
java.lang.Double[][]:
[11]
[12]
[13]
[21]
[22]
[23]
[ 0]
[ 0]
[ 0]
[14]
[24]
[ 0]
Creating a Copy of a Java Array
You can create an entirely new array from an existing Java array by indexing
into the array to describe a block of elements, (or subarray), and assigning this
subarray to a variable. The assignment copies the values in the original array
to the corresponding cells of the new array.
As with the example in section “Creating a New Array Reference” on page 5-44,
an original array is created and initialized. But, this time, a copy is made of the
array contents rather than copying the array reference. Changes made using
the reference to the new array do not affect the original.
origArray = javaArray('java.lang.Double', 3, 4);
for m = 1:3
for n = 1:4
origArray(m,n) = java.lang.Double((m * 10) + n);
end
end
origArray
origArray =
java.lang.Double[][]:
[11]
[12]
[13]
[21]
[22]
[23]
[31]
[32]
[33]
[14]
[24]
[34]
% ----- Make a copy of the array contents ----newArray = origArray(:,:);
newArray(3,:) = java.lang.Double(0);
origArray
origArray =
java.lang.Double[][]:
[11]
[12]
[13]
[21]
[22]
[23]
[31]
[32]
[33]
[14]
[24]
[34]
5-45
5
Calling Java from MATLAB
Passing Data to a Java Method
When you make a call from MATLAB to Java code, any MATLAB data types
you pass in the call are converted to data types native to the Java language.
MATLAB performs this conversion on each argument that is passed, except for
those arguments that are already Java objects. This section describes the
conversion that is performed on specific MATLAB data types and, at the end,
also takes a look at how argument types affect calls made to overloaded
methods.
If data is to be returned by the method being called, MATLAB receives this
data and converts it to the appropriate MATLAB format wherever necessary.
This process is covered in the next section entitled “Handling Data Returned
from a Java Method” on page 5-56.
This section addresses the following topics:
• “Conversion of MATLAB Argument Data”
• “Passing Built-In Data Types”
• “Passing String Arguments”
• “Passing Java Objects”
• “Other Data Conversion Topics”
• “Passing Data to Overloaded Methods”
Conversion of MATLAB Argument Data
MATLAB data, passed as arguments to Java methods, are converted by
MATLAB into data types that best represent the data to the Java language.
The table below shows all of the MATLAB base types for passed arguments and
the Java base types defined for input arguments. Each row shows a MATLAB
type followed by the possible Java argument matches, from left to right in order
of closeness of the match. The MATLAB types (except cell arrays) can all be
5-46
Passing Data to a Java Method
scalar (1-by-1) arrays or matrices. All of the Java types can be scalar values or
arrays.
Table 5-2: Conversion of MATLAB Types to Java Types
MATLAB
Argument
Closest
Type (7)
Java Input Argument (Scalar or Array)
Least Close
Type (1)
double
(logical)
boolean
byte
short
int
long
float
double
double
double
float
long
int
short
byte
boolean
single
float
double
char
String
char
uint8
byte
short
int
long
float
double
uint16
short
int
long
float
double
uint32
int
long
float
double
int8
byte
short
int
long
float
int16
short
int
long
float
double
int32
int
long
float
double
cell array
of strings
array of
String
Java object
Object
cell array
of object
array of
Object
double
Data type conversion of arguments passed to Java code will be discussed in the
following three categories. MATLAB handles each category differently.
• “Passing Built-In Data Types”
• “Passing String Arguments”
• “Passing Java Objects”
5-47
5
Calling Java from MATLAB
Passing Built-In Data Types
Java has eight data types that are intrinsic to the language and are not
represented as Java objects. These are often referred to as built-in, or
elemental, data types and they include boolean, byte, short, long, int,
double, float, and char. MATLAB converts its own data types to these Java
built-in types according to the table, “Conversion of MATLAB Types to Java
Types” on page 5-47. Built-in types are in the first 10 rows of the table.
When a Java method you are calling expects one of these data types, you can
pass it the type of MATLAB argument shown in the left-most column of the
table. If the method takes an array of one of these types, you can pass a
MATLAB array of the data type. MATLAB converts the data type of the
argument to the type assigned in the method declaration.
The MATLAB code shown below creates a top-level window frame and sets its
dimensions. The call to setBounds passes four MATLAB scalars of the double
type to the inherited Java Frame method, setBounds, that takes four
arguments of the int type. MATLAB converts each 64-bit double data type to
a 32-bit integer prior to making the call. Shown here is the setBounds method
declaration followed by the MATLAB code that calls the method.
public void setBounds(int x, int y, int width, int height)
frame=java.awt.Frame;
frame.setBounds(200,200,800,400);
frame.setVisible(1);
Passing Built-In Types in an Array
To call a Java method with an argument defined as an array of a built-in type,
you can create and pass a MATLAB matrix with a compatible base type. The
following code defines a polygon by sending four x and y coordinates to the
Polygon constructor. Two 1-by-4 MATLAB arrays of double are passed to
java.awt.Polygon, which expects integer arrays in the first two arguments.
Shown here is the Java method declaration followed by MATLAB code that
calls the method, and then verifies the set coordinates.
public Polygon(int xpoints[], int ypoints[], int npoints)
poly = java.awt.Polygon([14 42 98 124], [55 12 -2 62], 4);
5-48
Passing Data to a Java Method
[poly.xpoints poly.ypoints]
ans =
14
55
42
12
98
-2
124
62
% Verify the coordinates
MATLAB Arrays Are Passed by Value
Since MATLAB arrays are passed by value, any changes that a Java method
makes to them will not be visible to your MATLAB code. If you need to access
changes that a Java method makes to an array, then, rather than passing a
MATLAB array, you should create and pass a Java array, which is a reference.
For a description of using Java arrays in MATLAB, see “Working with Java
Arrays” on page 5-29.
Note Generally, it is preferable to have methods return data that has been
modified using the return argument mechanism as opposed to passing a
reference to that data in an argument list.
Passing String Arguments
To call a Java method that has an argument defined as an object of class
java.lang.String, you can pass either a String object that was returned from
an earlier Java call or a MATLAB 1-by-n character array. If you pass the
character array, MATLAB converts the array to a Java object of
java.lang.String for you.
For a programming example, see “Example – Reading a URL” on page 5-63.
This shows a MATLAB character array that holds a URL being passed to the
Java URL class constructor. The constructor, shown below, expects a Java
String argument.
public URL(String spec) throws MalformedURLException
5-49
5
Calling Java from MATLAB
In the MATLAB call to this constructor, a character array specifying the URL
is passed. MATLAB converts this array to a Java String object prior to calling
the constructor.
url = java.net.URL(...
'http://www.ncsa.uiuc.edu/demoweb/url-primer.html')
Passing Strings in an Array
When the method you are calling expects an argument of an array of type
String, you can create such an array by packaging the strings together in a
MATLAB cell array. The strings can be of varying lengths since you are storing
them in different cells of the array. As part of the method call, MATLAB
converts the cell array to a Java array of String objects.
In the following example, the echoPrompts method of a user-written class
accepts a string array argument that MATLAB converted from its original
format as a cell array of strings. The parameter list in the Java method appears
as follows.
public String[] echoPrompts(String s[])
You create the input argument by storing both strings in a MATLAB cell array.
MATLAB converts this structure to a Java array of String.
myaccount.echoPrompts({'Username: ','Password: '})
ans =
'Username: '
'Password: '
Passing Java Objects
When calling a method that has an argument belonging to a particular Java
class, you must pass an object that is an instance of that class. In the example
below, the add method belonging to the java.awt.Menu class requires, as an
argument, an object of the java.awt.MenuItem class. The method declaration
for this is
public MenuItem add(MenuItem mi)
The example operates on the frame created in the previous example in “Passing
Built-In Data Types” on page 5-48. The second, third, and fourth lines of code
shown here add items to a menu to be attached to the existing window frame.
5-50
Passing Data to a Java Method
In each of these calls to menu1.add, an object that is an instance of the
java.awt.MenuItem Java class is passed.
menu1 = java.awt.Menu('File Options');
menu1.add(java.awt.MenuItem('New'));
menu1.add(java.awt.MenuItem('Open'));
menu1.add(java.awt.MenuItem('Save'));
menuBar=java.awt.MenuBar;
menuBar.add(menu1);
frame.setMenuBar(menuBar);
Handling Objects of Class java.lang.Object
A special case exists when the method being called takes an argument of the
java.lang.Object class. Since this class is the root of the Java class hierarchy,
you can pass objects of any class in the argument. The hash table example
shown that follows, passes objects belonging to different classes to a common
method, put, which expects an argument of java.lang.Object. The method
declaration for put is
public synchronized Object put(Object key, Object value)
The following MATLAB code passes objects of different types (boolean, float,
and string) to the put method.
hTable = java.util.Hashtable;
hTable.put(0, java.lang.Boolean('TRUE'));
hTable.put(1, java.lang.Float(41.287));
hTable.put(2, java.lang.String('test string'));
hTable
% Verify hash table contents
hTable =
{1.0=41.287, 2.0=test string, 0.0=true}
When passing arguments to a method that takes java.lang.Object, it is not
necessary to specify the class name for objects of a built-in data type. Line
three, in the example above, specifies that 41.287 is an instance of class
java.lang.Float. You can omit this and simply say, 41.287, as shown in the
following example. Thus, MATLAB will create each object for you, choosing the
closest matching Java object representation for each argument.
5-51
5
Calling Java from MATLAB
The three calls to put from the preceding example can be rewritten as
hTable.put(0, 1);
hTable.put(1, 41.287);
hTable.put(2, 'test string');
Passing Objects in an Array
The only types of Java object arrays that you can pass to Java methods are
Java arrays and MATLAB cell arrays.
If the objects have already been placed into an array, either an array returned
from a Java constructor or constructed in MATLAB by the javaArray function,
then you simply pass it as is in the argument to the method being called. No
conversion is done by MATLAB, as this is already a Java array.
If you have objects that are not already in a Java array, then MATLAB allows
you to simply pass them in a MATLAB cell array. In this case, MATLAB
converts the cell array to a Java array prior to passing the argument.
The following example shows the mapPoints method of a user-written class
accepting an array of class java.awt.Point. The method declaration for this is
public Object mapPoints(java.awt.Point p[])
The MATLAB code shown below creates a 2-by-2 cell array containing four
Java Point objects. When the cell array is passed to the mapPoints method,
MATLAB converts it to a Java array of type java.awt.Point.
pointObj1
pointObj2
pointObj3
pointObj4
=
=
=
=
java.awt.Point(25,143);
java.awt.Point(31,147);
java.awt.Point(49,151);
java.awt.Point(52,176);
cellArray={pointObj1, pointObj2; pointObj3, pointObj4}
cellArray =
[1x1 java.awt.Point]
[1x1 java.awt.Point]
[1x1 java.awt.Point]
[1x1 java.awt.Point]
testData.mapPoints(cellArray);
5-52
Passing Data to a Java Method
Handling a Cell Array of Java Objects
You create a cell array of Java objects by using the MATLAB syntax
{a1,a2,...}. You index into a cell array of Java objects in the usual way, with
the syntax a{m,n,...}.
The following example creates a cell array of two Frame objects, frame1 and
frame2, and assigns it to variable frames.
frame1 = java.awt.Frame('Frame A');
frame2 = java.awt.Frame('Frame B');
frameArray = {frame1, frame2}
frameArray =
[1x1 java.awt.Frame]
[1x1 java.awt.Frame]
The next statement assigns element {1,2} of the cell array frameArray to
variable f.
f = frameArray {1,2}
f =
java.awt.Frame[frame2,0,0,0x0,invalid,hidden,layout =
java.awt.BorderLayout,resizable,title=Frame B]
Other Data Conversion Topics
There are several remaining items of interest regarding the way MATLAB
converts its data to a compatible Java type. This includes how MATLAB
matches array dimensions, and how it handles empty matrices and empty
strings.
How Array Dimensions Affect Conversion
The term dimension, as used in this section, refers more to the number of
subscripts required to address the elements of an array than to its length,
width, and height characteristics. For example, a 5-by-1 array is referred to as
having one dimension, as its individual elements can be indexed into using only
one array subscript.
In converting MATLAB to Java arrays, MATLAB handles dimension in a
special manner. For a MATLAB array, dimension can be considered as the
number of nonsingleton dimensions in the array. For example, a 10-by-1 array
has dimension 1, and a 1-by-1 array has dimension 0. In Java, dimension is
5-53
5
Calling Java from MATLAB
determined solely by the number of nested arrays. For example, double[][]
has dimension 2, and double has dimension 0.
If the Java array’s number of dimensions exactly matches the MATLAB array’s
number of dimensions n, then the conversion results in a Java array with n
dimensions. If the Java array has fewer than n dimensions, the conversion
drops singleton dimensions, starting with the first one, until the number of
remaining dimensions matches the number of dimensions in the Java array.
Empty Matrices and Nulls
The empty matrix is compatible with any method argument for which NULL is
a legal value in Java. The empty string ('') in MATLAB translates into an
empty (not NULL) String object in Java.
Passing Data to Overloaded Methods
When you invoke an overloaded method on a Java object, MATLAB determines
which method to invoke by comparing the arguments your call passes to the
arguments defined for the methods. Note that in this discussion, the term
method includes constructors. When it determines the method to call,
MATLAB converts the calling arguments to Java method types according to
Java conversion rules, except for conversions involving objects or cell arrays.
See “Passing Objects in an Array” on page 5-52.
How MATLAB Determines the Method to Call
When your MATLAB function calls a Java method, MATLAB:
1 Checks to make sure that the object (or class, for a static method) has a
method by that name
2 Determines whether the invocation passes the same number of arguments
of at least one method with that name
3 Makes sure that each passed argument can be converted to the Java type
defined for the method
If all of the preceding conditions are satisfied, MATLAB calls the method.
In a call to an overloaded method, if there is more than one candidate,
MATLAB selects the one with arguments that best fit the calling arguments.
5-54
Passing Data to a Java Method
First, MATLAB rejects all methods that have any argument types that are
incompatible with the passed arguments (for example, if the method has a
double argument and the passed argument is a char).
Among the remaining methods, MATLAB selects the one with the highest
fitness value, which is the sum of the fitness values of all its arguments. The
fitness value for each argument is the fitness of the base type minus the
difference between the MATLAB array dimension and the Java array
dimension. (Array dimensionality is explained in “How Array Dimensions
Affect Conversion” on page 5-53.) If two methods have the same fitness, the
first one defined in the Java class is chosen.
Example - Calling an Overloaded Method
Suppose a function constructs a java.io.OutputStreamWriter object, osw, and
then invokes a method on the object.
osw.write('Test data', 0, 9);
MATLAB finds that the class java.io.OutputStreamWriter defines three
write methods.
public void write(int c);
public void write(char[] cbuf, int off, int len);
public void write(String str, int off, int len);
MATLAB rejects the first write method, because it takes only one argument.
Then, MATLAB assesses the fitness of the remaining two write methods.
These differ only in their first argument, as explained below.
In the first of these two write methods, the first argument is defined with base
type, char. The table, Conversion of MATLAB Types to Java Types, shows that
for the type of the calling argument (MATLAB char), Java type, char, has a
value of 6. There is no difference between the dimension of the calling
argument and the Java argument. So the fitness value for the first argument
is 6.
In the other write method, the first argument has Java type String, which has
a fitness value of 7. The dimension of the Java argument is 0, so the difference
between it and the calling argument dimension is 1. Therefore, the fitness
value for the first argument is 6.
Because the fitness value of those two write methods is equal, MATLAB calls
the one listed first in the class definition, with char[] first argument.
5-55
5
Calling Java from MATLAB
Handling Data Returned from a Java Method
In many cases, data returned from Java is incompatible with the data types
operated on within MATLAB. When this is the case, MATLAB converts the
returned value to a data type native to the MATLAB language. This section
describes the conversion performed on the various data types that can be
returned from a call to a Java method.
This section addresses the following topics:
• “Conversion of Java Return Data”
• “Built-In Data Types”
• “Java Objects”
• “Converting Objects to MATLAB Data Types”
Conversion of Java Return Data
The following table lists Java return types and the resulting MATLAB types.
For some Java base return types, MATLAB treats scalar and array returns
differently, as described following the table.
Table 5-3: Conversion of Java Types to MATLAB Types
5-56
Java Return Type
If Scalar Return, Resulting
MATLAB Type
If Array Return, Resulting
MATLAB Type
boolean
double (logical)
double (logical)
byte
double
int8
short
double
int16
int
double
int32
long
double
double
float
double
single
double
double
double
char
char
char
Handling Data Returned from a Java Method
Built-In Data Types
Java built-in data types are described in “Passing Built-In Data Types” on
page 5-48. Basically, this data type includes boolean, byte, short, long, int,
double, float, and char. When the value returned from a method call is one of
these Java built-in types, MATLAB converts it according to the table,
“Conversion of Java Types to MATLAB Types” on page 5-56.
A single numeric or boolean value converts to a 1-by-1 matrix of double, which
is convenient for use in MATLAB. An array of a numeric or boolean return
values converts to an array of the closest base type, to minimize the required
storage space. Array conversions are listed in the right-hand column of the
table.
A return value of Java type char converts to a 1-by-1 matrix of char. And an
array of Java char converts to a MATLAB array of that type.
Java Objects
When a method call returns Java objects, MATLAB leaves them in their
original form. They remain as Java objects so you can continue to use them to
interact with other Java methods.
The only exception to this is when the method returns data of type,
java.lang.Object. This class is the root of the Java class hierarchy and is
frequently used as a catchall for objects and arrays of various types. When the
method being called returns a value of the Object class, MATLAB converts its
value according to the table, “Conversion of Java Types to MATLAB Types” on
page 5-56. That is, numeric and boolean objects such as java.lang.Integer or
java.lang.Boolean convert to a 1-by-1 MATLAB matrix of double. Object
arrays of these types convert to the MATLAB data types listed in the
right-hand column of the table. Other object types are not converted.
Converting Objects to MATLAB Data Types
With the exception of objects of class String and class Object, MATLAB does
not convert Java objects returned from method calls to a native MATLAB data
type. If you want to convert Java object data to a form more readily usable in
MATLAB, there are a few MATLAB functions that enable you to do this. These
are described in the following sections.
5-57
5
Calling Java from MATLAB
Converting to the MATLAB double Data Type
Using the double function in MATLAB, you can convert any Java object or
array of objects to the MATLAB double data type. The action taken by the
double function depends on the class of the object you specify:
• If the object is an instance of a numeric class (java.lang.Number or one of
the classes that inherit from that class), then MATLAB uses a preset
conversion algorithm to convert the object to a MATLAB double.
• If the object is not an instance of a numeric class, MATLAB checks the class
definition to see if it implements a method called toDouble. Note that
MATLAB uses toDouble to perform its conversion of Java objects to the
MATLAB double data type. If such a method is implemented for this class,
MATLAB executes it to perform the conversion.
• If you are using a class of your own design, you can write your own toDouble
method to perform conversions on objects of that class to a MATLAB double.
This enables you to specify your own means of data type conversion for
objects belonging to your own classes.
Note If the class of the specified object is not java.lang.Number, does not
inherit from that java.lang.Number, and it does not implement a toDouble
method, then an attempt to convert the object using the double function
results in an error in MATLAB.
The syntax for the double command is as follows, where object is a Java object
or Java array of objects.
double(object);
Converting to the MATLAB char Data Type
With the MATLAB char function, you can convert java.lang.String objects
and arrays to MATLAB data types. A single java.lang.String object converts
to a MATLAB character array. An array of java.lang.String objects converts
to a MATLAB cell array, with each cell holding a character array.
If the object specified in the char command is not an instance of the
java.lang.String class, then MATLAB checks its class to see if it implements
a method named toChar. If this is the case, then MATLAB executes the toChar
5-58
Handling Data Returned from a Java Method
method of the class to perform the conversion. If you write your own class
definitions, then you can make use of this feature by writing a toChar method
that performs the conversion according to your own needs.
Note If the class of the specified object is not java.lang.String and it does
not implement a toChar method, then an attempt to convert the object using
the char function results in an error in MATLAB.
The syntax for the char command is as follows, where object is a Java object
or Java array of objects.
char(object);
Converting to a MATLAB Structure
Java objects are similar to the MATLAB structure in that many of an object’s
characteristics are accessible via field names defined within the object. You
may want to convert a Java object into a MATLAB structure to facilitate the
handling of its data in MATLAB. Use the MATLAB struct function on the
object to do this.
The syntax for the struct command is as follows, where object is a Java object
or Java array of objects.
struct(object);
The following example converts a java.awt.Polygon object into a MATLAB
structure. You can access the fields of the object directly using MATLAB
structure operations. The last line indexes into the array, pstruct.xpoints,
to deposit a new value into the third array element.
polygon = java.awt.Polygon([14 42 98 124], [55 12 -2 62], 4);
pstruct = struct(polygon)
pstruct =
npoints: 4
xpoints: [4x1 int32]
ypoints: [4x1 int32]
5-59
5
Calling Java from MATLAB
pstruct.xpoints
ans =
14
42
98
124
pstruct.xpoints(3) = 101;
Converting to a MATLAB Cell Array
Use the cell function to convert a Java array or Java object into a MATLAB
cell array. Elements of the resulting cell array will be of the MATLAB type (if
any) closest to the Java array elements or Java object.
The syntax for the cell command is as follows, where object is a Java object
or Java array of objects.
cell(object);
In the following example, a MATLAB cell array is created in which each cell
holds an array of a different data type. The cell command used in the first line
converts each type of object array into a cell array.
import java.lang.* java.awt.*;
% Create a Java array of double
dblArray = javaArray('java.lang.Double', 1, 10);
for m = 1:10
dblArray(1, m) = Double(m * 7);
end
% Create a Java array of points
ptArray = javaArray('java.awt.Point', 3);
ptArray(1) = Point(7.1, 22);
ptArray(2) = Point(5.2, 35);
ptArray(3) = Point(3.1, 49);
% Create a Java array of strings
strArray = javaArray('java.lang.String', 2, 2);
strArray(1,1) = String('one');
strArray(1,2) = String('two');
strArray(2,1) = String('three'); strArray(2,2) = String('four');
5-60
Handling Data Returned from a Java Method
% Convert
cellArray
cellArray
{1x10
each to cell arrays
= {cell(dblArray), cell(ptArray), cell(strArray)}
=
cell}
{3x1 cell}
{2x2 cell}
cellArray{1,1}
ans =
[7]
[14]
% Array of type double
[21]
cellArray{1,2}
[28]
[35]
[42]
[49]
[56]
[63]
[70]
% Array of type Java.awt.Point
ans =
[1x1 java.awt.Point]
[1x1 java.awt.Point]
[1x1 java.awt.Point]
cellArray{1,3}
% Array of type char array
ans =
'one'
'three'
'two'
'four'
5-61
5
Calling Java from MATLAB
Introduction to Programming Examples
The following programming examples demonstrate the MATLAB interface to
Java classes and objects:
• “Example – Reading a URL”
• “Example – Finding an Internet Protocol Address”
• “Example – Communicating Through a Serial Port”
• “Example – Creating and Using a Phone Book”
Each example contains the following sections:
• Overview - Describes what the example does and how it uses the Java
interface to accomplish it. Highlighted are the most important Java objects
that are constructed and used in the example code.
• Description - provides a detailed description of all code in the example. For
longer functions, the description is divided into functional sections that focus
on a few statements.
• Running the Example - Shows a sample of the output from execution of the
example code.
The example descriptions concentrate on the Java-related functions. For
information on other MATLAB programming constructs, operators, and
functions used in the examples, see the applicable sections in the MATLAB
documentation.
5-62
Example – Reading a URL
Example – Reading a URL
This program, URLdemo, opens a connection to a web site specified by a URL
(Uniform Resource Locator), for the purpose of reading text from a file at that
site. It constructs an object of the Java API class, java.net.URL, which enables
convenient handling of URLs. Then, it calls a method on the URL object, to
open a connection.
To read and display the lines of text at the site, URLdemo uses classes from the
Java I/O package java.io. It creates an InputStreamReader object, and then
uses that object to construct a BufferedReader object. Finally, it calls a method
on the BufferedReader object to read the specified number of lines from the
site.
Description of URLdemo
The major tasks performed by URLdemo are:
1. Construct a URL Object
The example first calls a constructor on java.net.URL and assigns the
resulting object to variable url. The URL constructor takes a single argument,
the name of the URL to be accessed, as a string. The constructor checks
whether the input URL has a valid form.
url = java.net.URL(...
'http://www.ncsa.uiuc.edu/demoweb/url-primer.html')
2. Open a Connection to the URL
The second statement of the example calls the method, openStream, on the URL
object url, to establish a connection with the web site named by the object. The
method returns an InputStream object to variable, is, for reading bytes from
the site.
is = openStream(url)
3. Set Up a Buffered Stream Reader
The next two lines create a buffered stream reader for characters. The
java.io.InputStreamReader constructor is called with the input stream is, to
return to variable isr an object that can read characters. Then, the
java.io.BufferedReader constructor is called with isr, to return a
5-63
5
Calling Java from MATLAB
BufferedReader object to variable br. A buffered reader provides for efficient
reading of characters, arrays, and lines.
isr = java.io.InputStreamReader(is)
br = java.io.BufferedReader(isr)
4. Read and Display Lines of Text
The final statement reads and displays the first 10 lines of text from the site.
Within a MATLAB for statement that iterates 10 times, the BufferedReader
method readLine reads a line of text (terminated by a return and/or line feed
character) from the site. To display the output in MATLAB, assign it to
variable s, without terminating the statement with a semicolon.
for k = 1:10
s = readLine(br)
end
Running the Example
When you run this example, you see output similar to the following.
url =
http://www.ncsa.uiuc.edu/demoweb/url-primer.html
is =
java.io.BufferedInputStream@62d
isr =
java.io.InputStreamReader@67d
br =
java.io.BufferedReader@644
s =
<TITLE>A Beginner's Guide to URLs</TITLE>
s =
<H1>A Beginner's Guide to URLs</H1>
s =
''
s =
What is a URL? A URL is a <b>Uniform Resource Locator</b>. Think
s =
of it as a networked extension of the standard <i>filename</i>
5-64
Example – Reading a URL
s =
concept: not only can you point to a file in a directory, but that
s =
file and that directory can exist on any machine on the network,
s =
can be served via any of several different methods, and might not
s =
even be something as simple as a file: URLs can also point to
s =
queries, documents stored deep within databases, the results of a
s =
<i>finger</i> or <i>archie</i>
5-65
5
Calling Java from MATLAB
Example – Finding an Internet Protocol Address
The resolveip function returns either the name or address of an IP (internet
protocol) host. If you pass resolveip a hostname, it returns the IP address. If
you pass resolveip an IP address, it returns the hostname. The function uses
the Java API class java.net.InetAddress, which enables you to find an IP
address for a hostname, or the hostname for a given IP address, without
making DNS calls.
resolveip calls a static method on the InetAddress class to obtain an
InetAddress object. Then, it calls accessor methods on the InetAddress object
to get the hostname and IP address for the input argument. It displays either
the hostname or the IP address, depending on the program input argument.
Description of resolveip
The major tasks performed by resolveip are:
1. Create an InetAddress Object
Instead of constructors, the java.net.InetAddress class has static methods
that return an instance of the class. The try statement calls one of those
methods, getByName, passing the input argument that the user has passed to
resolveip. The input argument can be either a hostname or an IP address. If
getByName fails, the catch statement displays an error message.
function resolveip(input)
try
address = java.net.InetAddress.getByName(input);
catch
error(sprintf('Unknown host %s.', input));
end
2. Retrieve the Hostname and IP Address
The example uses calls to the getHostName and getHostAddress accessor
functions on the java.net.InetAddress object, to obtain the hostname and IP
address, respectively. These two functions return objects of class
java.lang.String, so we use the char function to convert them to character
arrays.
hostname = char(address.getHostName);
ipaddress = char(address.getHostAddress);
5-66
Example – Finding an Internet Protocol Address
3. Display the Hostname or IP Address
The example uses the MATLAB strcmp function to compare the input
argument to the resolved IP address. If it matches, MATLAB displays the
hostname for the internet address. If the input does not match, MATLAB
displays the IP address.
if strcmp(input,ipaddress)
disp(sprintf('Host name of %s is %s', input, hostname));
else
disp(sprintf('IP address of %s is %s', input, ipaddress));
end;
Running the Example
Here is an example of calling the resolveip function with a hostname.
resolveip ('www.mathworks.com')
IP address of www.mathworks.com is 144.212.100.10
Here is a call to the function with an IP address.
resolveip ('144.212.100.10')
Host name of 144.212.100.10 is www.mathworks.com
5-67
5
Calling Java from MATLAB
Example – Communicating Through a Serial Port
The serialexample program uses classes of the Java API javax.comm package,
which support access to communications ports. After defining port
configuration variables, serialexample constructs a
javax.comm.CommPortIdentifier object, to manage the serial
communications port. The program calls the open method on that object to
return an object of the javax.comm.SerialPort class, which describes the
low-level interface to the COM1 serial port, assumed to be connected to a
Tektronix oscilloscope. (The example can be run without an oscilloscope.) The
serialexample program then calls several methods on the SerialPort object
to configure the serial port.
The serialexample program uses the I/O package java.io, to write to and
read from the serial port. It calls a static method to return an OutputStream
object for the serial port. It then passes that object to the constructor for
java.io.OutputStreamWriter. It calls the write method on the
OutputStreamWriter object to write a command to the serial port, which sets
the contrast on the oscilloscope. It calls write again to write a command that
checks the contrast. It then constructs an object of the
java.io.InputStreamWriter class to read from the serial port.
It calls another static method on the SerialPort object to return an
OutputStream object for the serial port. It calls a method on that object to get
the number of bytes to read from the port. It passes the InputStream object to
the constructor for java.io.OutputStreamWriter. Then, if there is data to
read, it calls the read method on the OutputStreamWriter object to read the
contrast data returned by the oscilloscope.
Note MATLAB also provides built-in serial port support, described in
Chapter 8, “Serial Port I/O”.
5-68
Example – Communicating Through a Serial Port
Description of Serial Example
The major tasks performed by serialexample are:
1. Define Variables for Serial Port Configuration and Output
The first five statements define variables for configuring the serial port. The
first statement defines the baud rate to be 9600, the second defines number of
data bits to be 8, and the third defines the number of stop bits to be 1. The
fourth statement defines parity to be off, and the fifth statement defines flow
control (handshaking) to be off.
SerialPort_BAUD_9600 = 9600;
SerialPort_DATABITS_8 = 8;
SerialPort_STOPBITS_1 = 1;
SerialPort_PARITY_NONE = 0;
SerialPort_FLOWCTRL_NONE = 0;
The last variable definition sets the terminator character for writing to the
serial port, to a carriage return.
terminator = char(13);
2. Create a CommPortIdentifier Object
Instead of constructors, the javax.comm.CommPortIdentifier class has
static methods that return an instance of the class. The example calls one of
these, getPortIdentifier, to return a CommPortIdentifier object for port
COM1.
commPort = ...
javax.comm.CommPortIdentifier.getPortIdentifier('COM1');
3. Open the Serial Port
The example opens the serial port, by calling open on the CommPortIdentifier
object commPort. The open call returns a SerialPort object, assigning it to
serialPort. The first argument to open is the name (owner) for the port, the
second argument is the name for the port, and the third argument is the
number of milliseconds to wait for the open.
serialPort = open(commPort, 'serial', 1000);
5-69
5
Calling Java from MATLAB
4. Configure the Serial Port
The next three statements call configuration methods on the SerialPort object
serialPort. The first statement calls setSerialPortParams to set the baud
rate, data bits, stop bits, and parity. The next two statements call
setFlowControlMode to set the flow control, and then enableReceiveTimeout
to set the timeout for receiving data.
setSerialPortParams(serialPort, SerialPort_BAUD_9600,...
SerialPort_DATABITS_8, SerialPort_STOPBITS_1,...
SerialPort_PARITY_NONE);
setFlowControlMode(serialPort, SerialPort_FLOWCTRL_NONE);
enableReceiveTimeout(serialPort, 1000);
5. Set Up an Output Stream Writer
The example then calls a constructor to create and open a
java.io.OutputStreamWriter object. The constructor call passes the
java.io.OutputStream object, returned by a call to the getOutputStream
method serialPort, and assigns the OutputStreamWriter object to out.
out = java.io.OutputStreamWriter(getOutputStream(serialPort));
6. Write Data to Serial Port and Close Output Stream
The example writes a string to the serial port, by calling write on the object
out. The string is formed by concatenating (with MATLAB [ ] syntax) a
command to set the oscilloscope’s contrast to 45, with the command terminator
that is required by the instrument. The next statement calls flush on out to
flush the output stream.
write(out, ['Display:Contrast 45' terminator]);
flush(out);
Then, the example again calls write on out to send another string to the serial
port. This string is a query command, to determine the oscilloscope’s contrast
setting, concatenated with the command terminator. The example then calls
close on the output stream.
write(out, ['Display:Contrast?' terminator]);
close(out);
5-70
Example – Communicating Through a Serial Port
7. Open an Input Stream and Determine Number of Bytes to Read
To read the data expected from the oscilloscope in response to the contrast
query, the example opens an input stream by calling the static method,
InputStream.getInputStream, to obtain an InputStream object for the serial
port. Then, the example calls the method available on the InputStream object,
in, and assigns the returned number of bytes to numAvail.
in = getInputStream(serialPort);
numAvail = available(in);
8. Create an Input Stream Reader for the Serial Port
The example then calls a java.io.InputStreamReader constructor, with the
InputStream object, in, and assigns the new object to reader.
reader = java.io.InputStreamReader(in);
9. Read Data from Serial Port and Close Reader
The example reads from the serial port, by calling the read method on the
InputStreamReader object reader for each available byte. The read statement
uses MATLAB array concatenation to add each newly read byte to the array of
bytes already read. After reading the data, the example calls close on reader
to close the input stream reader.
result = [];
for k = 1:numAvail
result = [result read(reader)];
end
close(reader);
10. Close the Serial Port
The example closes the serial port, by calling close on the serialPort object.
close(serialPort);
11. Convert Input Argument to a MATLAB Character Array
The last statement of the example uses the MATLAB function, char, to convert
the array input bytes (integers) to an array of characters:
result = char(result);
5-71
5
Calling Java from MATLAB
Running the serialexample Program
The value of result depends upon whether your system’s COM1 port is cabled
to an oscilloscope. If you have run the example with an oscilloscope, you see the
result of reading the serial port.
result =
45
If you run the example without an oscilloscope attached, there is no data to
read. In that case, you see an empty character array.
result =
''
5-72
Example – Creating and Using a Phone Book
Example – Creating and Using a Phone Book
The example’s main function, phonebook, can be called either with no
arguments, or with one argument, which is the key of an entry that exists in
the phone book. The function first determines the directory to use for the phone
book file.
If no phone book file exists, it creates one by constructing a
java.io.FileOutputStream object, and then closing the output stream. Next,
it creates a data dictionary by constructing an object of the Java API class,
java.util.Properties, which is a subclass of java.util.Hashtable for
storing key/value pairs in a hash table. For the phonebook program, the key is
a name, and the value is one or more telephone numbers.
The phonebook function creates and opens an input stream for reading by
constructing a java.io.FileInputStream object. It calls load on that object to
load the hash table contents, if it exists. If the user passed the key to an entry
to look up, it looks up the entry by calling pb_lookup, which finds and displays
it. Then, the phonebook function returns.
If phonebook was called without the name argument, it then displays a textual
menu of the available phone book actions:
• Look up an entry
• Add an entry
• Remove an entry
• Change the phone number(s) in an entry
• List all entries
The menu also has a selection to exit the program. The function uses MATLAB
functions to display the menu and to input the user selection.
The phonebook function iterates accepting user selections and performing the
requested phone book action until the user selects the menu entry to exit. The
phonebook function then opens an output stream for the file by constructing a
java.io.FileOutputStream object. It calls save on the object to write the
current data dictionary to the phone book file. It finally closes the output
stream and returns.
5-73
5
Calling Java from MATLAB
Description of Function phonebook
The major tasks performed by phonebook are:
1. Determine the Data Directory and Full Filename
The first statement assigns the phone book filename, 'myphonebook', to the
variable pbname. If the phonebook program is running on a PC, it calls the
java.lang.System static method getProperty to find the directory to use for
the data dictionary. This will be set to the user’s current working directory.
Otherwise, it uses MATLAB function getenv to determine the directory, using
the system variable HOME which you can define beforehand to anything you
like. It then assigns to pbname the full pathname, consisting of the data
directory and filename 'myphonebook'.
function phonebook(varargin)
pbname = 'myphonebook'; % name of data dictionary
if ispc
datadir = char(java.lang.System.getProperty('user.dir'));
else
datadir = getenv('HOME');
end;
pbname = fullfile(datadir, pbname);
2. If Needed, Create a File Output Stream
If the phonebook file does not already exist, phonebook asks the user whether
to create a new one. If the user answers y, phonebook creates a new phone book
by constructing a FileOutputStream object. In the try clause of a try-catch
block, the argument pbname passed to the FileOutputStream constructor is the
full name of the file that the constructor creates and opens. The next statement
closes the file by calling close on the FileOutputStream object FOS. If the
output stream constructor fails, the catch statement prints a message and
terminates the program.
if ~exist(pbname)
disp(sprintf('Data file %s does not exist.', pbname));
r = input('Create a new phone book (y/n)?','s');
5-74
Example – Creating and Using a Phone Book
if r == 'y',
try
FOS = java.io.FileOutputStream(pbname);
FOS.close
catch
error(sprintf('Failed to create %s', pbname));
end;
else
return;
end;
end;
3. Create a Hash Table
The example constructs a java.util.Properties object to serve as the hash
table for the data dictionary.
pb_htable = java.util.Properties;
4. Create a File Input Stream
In a try block, the example invokes a FileInputStream constructor with the
name of the phone book file, assigning the object to FIS. If the call fails, the
catch statement displays an error message and terminates the program.
try
FIS = java.io.FileInputStream(pbname);
catch
error(sprintf('Failed to open %s for reading.', pbname));
end;
5. Load the Phone Book Keys and Close the File Input Stream
The example calls load on the FileInputStream object FIS, to load the phone
book keys and their values (if any) into the hash table. It then closes the file
input stream.
pb_htable.load(FIS);
FIS.close;
5-75
5
Calling Java from MATLAB
6. Display the Action Menu and Get the User’s Selection
Within a while loop, several disp statements display a menu of actions that
the user can perform on the phone book. Then, an input statement requests the
user’s typed selection.
while 1
disp ' '
disp ' Phonebook Menu:'
disp ' '
disp ' 1. Look up a phone number'
disp ' 2. Add an entry to the phone book'
disp ' 3. Remove an entry from the phone book'
disp ' 4. Change the contents of an entry in the phone book'
disp ' 5. Display entire contents of the phone book'
disp ' 6. Exit this program'
disp ' '
s = input('Please type the number for a menu selection: ','s');
7. Invoke the Function to Perform A Phone Book Action
Still within the while loop, a switch statement provides a case to handle each
user selection. Each of the first five cases invokes the function to perform a
phone book action.
Case 1 prompts for a name that is a key to an entry. It calls isempty to
determine whether the user has entered a name. If a name has not been
entered, it calls disp to display an error message. If a name has been input, it
passes it to pb_lookup. The pb_lookup routine looks up the entry and, if it finds
it, displays the entry contents.
switch s
case '1',
name = input('Enter the name to look up: ','s');
if isempty(name)
disp 'No name entered'
else
pb_lookup(pb_htable, name);
end;
5-76
Example – Creating and Using a Phone Book
Case 2 calls pb_add, which prompts the user for a new entry and then adds it
to the phone book.
case '2',
pb_add(pb_htable);
Case 3 uses input to prompt for the name of an entry to remove. If a name has
not been entered, it calls disp to display an error message. If a name has been
input, it passes it to pb_remove.
case '3',
name=input('Enter the name of the entry to remove: ', 's');
if isempty(name)
disp 'No name entered'
else
pb_remove(pb_htable, name);
end;
Case 4 uses input to prompt for the name of an entry to change. If a name has
not been entered, it calls disp to display an error message. If a name has been
input, it passes it to pb_change.
case '4',
name=input('Enter the name of the entry to change: ', 's');
if isempty(name)
disp 'No name entered'
else
pb_change(pb_htable, name);
end;
Case 5 calls pb_listall to display all entries.
case '5',
pb_listall(pb_htable);
8. Exit by Creating an Output Stream and Saving the Phone Book
If the user has selected case 6 to exit the program, a try statement calls the
constructor for a FileOuputStream object, passing it the name of the phone
book. If the constructor fails, the catch statement displays an error message.
If the object is created, the next statement saves the phone book data by calling
save on the Properties object pb_htable, passing the FileOutputStream
5-77
5
Calling Java from MATLAB
object FOS and a descriptive header string. It then calls close on the
FileOutputStream object, and returns.
case '6',
try
FOS = java.io.FileOutputStream(pbname);
catch
error(sprintf('Failed to open %s for writing.',...
pbname));
end;
pb_htable.save(FOS,'Data file for phonebook program');
FOS.close;
return;
otherwise
disp 'That selection is not on the menu.'
end;
end;
Description of Function pb_lookup
Arguments passed to pb_lookup are the Properties object pb_htable and the
name key for the requested entry. The pb_lookup function first calls get on
pb_htable with the name key, on which support function pb_keyfilter is
called to change spaces to underscores. The get method returns the entry (or
null, if the entry is not found) to variable entry. Note that get takes an
argument of type java.lang.Object and also returns an argument of that
type. In this invocation, the key passed to get and the entry returned from it
are actually character arrays.
pb_lookup then calls isempty to determine whether entry is null. If it is, it
uses disp to display an message stating that the name was not found. If entry
is not null, it calls pb_display to display the entry.
function pb_lookup(pb_htable,name)
entry = pb_htable.get(pb_keyfilter(name));
if isempty(entry),
disp(sprintf('The name %s is not in the phone book',name));
else
pb_display(entry);
end
5-78
Example – Creating and Using a Phone Book
Description of Function pb_add
1. Input the Entry to Add
The pb_add function takes one argument, the Properties object pb_htable.
pb_add uses disp to prompt for an entry. Using the up-arrow (^) character as
a line delimiter, input inputs a name to the variable entry. Then, within a
while loop, it uses input to get another line of the entry into variable line. If
the line is empty, indicating that the user has finished the entry, the code
breaks out of the while loop. If the line is not empty, the else statement appends
line to entry and then appends the line delimiter. At the end, the strcmp checks
the possibility that no input was entered and, if that is the case, returns.
function pb_add(pb_htable)
disp 'Type the name for the new entry, followed by Enter.'
disp 'Then, type the phone number(s), one per line.'
disp 'To complete the entry, type an extra Enter.'
name = input(':: ','s');
entry=[name '^'];
while 1
line = input(':: ','s');
if isempty(line)
break;
else
entry=[entry line '^'];
end;
end;
if strcmp(entry, '^')
disp 'No name entered'
return;
end;
2. Add the Entry to the Phone Book
After the input has completed, pb_add calls put on pb_htable with the hash
key name (on which pb_keyfilter is called to change spaces to underscores)
and entry. It then displays a message that the entry has been added.
pb_htable.put(pb_keyfilter(name),entry);
disp ' '
disp(sprintf('%s has been added to the phone book.', name));
5-79
5
Calling Java from MATLAB
Description of Function pb_remove
1. Look For the Key in the Phone Book
Arguments passed to pb_remove are the Properties object pb_htable and the
name key for the entry to remove. The pb_remove function calls containsKey on
pb_htable with the name key, on which support function pb_keyfilter is
called to change spaces to underscores. If name is not in the phone book, disp
displays a message and the function returns.
function pb_remove(pb_htable,name)
if ~pb_htable.containsKey(pb_keyfilter(name))
disp(sprintf('The name %s is not in the phone book',name))
return
end;
2. Ask for Confirmation and If Given, Remove the Key
If the key is in the hash table, pb_remove asks for user confirmation. If the user
confirms the removal by entering y, pb_remove calls remove on pb_htable with
the (filtered) name key, and displays a message that the entry has been
removed. If the user enters n, the removal is not performed and disp displays
a message that the removal has not been performed.
r = input(sprintf('Remove entry %s (y/n)? ',name), 's');
if r == 'y'
pb_htable.remove(pb_keyfilter(name));
disp(sprintf('%s has been removed from the phone book',name))
else
disp(sprintf('%s has not been removed',name))
end;
Description of Function pb_change
1. Find the Entry to Change, and Confirm
Arguments passed to pb_change are the Properties object pb_htable and the
name key for the requested entry. The pb_change function calls get on
pb_htable with the name key, on which pb_keyfilter is called to change
spaces to underscores. The get method returns the entry (or null, if the entry
is not found) to variable entry. pb_change calls isempty to determine whether
the entry is empty. If the entry is empty, pb_change displays a message that
5-80
Example – Creating and Using a Phone Book
the name will be added to the phone book, and allows the user to enter the
phone number(s) for the entry.
If the entry is found, in the else clause, pb_change calls pb_display to display
the entry. It then uses input to ask the user to confirm the replacement. If the
user enters anything other than y, the function returns.
function pb_change(pb_htable,name)
entry = pb_htable.get(pb_keyfilter(name));
if isempty(entry)
disp(sprintf('The name %s is not in the phone book', name));
return;
else
pb_display(entry);
r = input('Replace phone numbers in this entry (y/n)? ','s');
if r ~= 'y'
return;
end;
end;
2. Input New Phone Number(s) and Change the Phone Book Entry
pb_change uses disp to display a prompt for new phone number(s). Then,
pb_change inputs data into variable entry, with the same statements
described in “1. Input the Entry to Add” on page 5-79.
Then, to replace the existing entry with the new one, pb_change calls put on
pb_htable with the (filtered) key name and the new entry. It then displays a
message that the entry has been changed.
disp 'Type in the new phone number(s), one per line.'
disp 'To complete the entry, type an extra Enter.'
disp(sprintf(':: %s', name));
entry=[name '^'];
while 1
line = input(':: ','s');
if isempty(line)
break;
else
entry=[entry line '^'];
end;
end;
5-81
5
Calling Java from MATLAB
pb_htable.put(pb_keyfilter(name),entry);
disp ' '
disp(sprintf('The entry for %s has been changed', name));
Description of Function pb_listall
The pb_listall function takes one argument, the Properties object
pb_htable. The function calls propertyNames on the pb_htable object to return
to enum a java.util.Enumeration object, which supports convenient
enumeration of all the keys. In a while loop, pb_listall calls
hasMoreElements on enum, and if it returns true, pb_listall calls nextElement
on enum to return the next key. It then calls pb_display to display the key and
entry, which it retrieves by calling get on pb_htable with the key.
function pb_listall(pb_htable)
enum = pb_htable.propertyNames;
while enum.hasMoreElements
key = enum.nextElement;
pb_display(pb_htable.get(key));
end;
Description of Function pb_display
The pb_display function takes an argument entry, which is a phone book
entry. After displaying a horizontal line, pb_display calls MATLAB function
strtok to extract the first line the entry, up to the line delimiter (^), into t and
and the remainder into r. Then, within a while loop that terminates when t is
empty, it displays the current line in t. Then it calls strtok to extract the next
line from r, into t. When all lines have been displayed, pb_display indicates
the end of the entry by displaying another horizontal line.
function pb_display(entry)
disp ' '
disp '-------------------------'
[t,r] = strtok(entry,'^');
while ~isempty(t)
disp(sprintf(' %s',t));
[t,r] = strtok(r,'^');
end;
disp '-------------------------'
5-82
Example – Creating and Using a Phone Book
Description of Function pb_keyfilter
The pb_keyfilter function takes an argument key, which is a name used as a
key in the hash table, and either filters it for storage or unfilters it for display.
The filter, which replaces each space in the key with an underscore (_), makes
the key usable with the methods of java.util.Properties.
function out = pb_keyfilter(key)
if ~isempty(findstr(key,' '))
out = strrep(key,' ','_');
else
out = strrep(key,'_',' ');
end;
Running the phonebook Program
In this sample run, a user invokes phonebook with no arguments. The user
selects menu action 5, which displays the two entries currently in the phone
book (all entries are fictitious). Then, the user selects 2, to add an entry. After
adding the entry, the user again selects 5, which displays the new entry along
with the other two entries.
Phonebook Menu:
1.
2.
3.
4.
5.
6.
Look up a phone number
Add an entry to the phone book
Remove an entry from the phone book
Change the contents of an entry in the phone book
Display entire contents of the phone book
Exit this program
Please type the number for a menu selection: 5
------------------------Sylvia Woodland
(508) 111-3456
------------------------------------------------Russell Reddy
(617) 999-8765
-------------------------
5-83
5
Calling Java from MATLAB
Phonebook Menu:
1.
2.
3.
4.
5.
6.
Look up a phone number
Add an entry to the phone book
Remove an entry from the phone book
Change the contents of an entry in the phone book
Display entire contents of the phone book
Exit this program
Please type the number for a menu selection: 2
Type the name for the new entry, followed by Enter.
Then, type the phone number(s), one per line.
To complete the entry, type an extra Enter.
:: BriteLites Books
:: (781) 777-6868
::
BriteLites Books has been added to the phone book.
Phonebook Menu:
1.
2.
3.
4.
5.
6.
Look up a phone number
Add an entry to the phone book
Remove an entry from the phone book
Change the contents of an entry in the phone book
Display entire contents of the phone book
Exit this program
Please type the number for a menu selection: 5
------------------------BriteLites Books
(781) 777-6868
-------------------------
------------------------Sylvia Woodland
5-84
Example – Creating and Using a Phone Book
(508) 111-3456
------------------------------------------------Russell Reddy
(617) 999-8765
-------------------------
5-85
5
Calling Java from MATLAB
5-86
6
Importing and Exporting
Data
Using MAT-Files . . . . . . . . . .
Importing Data to MATLAB . . . . . .
Exporting Data from MATLAB . . . . .
Exchanging Data Files Between Platforms
Reading and Writing MAT-Files . . . .
Finding Associated Files . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Examples of MAT-Files . . .
Creating a MAT-File in C . . .
Reading a MAT-File in C . . .
Creating a MAT-File in Fortran .
Reading a MAT-File in Fortran .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 6-11
. 6-11
. 6-16
. 6-21
. 6-24
Compiling and Linking MAT-File Programs .
Masking Floating Point Exceptions . . . . . .
Compiling and Linking on UNIX . . . . . . .
Compiling and Linking on Windows . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
6-3
6-3
6-4
6-5
6-6
6-8
6-28
6-28
6-29
6-30
6
Importing and Exporting Data
You can use MAT-files, the data file format MATLAB uses for saving data to
disk, to import data to and export data from the MATLAB environment.
MAT-files provide a convenient mechanism for moving your MATLAB data
between different platforms in a highly portable manner. In addition, they
provide a means to import and export your data to other stand-alone MATLAB
applications.
To simplify your use of MAT-files in applications outside of MATLAB, we
provide a library of access routines that you can use in your own C or Fortran
programs to read and write MAT-files. Programs that access MAT-files also use
the mx API routines discussed in this book.
The following list summarizes the contents of this chapter:
• “Using MAT-Files”
• “Examples of MAT-Files”
• “Compiling and Linking MAT-File Programs”
For additional information and support in building your applications, see the
section entitled, “Additional Information” on page 1-42.
6-2
Using MAT-Files
Using MAT-Files
This section describes the various techniques for importing data to and
exporting data from the MATLAB environment. The main topics that are
discussed are:
• “Importing Data to MATLAB”
• “Exporting Data from MATLAB”
• “Exchanging Data Files Between Platforms”
• “Reading and Writing MAT-Files”
• “Finding Associated Files”
The most important approach to importing and exporting data involves the use
of MAT-files, the data file format that MATLAB uses for saving data to your
disk. MAT-files provide a convenient mechanism for moving your MATLAB
data between different platforms and for importing and exporting your data to
other stand-alone MATLAB applications.
To simplify your use of MAT-files in applications outside of MATLAB, we have
developed a library of access routines with a mat prefix that you can use in your
own C or Fortran programs to read and write MAT-files. Programs that access
MAT-files also use the mx prefixed API routines discussed in Chapter 2,
“Creating C Language MEX-Files” and Chapter 3, “Creating Fortran
MEX-Files.”
Importing Data to MATLAB
You can introduce data from other programs into MATLAB by several
methods. The best method for importing data depends on how much data there
is, whether the data is already in machine-readable form, and what format the
data is in. Here are some choices. Select the one that best meets your needs.
• Enter the data as an explicit list of elements.
If you have a small amount of data, less than 10-15 elements, it is easy to
type the data explicitly using brackets [ ]. This method is awkward for larger
amounts of data because you can’t edit your input if you make a mistake.
• Create data in an M-file.
Use your text editor to create an M-file that enters your data as an explicit
list of elements. This method is useful when the data isn’t already in
6-3
6
Importing and Exporting Data
computer-readable form and you have to type it in. Essentially the same as
the first method, this method has the advantage of allowing you to use your
editor to change the data and correct mistakes. You can then just rerun your
M-file to re-enter the data.
• Load data from an ASCII flat file.
A flat file stores the data in ASCII form, with fixed-length rows terminated
with new lines (carriage returns) and with spaces separating the numbers.
You can edit ASCII flat files using a normal text editor. Flat files can be read
directly into MATLAB using the load command. The result is to create a
variable with the same name as the filename.
See the load function reference page for more information.
• Read data using MATLAB’s I/O functions.
You can read data using fopen, fread, and MATLAB’s other low-level I/O
functions. This method is useful for loading data files from other applications
that have their own established file formats.
• Write a MEX-file to read the data.
This is the method of choice if subroutines are already available for reading
data files from other applications. See the section, “Introducing MEX-Files”
on page 1-3, for more information.
• Write a program to translate your data.
You can write a program in C or Fortran to translate your data into MAT-file
format. You can then read the MAT-file into MATLAB using the load
command. Refer to the section, “Reading and Writing MAT-Files” on
page 6-6, for more information.
Exporting Data from MATLAB
There are several methods for getting MATLAB data back to the outside world:
• Create a diary file.
For small matrices, use the diary command to create a diary file and display
the variables, echoing them into this file. You can use your text editor to
manipulate the diary file at a later time. The output of diary includes the
MATLAB commands used during the session, which is useful for inclusion
into documents and reports.
6-4
Using MAT-Files
• Use the Save command.
Save the data in ASCII form using the save command with the -ascii
option. For example,
A = rand(4,3);
save temp.dat A -ascii
creates an ASCII file called temp.dat containing
1.3889088e-001
2.0276522e-001
1.9872174e-001
6.0379248e-001
2.7218792e-001
1.9881427e-001
1.5273927e-002
7.4678568e-001
4.4509643e-001
9.3181458e-001
4.6599434e-001
4.1864947e-001
The -ascii option supports two-dimensional double and character arrays
only. Multidimensional arrays, cell arrays, and structures are not supported.
See the save function reference page for more information.
• Use MATLAB I/O functions.
Write the data in a special format using fopen, fwrite, and the other
low-level I/O functions. This method is useful for writing data files in the file
formats required by other applications.
• Develop a MEX-file to write the data.
You can develop a MEX-file to write the data. This is the method of choice if
subroutines are already available for writing data files in the form needed by
other applications. See the section, “Introducing MEX-Files” on page 1-3, for
more information.
• Translate data from a MAT-file.
You can write out the data as a MAT-file using the save command. You can
then write a program in C or Fortran to translate the MAT-file into your own
special format. See the section, “Reading and Writing MAT-Files” on
page 6-6, for more information.
Exchanging Data Files Between Platforms
You may want to work with MATLAB implementations on several different
computer systems, or need to transmit MATLAB applications to users on other
systems. MATLAB applications consist of M-files containing functions and
scripts, and MAT-files containing binary data.
6-5
6
Importing and Exporting Data
Both types of files can be transported directly between machines: M-files
because they are platform independent and MAT-files because they contain a
machine signature in the file header. MATLAB checks the signature when it
loads a file and, if a signature indicates that a file is foreign, performs the
necessary conversion.
Using MATLAB across several different machine architectures requires a
facility for exchanging both binary and ASCII data between the various
machines. Examples of this type of facility include FTP, NFS, Kermit, and
other communication programs. When using these programs, be careful to
transmit binary MAT-files in binary file mode and ASCII M-files in ASCII file
mode. Failure to set these modes correctly corrupts the data.
Reading and Writing MAT-Files
The save command in MATLAB saves the MATLAB arrays currently in
memory to a binary disk file called a MAT-file. The term MAT-file is used
because these files have the extension .mat. The load command performs the
reverse operation. It reads the MATLAB arrays from a MAT-file on disk back
into MATLAB’s workspace.
A MAT-file may contain one or more of any of the data types supported in
MATLAB 5 or later, including strings, matrices, multidimensional arrays,
structures, and cell arrays. MATLAB writes the data sequentially onto disk as
a continuous byte stream.
MAT-File Interface Library
The MAT-file interface library contains a set of routines for reading and
writing MAT-files. You can call these routines from within your own C and
Fortran programs. We recommend that you use these routines, rather than
attempt to write your own code, to perform these operations. By using the
routines in this library, you will be insulated from future changes to the
MAT-file structure.
6-6
Using MAT-Files
The MAT-file library contains routines for reading and writing MAT-files. They
all begin with the three-letter prefix mat. These tables list all the available
MAT-functions and their purposes.
Table 6-1: C MAT-File Routines
MAT-Function
Purpose
matOpen
Open a MAT-file
matClose
Close a MAT-file
matGetDir
Get a list of MATLAB arrays from a MAT-file
matGetFp
Get an ANSI C file pointer to a MAT-file
matGetArray
Read a MATLAB array from a MAT-file
matPutArray
Write a MATLAB array to a MAT-file
matGetNextArray
Read the next MATLAB array from a MAT-file
matDeleteArray
Remove a MATLAB array from a MAT-file
matPutArrayAsGlobal
Put a MATLAB array into a MAT-file such
that the load command will place it into the
global workspace
matGetArrayHeader
Load a MATLAB array header from a MAT-file
(no data)
matGetNextArrayHeader
Load the next MATLAB array header from a
MAT-file (no data)
Table 6-2: Fortran MAT-File Routines
MAT-Function
Purpose
matOpen
Open a MAT-file
matClose
Close a MAT-file
6-7
6
Importing and Exporting Data
Table 6-2: Fortran MAT-File Routines (Continued)
MAT-Function
Purpose
matGetDir
Get a list of MATLAB arrays from a
MAT-file
matGetMatrix
Get a named MATLAB array from a
MAT-file
matPutMatrix
Put a MATLAB array into a MAT-file
matGetNextMatrix
Get the next sequential MATLAB array
from a MAT-file
matDeleteMatrix
Remove a MATLAB array from a MAT-file
matGetString
Read a MATLAB string from a MAT-file
matPutString
Write a MATLAB string to a MAT-file
Finding Associated Files
A collection of files associated with reading and writing MAT-files is located on
your disk. The following table, MAT-Function Subdirectories, lists the path to
the required subdirectories for importing and exporting data using
MAT-functions.
Table 6-3: MAT-Function Subdirectories
Platform
Contents
Directories
Windows
Include Files
<matlab>\extern\include
Libraries
<matlab>\bin\win32
Examples
<matlab>\extern\examples/eng_mat
Include Files
<matlab>/extern/include
Libraries
<matlab>/extern/lib/$arch
Examples
<matlab>/extern/examples/eng_mat
UNIX
6-8
Using MAT-Files
Include Files
The include directory holds header files containing function declarations with
prototypes for the routines that you can access in the API Library. These files
are the same for both Windows and UNIX. Included in the subdirectory are:
• matrix.h, the header file that defines MATLAB array access and creation
methods
• mat.h, the header file that defines MAT-file access and creation methods
Libraries
The subdirectory that contains shared (dynamically linkable) libraries for
linking your programs is platform dependent.
Shared Libraries on Windows. The bin subdirectory contains the shared libraries
for linking your programs:
• libmat.dll, the library of MAT-file routines (C and Fortran)
• libmx.dll, the library of array access and creation routines
Shared Libraries on UNIX. The extern/lib/$arch subdirectory, where $arch is
your machine’s architecture, contains the shared libraries for linking your
programs. For example, on sol2, the subdirectory is extern/lib/sol2:
• libmat.so, the library of MAT-file routines (C and Fortran)
• libmx.so, the library of array access and creation routines
Example Files
The examples/eng_mat subdirectory contains C and Fortran source code for a
number of example files that demonstrate how to use the MAT-file routines.
The source code files are the same for both Windows and UNIX.
6-9
6
Importing and Exporting Data
Table 6-4: C and Fortran Examples
Library
Description
matcreat.c
Example C program that demonstrates how to use the
library routines to create a MAT-file that can be loaded
into MATLAB
matdgns.c
Example C program that demonstrates how to use the
library routines to read and diagnose a MAT-file
matdemo1.f
Example Fortran program that demonstrates how to call
the MATLAB MAT-file functions from a Fortran program
matdemo2.f
Example Fortran program that demonstrates how to use
the library routines to read in the MAT-file created by
matdemo1.f and describe its contents
For additional information about the MATLAB API files and directories, see
“Additional Information” on page 1-42.
6-10
Examples of MAT-Files
Examples of MAT-Files
This section includes C and Fortran examples of writing, reading, and
diagnosing MAT-files. The examples cover the following topics:
• “Creating a MAT-File in C”
• “Reading a MAT-File in C”
• “Creating a MAT-File in Fortran”
• “Reading a MAT-File in Fortran”
Creating a MAT-File in C
This sample program illustrates how to use the library routines to create a
MAT-file that can be loaded into MATLAB. The program also demonstrates
how to check the return values of MAT-function calls for read or write failures.
/*
* MAT-file creation program
*
* See the MATLAB API Guide for compiling information.
*
* Calling syntax:
*
*
matcreat
*
* Create a MAT-file which can be loaded into MATLAB.
*
* This program demonstrates the use of the following functions:
*
* matClose
* matGetArray
* matOpen
* matPutArray
* matPutArrayAsGlobal
*
* Copyright 1984-2000 The MathWorks, Inc.
*/
/* $Revision: 1.10 $ */
#include <stdio.h>
#include <string.h> /* For strcmp() */
6-11
6
Importing and Exporting Data
#include <stdlib.h> /* For EXIT_FAILURE, EXIT_SUCCESS */
#include "mat.h"
#define BUFSIZE 256
int main() {
MATFile *pmat;
mxArray *pa1, *pa2, *pa3;
double data[9] = { 1.0, 4.0, 7.0, 2.0, 5.0, 8.0, 3.0, 6.0, 9.0 };
const char *file = "mattest.mat";
char str[BUFSIZE];
int status;
printf("Creating file %s...\n\n", file);
pmat = matOpen(file, "w");
if (pmat == NULL) {
printf("Error creating file %s\n", file);
printf("(Do you have write permission in this directory?)\n");
return(EXIT_FAILURE);
}
pa1 = mxCreateDoubleMatrix(3,3,mxREAL);
if (pa1 == NULL) {
printf("%s : Out of memory on line %d\n",
__FILE__, __LINE__);
printf("Unable to create mxArray.\n");
return(EXIT_FAILURE);
}
mxSetName(pa1, "LocalDouble");
pa2 = mxCreateDoubleMatrix(3,3,mxREAL);
if (pa2 == NULL) {
printf("%s : Out of memory on line %d\n",
__FILE__, __LINE__);
printf("Unable to create mxArray.\n");
return(EXIT_FAILURE);
}
mxSetName(pa2, "GlobalDouble");
memcpy((void *)(mxGetPr(pa2)), (void *)data, sizeof(data));
6-12
Examples of MAT-Files
pa3 = mxCreateString("MATLAB: the language of technical
computing");
if (pa3 == NULL) {
printf("%s : Out of memory on line %d\n",
__FILE__, __LINE__);
printf("Unable to create string mxArray.\n");
return(EXIT_FAILURE);
}
mxSetName(pa3, "LocalString");
status = matPutArray(pmat, pa1);
if (status != 0) {
printf("%s : Error using matPutArray on line %d\n",
__FILE__, __LINE__);
return(EXIT_FAILURE);
}
status = matPutArrayAsGlobal(pmat, pa2);
if (status != 0) {
printf("Error using matPutArrayAsGlobal\n");
return(EXIT_FAILURE);
}
status = matPutArray(pmat, pa3);
if (status != 0) {
printf("%s : Error using matPutArray on line %d\n",
__FILE__, __LINE__);
return(EXIT_FAILURE);
}
/*
* Ooops! we need to copy data before writing the array. (Well,
* ok, this was really intentional.) This demonstrates that
* matPutArray will overwrite an existing array in a MAT-file.
*/
memcpy((void *)(mxGetPr(pa1)), (void *)data, sizeof(data));
status = matPutArray(pmat, pa1);
6-13
6
Importing and Exporting Data
if (status != 0) {
printf("%s : Error using matPutArray on line %d\n",
__FILE__, __LINE__);
return(EXIT_FAILURE);
}
/* clean up */
mxDestroyArray(pa1);
mxDestroyArray(pa2);
mxDestroyArray(pa3);
if (matClose(pmat) != 0) {
printf("Error closing file %s\n",file);
return(EXIT_FAILURE);
}
/*
* Reopen file and verify its contents with matGetArray
*/
pmat = matOpen(file, "r");
if (pmat == NULL) {
printf("Error reopening file %s\n", file);
return(EXIT_FAILURE);
}
/*
* Read in each array we just wrote
*/
pa1 = matGetArray(pmat, "LocalDouble");
if (pa1 == NULL) {
printf("Error reading existing matrix LocalDouble\n");
return(EXIT_FAILURE);
}
if (mxGetNumberOfDimensions(pa1) != 2) {
printf("Error saving matrix: result does not have two
dimensions\n");
return(EXIT_FAILURE);
}
6-14
Examples of MAT-Files
pa2 = matGetArray(pmat, "GlobalDouble");
if (pa2 == NULL) {
printf("Error reading existing matrix GlobalDouble\n");
return(EXIT_FAILURE);
}
if (!(mxIsFromGlobalWS(pa2))) {
printf("Error saving global matrix: result is not global\n");
return(EXIT_FAILURE);
}
pa3 = matGetArray(pmat, "LocalString");
if (pa3 == NULL) {
printf("Error reading existing matrix LocalDouble\n");
return(EXIT_FAILURE);
}
status = mxGetString(pa3, str, sizeof(str));
if(status != 0) {
printf("Not enough space. String is truncated.");
return(EXIT_FAILURE);
}
if (strcmp(str, "MATLAB: the language of technical
computing")) {
printf("Error saving string: result has incorrect
contents\n");
return(EXIT_FAILURE);
}
/* clean up before exit */
mxDestroyArray(pa1);
mxDestroyArray(pa2);
mxDestroyArray(pa3);
if (matClose(pmat) != 0) {
printf("Error closing file %s\n",file);
return(EXIT_FAILURE);
}
printf("Done\n");
return(EXIT_SUCCESS);
}
6-15
6
Importing and Exporting Data
To produce an executable version of this example program, compile the file and
link it with the appropriate library. Details on how to compile and link
MAT-file programs on the various platforms are discussed in the section,
“Compiling and Linking MAT-File Programs” on page 6-28.
Once you have compiled and linked your MAT-file program, you can run the
stand-alone application you have just produced. This program creates a
MAT-file, mattest.mat, that can be loaded into MATLAB. To run the
application, depending on your platform, either double-click on its icon or enter
matcreat at the system prompt.
matcreat
Creating file mattest.mat...
To verify that the MAT-file has been created, at the MATLAB prompt enter
whos -file mattest.mat
Name
Size
GlobalDouble
LocalDouble
LocalString
3x3
3x3
1x43
Bytes
72
72
86
Class
double array (global)
double array
char array
Grand total is 61 elements using 230 bytes
Reading a MAT-File in C
This sample program illustrates how to use the library routines to read and
diagnose a MAT-file.
/*
/*
*
*
*
*
*
*
*
*
*
*
6-16
$Revision: 1.1 $ */
MAT-file diagnose program
Calling syntax:
matdgns <matfile.mat>
It diagnoses the MAT-file named <matfile.mat>.
This program demonstrates the use of the following functions:
Examples of MAT-Files
*
*
*
*
*
*
*
*/
matClose
matGetDir
matGetNextArray
matGetNextArrayHeader
matOpen
Copyright (c) 1984-1998 The MathWorks, Inc.
#include
#include
#include
#include
<stdio.h>
<stdlib.h>
"string.h"
"mat.h"
int diagnose(const char *file) {
MATFile *pmat;
char **dir;
int ndir;
int i;
mxArray *pa;
printf("Reading file %s...\n\n", file);
/* Open file to get directory. */
pmat = matOpen(file, "r");
if (pmat == NULL) {
printf("Error opening file %s\n", file);
return(1);
}
/* Get directory of MAT-file. */
dir = matGetDir(pmat, &ndir);
if (dir == NULL) {
printf("Error reading directory of file %s\n", file);
return(1);
} else {
printf("Directory of %s:\n", file);
for (i=0; i < ndir; i++)
printf("%s\n",dir[i]);
}
6-17
6
Importing and Exporting Data
mxFree(dir);
/* In order to use matGetNextXXX correctly, reopen file to read
in headers. */
if (matClose(pmat) != 0) {
printf("Error closing file %s\n",file);
return(1);
}
pmat = matOpen(file, "r");
if (pmat == NULL) {
printf("Error reopening file %s\n", file);
return(1);
}
/* Get headers of all variables. */
printf("\nExamining the header for each variable:\n");
for (i=0; i < ndir; i++) {
pa = matGetNextArrayHeader(pmat);
if (pa == NULL) {
printf("Error reading in file %s\n", file);
return(1);
}
/* Diagnose header pa. */
printf("According to its header, array %s has %d
dimensions\n", mxGetName(pa),
mxGetNumberOfDimensions(pa));
if (mxIsFromGlobalWS(pa))
printf(" and was a global variable when saved\n");
else
printf(" and was a local variable when saved\n");
mxDestroyArray(pa);
}
/* Reopen file to read in actual arrays. */
if (matClose(pmat) != 0) {
printf("Error closing file %s\n",file);
return(1);
}
pmat = matOpen(file, "r");
6-18
Examples of MAT-Files
if (pmat == NULL) {
printf("Error reopening file %s\n", file);
return(1);
}
/* Read in each array. */
printf("\nReading in the actual array contents:\n");
for (i=0; i<ndir; i++) {
pa = matGetNextArray(pmat);
if (pa == NULL) {
printf("Error reading in file %s\n", file);
return(1);
}
/* Diagnose array pa. */
printf("According to its contents, array %s has %d
dimensions\n", mxGetName(pa),
mxGetNumberOfDimensions(pa));
if (mxIsFromGlobalWS(pa))
printf(" and was a global variable when saved\n");
else
printf(" and was a local variable when saved\n");
mxDestroyArray(pa);
}
if (matClose(pmat) != 0) {
printf("Error closing file %s\n",file);
return(1);
}
printf("Done\n");
return(0);
}
int main(int argc, char **argv)
{
int result;
if (argc > 1)
result = diagnose(argv[1]);
6-19
6
Importing and Exporting Data
else {
result = 0;
printf("Usage: matdgns <matfile>");
printf("where <matfile> is the name of the MAT-file");
printf("to be diagnosed");
}
return (result==0)?EXIT_SUCCESS:EXIT_FAILURE;
}
After compiling and linking this program, you can view its results.
matdgns mattest.mat
Reading file mattest.mat...
Directory of mattest.mat:
GlobalDouble
LocalString
LocalDouble
Examining
According
and was
According
and was
According
and was
the header for each variable:
to its header, array GlobalDouble has 2 dimensions
a global variable when saved
to its header, array LocalString has 2 dimensions
a local variable when saved
to its header, array LocalDouble has 2 dimensions
a local variable when saved
Reading in the actual array contents:
According to its contents, array GlobalDouble has 2 dimensions
and was a global variable when saved
According to its contents, array LocalString has 2 dimensions
and was a local variable when saved
According to its contents, array LocalDouble has 2 dimensions
and was a local variable when saved
Done
6-20
Examples of MAT-Files
Creating a MAT-File in Fortran
This example creates a MAT-file, matdemo.mat.
C $Revision: 1.6 $
C
C
matdemo1.f
C
C
This is a simple program that illustrates how to call the
C
MATLAB MAT-file functions from a Fortran program. This
C
demonstration focuses on writing MAT-files.
C
C
Copyright (c) 1984-2000 The MathWorks, Inc.
C
C
C
C
C
matdemo1 - Create a new MAT-file from scratch.
program matdemo1
C-------------------------------------------------------------C
(integer) Replace integer by integer*8 on the DEC Alpha
C
platform.
C
integer matOpen, mxCreateFull, mxCreateString
integer matGetMatrix, mxGetPr
integer mp, pa1, pa2, pa3
C-------------------------------------------------------------C
C
Other variable declarations here
C
integer status, matClose
double precision dat(9)
data dat / 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0 /
C
C
Open MAT-file for writing.
C
write(6,*) 'Creating MAT-file matdemo.mat ...'
mp = matOpen('matdemo.mat', 'w')
6-21
6
Importing and Exporting Data
if (mp .eq. 0) then
write(6,*) 'Can''t open ''matdemo.mat'' for writing.'
write(6,*) '(Do you have write permission in this directory?)'
stop
end if
C
C
C
Create variables.
pa1 = mxCreateFull(3,3,0)
call mxSetName(pa1, 'Numeric')
C
pa2 = mxCreateString('MATLAB: The language of computing')
call mxSetName(pa2, 'String')
C
pa3 = mxCreateString('MATLAB: The language of computing')
call mxSetName(pa3, 'String2')
C
call matPutMatrix(mp, pa1)
call matPutMatrix(mp, pa2)
call matPutMatrix(mp, pa3)
C
C
C
C
C
C
Whoops! Forgot to copy the data into the first matrix -it's now blank. (Well, ok, this was deliberate.) This
demonstrates that matPutMatrix will overwrite existing
matrices.
call mxCopyReal8ToPtr(dat, mxGetPr(pa1), 9)
call matPutMatrix(mp, pa1)
C
C
C
Now, we'll delete String2 from the MAT-file.
call matDeleteMatrix(mp, 'String2')
C
C
C
C
Finally, read back in MAT-file to make sure we know what we
put into it.
status = matClose(mp)
6-22
Examples of MAT-Files
if (status .ne. 0) then
write(6,*) 'Error closing MAT-file'
stop
end if
C
mp = matOpen('matdemo.mat', 'r')
if (status .ne. 0) then
write(6,*) 'Can''t open ''matdemo.mat'' for reading.'
stop
end if
C
pa1 = matGetMatrix(mp, 'Numeric')
if (mxIsNumeric(pa1) .eq. 0) then
write(6,*) 'Invalid non-numeric matrix written to MAT-file'
stop
end if
C
pa2 = matGetMatrix(mp, 'String')
if (mxIsString(pa2) .eq. 0) then
write(6,*) 'Invalid non-numeric matrix written to MAT-file'
stop
end if
C
pa3 = matGetMatrix(mp, 'String2')
if (pa3 .ne. 0) then
write(6,*) 'String2 not deleted from MAT-file'
stop
end if
C
status = matClose(mp)
if (status .ne. 0) then
write(6,*) 'Error closing MAT-file'
stop
end if
C
write(6,*) 'Done creating MAT-file'
stop
end
6-23
6
Importing and Exporting Data
Once you have compiled and linked your MAT-file program, you can run the
stand-alone application you have just produced. This program creates a
MAT-file, matdemo.mat, that can be loaded into MATLAB. To run the
application, depending on your platform, either double-click on its icon or enter
matdemo1 at the system prompt.
matdemo1
Creating MAT-file matdemo.mat ...
Done creating MAT-file
To verify that the MAT-file has been created, at the MATLAB prompt enter
whos -file matdemo.mat
Name
Size
Numeric
String
3x3
1x33
Bytes
72
66
Class
double array
char array
Grand total is 42 elements using 138 bytes
Note For an example of a Windows stand-alone program (not MAT-file
specific), see engwindemo.c in the <matlab>\extern\examples\eng_mat
directory.
Reading a MAT-File in Fortran
This sample program illustrates how to use the library routines to read in the
MAT-file created by matdemo1.f and describe its contents.
C
C
C
C
C
C
C
C
C
C
6-24
matdemo2.f
This is a simple program that illustrates how to call the
MATLAB MAT-file functions from a Fortran program. This
demonstration focuses on reading MAT-files. It reads in
the MAT-file created by matdemo1.f and describes its
contents.
Copyright (c) 1996-2000 The MathWorks, Inc.
Examples of MAT-Files
C-------------------------------------------------------------C
$Revision: 1.7 $
C
program matdemo2
C-------------------------------------------------------------C
(integer) Replace integer by integer*8 on the DEC Alpha
C
platform.
C
integer matOpen, matGetDir, matGetNextMatrix
integer mp, dir, adir(100), pa
C-------------------------------------------------------------C
C
Other variable declarations here
C
integer
mxGetM, mxGetN, matClose
integer
ndir, i, stat
character*32 names(100), name, mxGetName
C
C------------------------------------------------------------C
Open file and read directory.
C------------------------------------------------------------C
mp = matOpen('matdemo.mat', 'r')
if (mp .eq. 0) then
write(6,*) 'Can''t open ''matdemo.mat''.'
stop
end if
C
C
Read directory.
C
dir = matgetdir(mp, ndir)
if (dir .eq. 0) then
write(6,*) 'Can''t read directory.'
stop
endif
C
C
Copy integer into an array of pointers.
C
call mxCopyPtrToPtrArray(dir, adir, ndir)
6-25
6
Importing and Exporting Data
C
C
C
Copy integer to character string
do 20 i=1,ndir
call mxCopyPtrToCharacter(adir(i), names(i), 32)
20 continue
C
write(6,*) 'Directory of Mat-file:'
do 30 i=1,ndir
write(6,*) names(i)
30 continue
C
stat = matClose(mp)
if (stat .ne. 0) then
write(6,*) 'Error closing ''matdemo.mat''.'
stop
end if
C
C------------------------------------------------------------C
Reopen file and read full arrays.
C------------------------------------------------------------C
mp = matOpen('matdemo.mat', 'r')
if (mp .eq. 0) then
write(6,*) 'Can''t open ''matdemo.mat''.'
stop
end if
C
C
Read directory.
C
write(6,*) 'Getting full array contents:'
pa = matGetNextMatrix(mp)
do while (pa .ne. 0)
C
C
Copy name to character string.
C
name = mxGetName(pa)
write(6,*) 'Retrieved ', name
write(6,*) ' With size ', mxGetM(pa), '-by-', mxGetN(pa)
pa = matGetNextMatrix(mp)
6-26
Examples of MAT-Files
end do
C
stat = matClose(mp)
if (stat .ne. 0) then
write(6,*) 'Error closing ''matdemo.mat''.'
stop
end if
stop
C
end
After compiling and linking this program, you can view its results.
matdemo2
Directory of Mat-file:
String
Numeric
Getting full array contents:
1
Retrieved String
With size
1-by- 33
3
Retrieved Numeric
With size
3-by- 3
6-27
6
Importing and Exporting Data
Compiling and Linking MAT-File Programs
This section describes the steps required to compile and link MAT-file
programs on UNIX and Windows systems. It begins by looking at a special
consideration for compilers that do not mask floating-point exceptions. Topics
covered are:
• “Masking Floating Point Exceptions”
• “Compiling and Linking on UNIX”
• “Compiling and Linking on Windows”
Masking Floating Point Exceptions
Certain mathematical operations can result in nonfinite values. For example,
division by zero results in the nonfinite IEEE value, inf. A floating-point
exception occurs when such an operation is performed. Because MATLAB uses
an IEEE model that supports nonfinite values such as inf and NaN, MATLAB
disables, or masks, floating-point exceptions.
Some compilers do not mask floating-point exceptions by default. This causes
MAT-file applications built with such compilers to terminate when a
floating-point exception occurs. Consequently, you need to take special
precautions when using these compilers to mask floating-point exceptions so
that your MAT-file application will perform properly.
Note MATLAB-based applications should never get floating-point
exceptions. If you do get a floating-point exception, verify that any third party
libraries that you link against do not enable floating-point exception handling.
The only compiler and platform on which you need to mask floating-point
exceptions is the Borland C++ compiler on Windows.
Borland C++ Compiler on Windows
To mask floating-point exceptions when using the Borland C++ compiler on
the Windows platform, you must add some code to your program. Include the
following at the beginning of your main() or WinMain() function, before any
calls to MATLAB API functions.
6-28
Compiling and Linking MAT-File Programs
#include <float.h>
.
.
.
_control87(MCW_EM,MCW_EM);
.
.
.
Compiling and Linking on UNIX
Under UNIX at runtime, you must tell the system where the API shared
libraries reside. These sections provide the necessary UNIX commands
depending on your shell and system architecture.
Setting Runtime Library Path
In C shell, the command to set the library path is
setenv LD_LIBRARY_PATH <matlab>/extern/lib/$Arch:$LD_LIBRARY_PATH
In Bourne shell, the commands to set the library path are
LD_LIBRARY_PATH=<matlab>/extern/lib/$Arch:$LD_LIBRARY_PATH
export LD_LIBRARY_PATH
where <matlab> is the MATLAB root directory and $Arch is your system
architecture (alpha, glnx86, sgi, sol2, hp700, or ibm_rs).
Note that the environment variable (LD_LIBRARY_PATH in this example) varies
on several platforms. The following table lists the different environment
variable names you should use on these systems.
Table 6-5: Environment Variable Names
Architecture
Environment Variable
HP700
SHLIB_PATH
IBM RS/6000
LIBPATH
It is convenient to place these commands in a startup script such as ~/.cshrc
for C shell or ~/.profile for Bourne shell.
6-29
6
Importing and Exporting Data
Using the Options File
MATLAB provides an options file, matopts.sh, that lets you use the mex script
to easily compile and link MAT-file applications. For example, to compile and
link the matcreat.c example, you can use
mex -f <matlab>/bin/matopts.sh <pathname>/matcreat.c
where <pathname> specifies the complete path to the specified file.
If you need to modify the options file for your particular compiler or platform,
use the -v switch to view the current compiler and linker settings and then
make the appropriate changes in a local copy of the matopts.sh file.
Compiling and Linking on Windows
To compile and link Fortran or C MAT-file programs, use the mex script with a
MAT options file. Table 1-2, Options Files, on page 1-17 lists the MAT options
files included in this release. All of these options files are located in
<matlab>\bin\win32\mexopts.
As an example, to compile and link the stand-alone MAT application
matcreat.c on Windows using MSVC (Version 5.0), use
mex -f <matlab>\bin\win32\mexopts\msvc50engmatopts.bat <pathname>\matcreat.c
where <pathname> specifies the complete path to the specified file. If you need
to modify the options file for your particular compiler, use the -v switch to view
the current compiler and linker settings and then make the appropriate
changes in a local copy of the options file.
6-30
7
ActiveX and DDE Support
Introducing MATLAB ActiveX Integration . . . . . . 7-3
ActiveX Concepts and Terminology . . . . . . . . . . . 7-3
MATLAB ActiveX Support Overview . . . . . . . . . . 7-4
MATLAB ActiveX Client Support . . . . . . . . . . 7-6
Using ActiveX Objects . . . . . . . . . . . . . . . . 7-6
Writing Event Handlers . . . . . . . . . . . . . . . 7-8
Additional ActiveX Client Information
Releasing Interfaces . . . . . . . . .
Using ActiveX Collections . . . . . . .
Converting Data . . . . . . . . . . .
Using MATLAB As a DCOM Server Client
MATLAB ActiveX Support Limitations . .
MATLAB Sample Control . . . . . . .
Using MATLAB As an Automation Client
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 7-10
. 7-10
. 7-10
. 7-11
. 7-12
. 7-13
. 7-13
. 7-13
MATLAB ActiveX Automation Server Support . . . . 7-16
MATLAB ActiveX Automation Methods . . . . . . . . . 7-17
MATLAB ActiveX Automation Properties . . . . . . . . . 7-20
Additional ActiveX Server Information
Launching the MATLAB ActiveX Server .
Specifying a Shared or Dedicated Server .
Using MATLAB As a DCOM Server . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Dynamic Data Exchange (DDE) . . . . . . . . . .
DDE Concepts and Terminology . . . . . . . . . . .
Accessing MATLAB As a Server . . . . . . . . . . .
The DDE Name Hierarchy . . . . . . . . . . . . . .
Example: Using Visual Basic and the MATLAB DDE Server
Using MATLAB As a Client . . . . . . . . . . . . .
DDE Advisory Links . . . . . . . . . . . . . . . .
.
.
.
.
7-21
7-21
7-21
7-22
. 7-23
. 7-23
. 7-25
. 7-26
. 7-29
. 7-31
. 7-32
7
ActiveX and DDE Support
ActiveX is a set of object-oriented technologies and tools that allow software
developers to integrate application-specific components from different vendors
into their own application solution. ActiveX Control components can be both
visually and programmatically integrated into an ActiveX control container,
such as a MATLAB figure window. ActiveX Automation allows MATLAB to
both control and be controlled by other ActiveX components.
Dynamic Data Exchange, or DDE, is a feature of Windows that allows two
programs to share data or send commands directly to each other. MATLAB
provides functions that use this data exchange to enable access between
MATLAB and other Windows applications in a wide range of contexts.
This chapter discusses the following topics:
• “Introducing MATLAB ActiveX Integration”
• “MATLAB ActiveX Client Support”
• “Writing Event Handlers”
• “Additional ActiveX Client Information”
• “MATLAB ActiveX Automation Server Support”
• “Additional ActiveX Server Information”
• “Dynamic Data Exchange (DDE)”
7-2
Introducing MATLAB ActiveX Integration
Introducing MATLAB ActiveX Integration
ActiveX is a Microsoft Windows protocol for component integration. Using
ActiveX, developers and end users can select application-specific, ActiveX
components produced by different vendors and seamlessly integrate them into
a complete application solution. For example, a single application may require
database access, mathematical analysis, and presentation quality business
graphs. Using ActiveX, a developer may choose a database access component
by one vendor, a business graph component by another, and integrate these
into a mathematical analysis package produced by yet a third.
ActiveX Concepts and Terminology
COM
ActiveX is a family of related object-oriented technologies that have a common
root, called the Component Object Model, or COM. Each object-oriented
language or environment has an object model that defines certain
characteristics of objects in that environment, such as how objects are located,
instantiated, or identified. COM defines the object model for all ActiveX
objects.
ActiveX Interfaces
Each ActiveX object supports one or more named interfaces. An interface is a
logically related collection of methods, properties, and events. Methods are
similar to function calls in that they are a request for the object to perform
some action. Properties are state variables maintained by the object, such as
the color of text, or the name of a file on which the control is acting. Events are
notifications that the control forwards back to its client (similar to Handle
Graphics® callbacks.)
For example, the sample control shipped with MATLAB has the following
methods, properties, and events:
• Methods
Redraw - causes the control to redraw
Beep - causes the control to beep
AboutBox - display the control’s “About” dialog
7-3
7
ActiveX and DDE Support
• Properties
Radius (integer) - sets the radius of the circle drawn by the control
Label (string) - text to be drawn in the control
• Events
Click - fired when the user clicks on the control
One important characteristic of COM is that it defines an object model in which
objects support multiple interfaces. Some interfaces are standard interfaces,
which are defined by Microsoft and are part of ActiveX, and some interfaces are
custom interfaces, which are defined by individual component vendors. In
order to use any ActiveX object, you must learn about which custom interfaces
it supports, and the interface’s methods, properties, and events. The ActiveX
object’s vendor provides this information.
MATLAB ActiveX Support Overview
MATLAB supports two ActiveX technologies: ActiveX control containment and
ActiveX Automation. ActiveX controls are application components that can be
both visually and programmatically integrated into an ActiveX control
container, such as MATLAB figure windows. Some examples of useful ActiveX
controls are the Microsoft Internet Explorer Web Browser control, the
Microsoft Windows Communications control for serial port access, and the
graphical user interface controls delivered with the Visual Basic development
environment.
ActiveX Automation allows MATLAB to both control and be controlled by other
ActiveX components. When MATLAB is controlled by another component, it is
acting as an automation server. When MATLAB controls another component,
MATLAB is the automation client, and the other component is the automation
server.
MATLAB automation server capabilities include the ability to execute
commands in the MATLAB workspace, and to get and put matrices directly
from and into the workspace. MATLAB automation client capabilities allow
MATLAB, through M-code, to programmatically instantiate and manipulate
automation servers. The MATLAB automation client capabilities are a subset
of the MATLAB control containment support, since you use the automation
client capabilities to manipulate controls as well as automation servers. In
other words, all ActiveX controls are ActiveX automation servers, but not all
automation servers are necessarily controls.
7-4
Introducing MATLAB ActiveX Integration
In general, servers that are not controls will not be physically or visually
embedded in the client application. (MATLAB is a good example — MATLAB
is not itself a control, but it is a server. So, MATLAB cannot be physically
embedded within another client. However, since MATLAB is a control
container, other ActiveX controls can be embedded within MATLAB.)
In addition, MATLAB ships with a very simple sample ActiveX control that
draws a circle on the screen and displays some text. This allows MATLAB users
to try out MATLAB’s ActiveX control support with a known control. For more
information, see “MATLAB Sample Control” on page 7-13.
7-5
7
ActiveX and DDE Support
MATLAB ActiveX Client Support
In order to use an ActiveX component with MATLAB or with any ActiveX
client, you first need to consult the documentation for that object and find out
the name of the object itself (known as the ProgID), as well as the names of the
interfaces, methods, properties, and events that the object uses. Once you have
this information, you can integrate that object with MATLAB by using the
ActiveX client support.
Using ActiveX Objects
You create an ActiveX control or server in MATLAB by creating an instance of
the MATLAB activex class. Each instance represents one interface to the
object.
Note This book uses ActiveX to refer to the generic ActiveX control/server
and activex to refer to the MATLAB class/object.
Creating ActiveX Objects
There are two commands used to create activex objects initially.
actxcontrol
Creates an ActiveX control
actxserver
Creates an ActiveX automation server
Manipulating the Interface
Once you create an activex object that represents an interface, you can
manipulate it by invoking methods on the object to perform various actions.
These methods are implemented for the activex class.
7-6
Method
Description
delete
Deletes an activex object
get
Gets a property value from an interface or displays a
list of properties
MATLAB ActiveX Client Support
invoke
Invokes a method on an interface or displays a list of
methods
load
Initializes an ActiveX object from a file
move
Moves and/or resizes an ActiveX control in its parent
window
propedit
Asks the control to display its built-in property page
release
Releases an activex object
save
Serializes an ActiveX control object to a file
send
Displays a list of events
set
Sets a property on an interface
The creation commands, actxcontrol and actxserver, both return a
MATLAB activex object, which represents the default interface for the object
that was created. However, these objects may have other interfaces. It is
possible (and common) for interfaces to also be obtained by invoking a method
on, or getting a property from, an existing interface. The ActiveX get and
invoke methods automatically create and return new activex objects to
represent these additional interfaces.
When each interface is no longer needed, use the release method to release the
interface. When the entire control or server is no longer needed, use the delete
command to delete it. See the section, “Releasing Interfaces” on page 7-10, for
more details.
7-7
7
ActiveX and DDE Support
Writing Event Handlers
ActiveX events are invoked when a control wants to notify its container that
something of interest has occurred. For example, many controls trigger an
event when the user single-clicks on the control. In MATLAB, when a control
is created, you may optionally supply a callback (also known as an event
handler function) as the last argument to the actxcontrol command or a cell
array that contains the event handlers.
h = actxcontrol (progid [, position [, handle ...
[, callback | {event1 eventhandler1; event2 eventhandler2; …}]]])
The event handler function (callback) is called whenever the control triggers
any event. The event handler function must be an M-function that accepts a
variable number of arguments of the following form:
function event (varargin)
if (str2num(vararg{1}) == -600)
disp ('Click Event Fired');
end
All arguments passed to this function are MATLAB strings. The first argument
to the event handler is a string that represents the number of the event that
caused the event handler to be called. The remaining arguments are the values
passed by the control with the event. These values will vary with the particular
event and control being used. The list of events that control invocations and
their corresponding event numbers and parameters must be obtained from the
control’s documentation. In order to use events with MATLAB, you will need to
find out the numerical values that the control uses for each event so that you
can use these in the event handler.
The event handlers that are used in the cell array style are slightly different
than those used in the callback style. The first argument in the cell array style
is a reference to the object itself. The second argument is the event number,
which, unlike the callback style, is not a string that represents the number.
(Subsequent arguments vary depending on the event.)
These sample event handlers, myclick.m, my2click.m, and mymoused.m
correspond to the actxcontrol example.
7-8
Writing Event Handlers
function myclick(varargin)
disp('Single click function')
function my2click(varargin)
disp('Double click function')
function mymoused(varargin)
disp('You have reached the mouse down function')
disp('The X position is: ')
varargin(5)
disp('The Y position is: ')
varargin(6)
You can use the same event handler for all the events you want to monitor
using the cell array pairs. Response time will be better than using the callback
style.
For instance
h = actxcontrol('SELECTOR.SelectorCtrl.1', [0 0 200 200], f, ...
{'Click' 'allevents'; 'DblClick' 'allevents'; ...
'MouseDown' 'allevents'})
and allevents.m is
function allevents(varargin)
if (varargin{2} = -600)
disp ('Single Click Event Fired')
elseif (varargin{2} = -601)
disp ('Double Click Event Fired')
elseif (varargin{2} = -605)
disp ('Mousedown Event Fired')
end
Note MATLAB does not support event arguments passed by reference or
return values from events.
7-9
7
ActiveX and DDE Support
Additional ActiveX Client Information
Releasing Interfaces
Each ActiveX object can support one or more interfaces. In MATLAB, an
interface is represented by an instance of the activex class. There are three
ways to get a valid interface object into the MATLAB workspace:
• Return value from actxcontrol/actxserver
• Return value from a property via get
• Return value from a method invocation via invoke
In each case, once the interface is represented by an activex object in the
workspace, it must be released when you are finished using it. Failure to
release interface handles results in memory and resources being consumed.
Alternatively, you can use the delete command on any valid interface object,
and all interfaces for that object are automatically released, and thus
invalidated, and the ActiveX server or control itself is deleted.
MATLAB automatically releases all interfaces for an ActiveX control when the
figure window that contains that control is deleted or closed. MATLAB also
automatically releases all handles for an ActiveX automation server when
MATLAB is shut down.
Using ActiveX Collections
ActiveX collections are a way to support groups of related ActiveX objects that
can be iterated over. A collection is itself a special interface with a Count
property (read only), which contains the number of items in the collection, and
an Item method, which allows you to retrieve a single item from the collection.
The Item method is indexed, which means that it requires an argument that
specifies which item in the collection is being requested. The data type of the
index can be any data type that is appropriate for the particular collection and
is specific to the control or server that supports the collection. Although integer
indices are common, the index could just as easily be a string value. Often, the
return value from the Item method is itself an interface. Like all interfaces,
this interface should be released when you are finished with it.
This example iterates through the members of a collection. Each member of the
collection is itself an interface (called Plot and represented by a MATLAB
7-10
Additional ActiveX Client Information
activex object called hPlot.) In particular, this example iterates through a
collection of Plot interfaces, invokes the Redraw method for each interface, and
then releases each interface.
hCollection = get (hControl, 'Plots');
for i=1:get (hCollection, 'Count')
hPlot = invoke (hCollection, 'Item', i);
invoke (hPlot, 'Redraw');
release (hPlot);
end;
release (hCollection);
Converting Data
Since ActiveX defines a number of different data formats and types, you will
need to know how MATLAB converts data from activex objects into variables
in the MATLAB workspace. Data from activex objects must be converted:
• When a property value is retrieved
• When a value is returned from a method invocation
This chart shows how ActiveX data types are converted into variables in the
MATLAB workspace.
ActiveX Data Type
MATLAB Variable
String
File Time
Error
Decimal Date
MATLAB String
Currency
Hresult
Int/Unsigned (2, 4, 8)
Bool
Real (Single/Double
Precision)
Scalar Double
Null
NaN
7-11
7
ActiveX and DDE Support
ActiveX Data Type
MATLAB Variable
Array of
Currency
Hresult
Int/Unsigned (2, 4, 8)
Bool
Real (Single/Double
Precision)
Matrix of Double
Variant
Array of Variant
Cell Array
IDispatch *
activex Object
Empty
Unknown
Void
Ptr
Carray
Userdefined
Blob
Stream
Storage
Streamed Object
Stored Object
Blob Object
CF
Not Converted (error)
Using MATLAB As a DCOM Server Client
Distributed Component Object Model (DCOM) is an object distribution
mechanism that allows ActiveX clients to use remote ActiveX objects over a
network. At the time of this writing, DCOM is shipped with NT 4.0, and can be
obtained from Microsoft for Windows 95.
7-12
Additional ActiveX Client Information
MATLAB has been tested as a DCOM server with Windows NT 4.0 only.
Additionally, MATLAB can be used as a DCOM client with remote automation
servers if the operating system on which MATLAB is running is DCOM
enabled.
Note If you use MATLAB as a remote DCOM server, all MATLAB windows
will appear on the remote machine.
MATLAB ActiveX Support Limitations
The following is a list of limitations of MATLAB ActiveX support:
• MATLAB only supports indexed collections.
• ActiveX controls are not printed with figure windows.
• MATLAB does not support event arguments passed by reference.
• MATLAB does not support returning values from event handler functions.
• The position vector of a control cannot be changed or queried.
• MATLAB does not support method or event arguments passed by reference.
MATLAB Sample Control
MATLAB ships with a very simple example ActiveX control that draws a circle
on the screen, displays some text, and fires a clicked event when the user clicks
on the control. This control makes it easy to try out ActiveX control support
with a known control. The control can be created by running the mwsamp file in
the winfun directory.
The control is stored in the MATLAB bin, or executable, directory along with
the control’s type library. The type library is a binary file used by ActiveX tools
to decipher the control’s capabilities.
Using MATLAB As an Automation Client
This example uses MATLAB as an automation client and Microsoft Excel as
the server. It provides a good overview of typical functions. In addition, it is a
good example of using the automation interface of another application.
7-13
7
ActiveX and DDE Support
% MATLAB ActiveX automation client example
%
% Open Excel, add workbook, change active worksheet,
% get/put array, save.
% First, open an Excel Server.
Excel = actxserver('Excel.Application');
set(Excel, 'Visible', 1);
% Insert a new workbook.
Workbooks = Excel.Workbooks;
Workbook = invoke(Workbooks, 'Add');
% Make the second sheet active.
Sheets = Excel.ActiveWorkBook.Sheets;
sheet2 = get(Sheets, 'Item', 2);
invoke(sheet2, 'Activate');
% Get a handle to the active sheet.
Activesheet = Excel.Activesheet;
% Put a MATLAB array into Excel.
A = [1 2; 3 4];
ActivesheetRange = get(Activesheet,'Range','A1','B2');
set(ActivesheetRange, 'Value', A);
% Get back a range. It will be a cell array, since the cell range
% can contain different types of data.
Range = get(Activesheet, 'Range', 'A1', 'B2');
B = Range.value;
% Convert to a double matrix.
% scalars.
B = reshape([B{:}], size(B));
The cell array must contain only
% Now, save the workbook.
invoke(Workbook, 'SaveAs', 'myfile.xls');
% To avoid saving the workbook and being prompt to do so,
% uncomment the following code.
7-14
Additional ActiveX Client Information
% Workbook.Saved = 1;
% invoke(Workbook, 'Close');
% Quit Excel.
invoke(Excel, 'Quit');
7-15
7
ActiveX and DDE Support
MATLAB ActiveX Automation Server Support
MATLAB on Microsoft Windows supports ActiveX Automation server
capabilities. Automation is an ActiveX protocol that allows one application or
component (the controller) to control another application or component (the
server). Thus, MATLAB can be launched and controlled by any Windows
program that can be an Automation Controller. Some examples of applications
that can be Automation Controllers are Microsoft Excel, Microsoft Access,
Microsoft Project, and many Visual Basic and Visual C++ programs. Using
Automation, you can execute MATLAB commands, and get and put mxArrays
from and to the MATLAB workspace.
To use MATLAB as an automation server, follow steps 1 and 2:
1 Invoke an ActiveX Automation server.
Consult the documentation of your controller to find out how to invoke an
ActiveX Automation server. The name of the MATLAB ActiveX object that
is placed in the registry is Matlab.Application. Exactly how you invoke the
MATLAB server depends on which controller you choose, but all controllers
require this name to identify the server.
2 Use the Execute method on command strings.
The ActiveX Automation interface to MATLAB supports several methods.
Here is a Visual Basic code fragment that invokes the MATLAB Automation
Execute method, and that works in Microsoft Excel or any other Visual
Basic or Visual Basic for Applications (VBA)-enabled application. The
Execute method takes a command string as an argument and returns the
results as a string. The command string can be any command that would
normally be typed in the command window; the result contains any output
that would have been printed to the command window as a result of
executing the string, including errors.
Dim MatLab As Object
Dim Result As String
Set MatLab = CreateObject("Matlab.Application")
Result = MatLab.Execute("surf(peaks)")
7-16
MATLAB ActiveX Automation Server Support
Note The first statement above should be declared in the general
declarations section in order to keep the scope throughout the application.
MATLAB ActiveX Automation Methods
This section lists the methods that are supported by the MATLAB Automation
Server. The data types for the arguments and return values are expressed as
ActiveX Automation data types, which are language-independent types
defined by the ActiveX Automation protocol. For example, BSTR is a
wide-character string type defined as an Automation type, and is the same data
format used by Visual Basic to store strings. Any ActiveX-compliant controller
should support these data types, although the details of how you declare and
manipulate these are controller specific.
BSTR Execute([in] BSTR Command);
This command accepts a single string (Command), which contains any command
that can be typed at the MATLAB command window prompt. MATLAB will
execute the command and return the results as a string. Any figure windows
generated by the command are displayed on the screen as if the command were
executed directly from the command window or an M-file. A Visual Basic
example is
Dim MatLab As Object
Dim Result As String
Set MatLab = CreateObject("Matlab.Application")
Result = MatLab.Execute("surf(peaks)")
void GetFullMatrix(
[in] BSTR
[in] BSTR
[in, out]
[in, out]
Name,
Workspace,
SAFEARRAY(double)* pr,
SAFEARRAY(double)* pi);
Note The first statement above should be declared in the general
declarations section in order to keep the scope throughout the application.
7-17
7
ActiveX and DDE Support
This method retrieves a full, one- or two-dimensional real or imaginary
mxArray from the named workspace. The real and (optional) imaginary parts
are retrieved into separate arrays of doubles.
Name. Identifies the name of the mxArray to be retrieved.
Workspace. Identifies the workspace that contains the mxArray. Use the
workspace name “base” to retrieve an mxArray from the default MATLAB
workspace. Use the workspace name “global” to put the mxArray into the global
MATLAB workspace. The “caller” workspace does not have any context in the
API when used outside of MEX-files.
pr. Array of reals that is dimensioned to be the same size as the mxArray being
retrieved. On return, this array will contain the real values of the mxArray.
pi. Array of reals that is dimensioned to be the same size as the mxArray being
retrieved. On return, this array will contain the imaginary values of the
mxArray. If the requested mxArray is not complex, an empty array must be
passed. In Visual Basic, an empty array is declared as
Dim Mempty() as Double. A Visual Basic example of this method is
Dim MatLab As Object
Dim
Dim
Dim
Dim
Dim
Result As String
MReal(1, 3) As Double
MImag() As Double
RealValue As Double
i, j As Integer
rem We assume that the connection to MATLAB exists.
Result = MatLab.Execute("a = [1 2 3 4; 5 6 7 8;]")
Call MatLab.GetFullMatrix("a", "base", MReal, MImag)
For i = 0 To 1
For j = 0 To 3
RealValue = MReal(i, j)
Next j
Next i
7-18
MATLAB ActiveX Automation Server Support
void PutFullMatrix(
[in] BSTR Name,
[in] BSTR Workspace,
[in] SAFEARRAY(double) pr,
[in] SAFEARRAY(double) pi);
Note The first statement above should be declared in the general
declarations section in order to keep the scope throughout the application.
This method puts a full, one- or two-dimensional real or imaginary mxArray
into the named workspace. The real and (optional) imaginary parts are passed
in through separate arrays of doubles.
Name. Identifies the name of the mxArray to be placed.
Workspace. Identifies the workspace into which the mxArray should be placed.
Use the workspace name “base” to put the mxArray into the default MATLAB
workspace. Use the workspace name “global” to put the mxArray into the global
MATLAB workspace. The “caller” workspace does not have any context in the
API when used outside of MEX-files.
pr. Array of reals that contains the real values for the mxArray.
pi. Array of reals that contains the imaginary values for the mxArray. If the
mxArray that is being sent is not complex, an empty array must be passed for
this parameter. In Visual Basic, an empty array is declared as Dim Mempty()
as Double. A Visual Basic example of this method is
Dim MatLab As Object
Dim
Dim
Dim
For
MReal(1, 3) As Double
MImag() As Double
i, j As Integer
i = 0 To 1
For j = 0 To 3
MReal(i, j) = I * j;
Next j
Next I
rem We assume that the connection to MATLAB exists.
Call MatLab.PutFullMatrix("a", "base", MReal, MImag)
7-19
7
ActiveX and DDE Support
Note The first statement above should be declared in the general
declarations section in order to keep the scope throughout the application.
MATLAB ActiveX Automation Properties
You have the option of making your server application visible or not by setting
the Visible property. When visible, the server window appears on the desktop,
enabling the user to interact with the server application. This may be useful for
such purposes as debugging. When not visible, the server window does not
appear, thus perhaps making for a cleaner interface and also preventing any
interaction with the server application.
By default, the Visible property is enabled, or set to 1.
h = actxserver('Matlab.Application');
h.visible
ans =
1
You can change the setting of Visible by setting it to 0 (invisible) or 1 (visible).
The following command removes the server application window from the
desktop.
h.visible = 0;
h.visible
ans =
0
7-20
Additional ActiveX Server Information
Additional ActiveX Server Information
Launching the MATLAB ActiveX Server
For MATLAB to act as an automation server, it must be started with the /
Automation command line argument. Microsoft Windows does this
automatically when an ActiveX connection is established by a controller.
However, if MATLAB is already running and was launched without this
parameter, any request by an automation controller to connect to MATLAB as
a server will cause Windows to launch another instance of MATLAB with the
/Automation parameter. This protects controllers from interfering with any
interactive MATLAB sessions that may be running.
Specifying a Shared or Dedicated Server
You can launch the MATLAB Automation server in one of two modes — shared
or dedicated. A dedicated server is dedicated to a single client; a shared server
is shared by multiple clients.
The mode is determined by the Program ID (ProgID) used by the client to
launch MATLAB. The ProgID, Matlab.Application, (or the version-specific
ProgID, Matlab.Application.5) specifies the default mode, which is shared.
To specify a dedicated server, use the ProgID, Matlab.Application.Single,
(or the version-specific ProgID, Matlab.Application.Single.5). The term
Single refers to the number of clients that may connect to this server.
Once MATLAB is launched as a shared server, all clients that request a
connection to MATLAB by using the shared server ProgID will connect to the
already running instance of MATLAB. In other words, there is never more than
one instance of a shared server running, since it is shared by all clients that use
the ProgID that specifies a shared server.
Note Clients will not connect to an interactive instance of MATLAB, that is,
an instance of MATLAB that was launched manually and without the /
Automation command line flag.
Each client that requests a connection to MATLAB using a dedicated ProgID
will cause a separate instance of MATLAB to be launched, and that server will
not be shared with any other client. Therefore, there can be several instances
7-21
7
ActiveX and DDE Support
of a dedicated server running simultaneously, since the dedicated server is not
shared by multiple clients.
Using MATLAB As a DCOM Server
DCOM is a protocol that allows ActiveX connections to be established over a
network. If you are using a version of Windows that supports DCOM (Windows
NT 4.0 at the time of this writing) and a controller that supports DCOM, you
can use the controller to launch MATLAB on a remote machine. To do this,
DCOM must be configured properly, and MATLAB must be installed on each
machine that is used as a client or server. (Even though the client machine will
not be running MATLAB in such a configuration, the client machine must have
a MATLAB installation because certain MATLAB components are required to
establish the remote connection.) Consult the DCOM documentation for how to
configure DCOM for your environment.
7-22
Dynamic Data Exchange (DDE)
Dynamic Data Exchange (DDE)
MATLAB provides functions that enable MATLAB to access other Windows
applications and for other Windows applications to access MATLAB in a wide
range of contexts. These functions use dynamic data exchange (DDE), software
that allows Microsoft Windows applications to communicate with each other by
exchanging data.
This section describes using DDE in MATLAB:
• “DDE Concepts and Terminology”
• “Accessing MATLAB As a Server”
• “Using MATLAB As a Client”
• “DDE Advisory Links”
DDE Concepts and Terminology
Applications communicate with each other by establishing a DDE
conversation. The application that initiates the conversation is called the client.
The application that responds to the client application is called the server.
When a client application initiates a DDE conversation, it must identify two
DDE parameters that are defined by the server:
• The name of the application it intends to have the conversation with, called
the service name
• The subject of the conversation, called the topic
When a server application receives a request for a conversation involving a
supported topic, it acknowledges the request, establishing a DDE conversation.
The combination of a service and a topic identifies a conversation uniquely. The
service or topic cannot be changed for the duration of the conversation,
although the service can maintain more than one conversation.
During a DDE conversation, the client and server applications exchange data
concerning items. An item is a reference to data that is meaningful to both
applications in a conversation. Either application can change the item during
a conversation. These concepts are discussed in more detail below.
7-23
7
ActiveX and DDE Support
The Service Name
Every application that can be a DDE server has a unique service name. The
service name is usually the application’s executable filename without any
extension. Service names are not case sensitive. Here are some commonly used
service names:
• The service name for MATLAB is Matlab.
• The service name for Microsoft Word for Windows is WinWord.
• The service name for Microsoft Excel is Excel.
For the service names of other Windows applications, refer to the application
documentation.
The Topic
The topic defines the subject of a DDE conversation and is usually meaningful
to both the client and server applications. Topic names are not case sensitive.
MATLAB topics are System and Engine and are discussed in “Accessing
MATLAB As a Server” on page 7-25. Most applications support the System
topic and at least one other topic. Consult your application documentation for
information about supported topics.
The Item
Each topic supports one or more items. An item identifies the data being passed
during the DDE conversation. Case sensitivity of items depends on the
application. MATLAB Engine items are case sensitive if they refer to matrices
because matrix names are case sensitive.
Clipboard Formats
DDE uses the Windows clipboard formats for formatting data sent between
applications. As a client, MATLAB supports only Text format. As a server,
MATLAB supports Text, Metafilepict, and XLTable formats, described below:
• Text – Data in Text format is a buffer of characters terminated by the null
character. Lines of text in the buffer are delimited by a carriage return
line-feed combination. If the buffer contains columns of data, those columns
are delimited by the tab character. MATLAB supports Text format for
obtaining the results of a remote EvalString command and requests for
matrix data. Also, matrix data can be sent to MATLAB in Text format.
7-24
Dynamic Data Exchange (DDE)
• Metafilepict – Metafilepict format is a description of graphical data
containing the drawing commands for graphics. As a result, data stored in
this format is scalable and device independent. MATLAB supports
Metafilepict format for obtaining the result of a remote command that causes
some graphic action to occur.
• XLTable – XLTable format is the clipboard format used by Microsoft Excel
and is supported for ease and efficiency in exchanging data with Excel.
XLTable format is a binary buffer with a header that describes the data held
in the buffer. For a full description of XLTable format, consult the Microsoft
Excel SDK documentation.
Accessing MATLAB As a Server
A client application can access MATLAB as a DDE server in the following
ways, depending on the client application:
• If you are using an application that provides functions or macros to conduct
DDE conversations, you can use these functions or macros. For example,
Microsoft Excel, Word for Windows, and Visual Basic provide DDE functions
or macros. For more information about using these functions or macros, see
the appropriate Microsoft documentation.
• If you are creating your own application, you can use the MATLAB Engine
Library or DDE directly. For more information about using the Engine
Library, see “Using the MATLAB Engine” on page 4-3. For more information
about using DDE routines, see the Microsoft Windows Programmer’s Guide.
The figure below illustrates how MATLAB communicates as a server. DDE
functions in the client application communicate with MATLAB’s DDE server
module. The client’s DDE functions can be provided by either the application
or the MATLAB Engine Library.
MATLAB
DDE Server
Module
Client Application
Conversation
DDE Functions
7-25
7
ActiveX and DDE Support
The DDE Name Hierarchy
When you access MATLAB as a server, you must specify its service name, topic,
and item. The figure below illustrates the MATLAB DDE name hierarchy.
Topics and items are described in more detail below.
items
SysItems
topics
System
Format
Topics
service
MATLAB
EngEvalString
EngStringResult
Engine
EngFigureResult
<matrix name>
The two MATLAB topics are System and Engine.
MATLAB System Topic
The System topic allows users to browse the list of topics provided by the
server, the list of System topic items provided by the server, and the formats
supported by the server.
The MATLAB System topic supports these items:
• SysItems
Provides a tab-delimited list of items supported under the System topic (this
list).
7-26
Dynamic Data Exchange (DDE)
• Format
Provides a tab-delimited list of string names of all the formats supported by
the server. MATLAB supports Text, Metafilepict, and XLTable. These
formats are described in “Clipboard Formats” on page 7-24.
• Topics
Provides a tab-delimited list of the names of the topics supported by
MATLAB.
MATLAB Engine Topic
The Engine topic allows users to use MATLAB as a server by passing it a
command to execute, requesting data, or sending data.
The MATLAB Engine topic supports these items:
• EngEvalString
Specifies an item name, if required, when you send a command to MATLAB
for evaluation.
• EngStringResult
Provides the string result of a DDE execute command when you request data
from MATLAB.
• EngFigureResult
Provides the graphical result of a DDE execute command when you request
data from MATLAB.
• <matrix name>
When requesting data from MATLAB, this is the name of the matrix for
which data is being requested. When sending data to MATLAB, this is the
name of the matrix to be created or updated.
The MATLAB Engine topic supports three operations that may be used by
applications with a DDE client interface. These operations include sending
commands to MATLAB for evaluation, requesting data from MATLAB, and
sending data to MATLAB.
Sending Commands to MATLAB for Evaluation
Clients send commands to MATLAB using the DDE execute operation. The
Engine topic supports DDE execute in two forms because some clients require
that you specify the item name and the command to execute, while others
7-27
7
ActiveX and DDE Support
require only the command. Where an item name is required, use
EngEvalString. In both forms, the format of the command must be Text. Most
clients default to Text for DDE execute. If the format cannot be specified, it is
probably Text. The table summarizes the DDE execute parameters.
Item
Format
Command
EngEvalString
Text
String
null
Text
String
Requesting Data from MATLAB
Clients request data from MATLAB using the DDE request operation. The
Engine topic supports DDE requests for three functions:
Text that is the result of the previous DDE execute command. You
request the string result of a DDE execute command using the
EngStringResult item with Text format.
Graphical results of the previous DDE execute command. You request
the graphical result of a DDE execute command using the EngFigureResult
item. The EngFigureResult item can be used with Text or Metafilepict
formats.
• Specifying the Text format results in a string having a value of “yes” or “no.”
If the result is “yes,” the metafile for the current figure is placed on the
clipboard. This functionality is provided for DDE clients that can retrieve
only text from DDE requests, such as Word for Windows. If the result is “no,”
no metafile is placed on the clipboard.
• Specifying the Metafilepict format when there is a graphical result causes a
metafile to be returned directly from the DDE request.
The data for a specified matrix. You request the data for a matrix by
specifying the name of the matrix as the item. You can specify either the Text
or XLTable format.
7-28
Dynamic Data Exchange (DDE)
The table summarizes the DDE request parameters.
Item
Format
Result
EngStringResult
Text
String
EngFigureResult
Text
Yes/No
EngFigureResult
Metafilepict
Metafile of the current figure
<matrix name>
Text
Character buffer, tab-delimited
columns, CR/LF-delimited rows
<matrix name>
XLTable
Binary data in a format compatible
with Microsoft Excel
Sending Data to MATLAB
Clients send data to MATLAB using the DDE poke operation. The Engine topic
supports DDE poke for updating or creating new matrices in the MATLAB
workspace. The item specified is the name of the matrix to be updated or
created. If a matrix with the specified name already exists in the workspace it
will be updated; otherwise it will be created. The matrix data can be in Text or
XLTable format.
The table summarizes the DDE poke parameters.
Item
Format
Poke Data
<matrix name>
Text
Character buffer, tab-delimited
columns, CR/LF-delimited rows
<matrix name>
XLTable
Binary data in a format compatible
with Microsoft Excel
Example: Using Visual Basic and the MATLAB DDE
Server
This example shows a Visual Basic form that contains two text edit controls,
TextInput and TextOutput. This code is the TextInput_KeyPress method.
7-29
7
ActiveX and DDE Support
Sub TextInput_KeyPress(KeyAscii As Integer)
rem If the user presses the return key
rem in the TextInput control.
If KeyAscii = vbKeyReturn then
rem Initiate the conversation between the TextInput
rem control and MATLAB under the Engine topic.
rem Set the item to EngEvalString.
TextInput.LinkMode = vbLinkNone
TextInput.LinkTopic = "MATLAB|Engine"
TextInput.LinkItem = "EngEvalString"
TextInput.LinkMode = vbLinkManual
rem Get the current string in the TextInput control.
rem This text is the command string to send to MATLAB.
szCommand = TextInput.Text
rem Perform DDE Execute with the command string.
TextInput.LinkExecute szCommand
TextInput.LinkMode = vbLinkNone
rem Initiate the conversation between the TextOutput
rem control and MATLAB under the Engine topic.
rem Set the item to EngStringResult.
TextOutput.LinkMode = vbLinkNone
TextOutput.LinkTopic = "MATLAB|Engine"
TextOutput.LinkItem = "EngStringResult"
TextOutput.LinkMode = vbLinkManual
rem Request the string result of the previous EngEvalString
rem command. The string ends up in the text field of the
rem control TextOutput.text.
TextOutput.LinkRequest
TextOutput.LinkMode = vbLinkNone
End If
End Sub
7-30
Dynamic Data Exchange (DDE)
Using MATLAB As a Client
For MATLAB to act as a client application, you can use the MATLAB DDE
client functions to establish and maintain conversations.
This figure illustrates how MATLAB communicates as a client to a server
application.
Server Application
MATLAB
DDE Client
Module
Conversation
DDE Server
Module
MATLAB’s DDE client module includes a set of functions. This table describes
the functions that enable you to use MATLAB as a client.
Function
Description
ddeadv
Sets up advisory link between MATLAB and DDE server
application.
ddeexec
Sends execution string to DDE server application.
ddeinit
Initiates DDE conversation between MATLAB and another
application.
ddepoke
Sends data from MATLAB to DDE server application.
ddereq
Requests data from DDE server application.
ddeterm
Terminates DDE conversation between MATLAB and server
application.
ddeunadv
Releases advisory link between MATLAB and DDE server
application.
If the server application is Microsoft Excel, you can specify the System topic or
a topic that is a filename. If you specify the latter, the filename ends in .XLS or
7-31
7
ActiveX and DDE Support
.XLC and includes the full path if necessary. A Microsoft Excel item is a cell
reference, which can be an individual cell or a range of cells.
Microsoft Word for Windows topics are System and document names that are
stored in files whose names end in .DOC or .DOT. A Word for Windows item is
any bookmark in the document specified by the topic.
The following example is an M-file that establishes a DDE conversation with
Microsoft Excel, and then passes a 20-by-20 matrix of data to Excel.
% Initialize conversation with Excel.
chan = ddeinit('excel', 'Sheet1');
%
h
%
z
Create a surface of peaks plot.
= surf(peaks(20));
Get the z data of the surface
= get(h, 'zdata');
% Set range of cells in Excel for poking.
range = 'r1c1:r20c20';
% Poke the z data to the Excel spread sheet.
rc = ddepoke(chan, range, z);
DDE Advisory Links
You can use DDE to notify a client application when data at a server has
changed. For example, if you use MATLAB to analyze data entered in an Excel
spreadsheet, you can establish a link that causes Excel to notify MATLAB
when this data changes. You can also establish a link that automatically
updates a matrix with the new or modified spreadsheet data.
MATLAB supports two kinds of advisory links, distinguished by the way in
which the server application advises MATLAB when the data that is the
subject of the item changes at the server:
• A hot link causes the server to supply the data to MATLAB when the data
defined by the item changes.
• A warm link causes the server to notify MATLAB when the data changes but
supplies the data only when MATLAB requests it.
7-32
Dynamic Data Exchange (DDE)
You set up and release advisory links with the ddeadv and ddeunadv functions.
MATLAB only supports links when MATLAB is a client.
This example establishes a DDE conversation between MATLAB, acting as a
client, and Microsoft Excel. The example extends the example in the previous
section by creating a hot link with Excel. The link updates matrix z and
evaluates a callback when the range of cells changes. A push-button, user
interface control terminates the advisory link and the DDE conversation when
pressed. (For more information about creating a graphical user interface, see
the online MATLAB manual, Creating Graphical User Interfaces.)
% Initialize conversation with Excel.
chan = ddeinit('excel', 'Sheet1');
% Set range of cells in Excel for poking.
range = 'r1c1:r20c20';
% Create a surface of peaks plot.
h = surf(peaks(20));
% Get the z data of the surface.
z = get(h, 'zdata');
% Poke the z data to the Excel spread sheet.
rc = ddepoke(chan, range, z);
% Set up a hot link ADVISE loop with Excel
% and the MATLAB matrix 'z'.
% The callback sets the zdata and cdata for
% the surface h to be the new data sent from Excel.
rc = ddeadv(chan, range,...
'set(h,''zdata'',z);set(h,''cdata'',z);','z');
%
%
%
c
Create a push button that will end the ADVISE link,
terminate the DDE conversation,
and close the figure window.
= uicontrol('String','&Close','Position',[5 5 80 30],...
'Callback',...
'rc = ddeunadv(chan,range);ddeterm(chan);close;');
7-33
7
ActiveX and DDE Support
7-34
8
Serial Port I/O
Introduction . . . . . . . . . . . . . . . . . . . . 8-2
Overview of the Serial Port . . . . . . . . . . . . . 8-4
Getting Started with Serial I/O
Creating a Serial Port Object
. . . . . . . . . . . 8-18
. . . . . . . . . . . . 8-24
Connecting to the Device . . . . . . . . . . . . . . 8-27
Configuring Communication Settings
. . . . . . . . 8-28
Writing and Reading Data . . . . . . . . . . . . . . 8-29
Events and Callbacks . . . . . . . . . . . . . . . . 8-48
Using Control Pins . . . . . . . . . . . . . . . . . 8-56
Debugging: Recording Information to Disk . . . . . . 8-62
Saving and Loading
. . . . . . . . . . . . . . . . 8-68
Disconnecting and Cleaning Up . . . . . . . . . . . 8-69
Property Reference . . . . . . . . . . . . . . . . . 8-70
8
Serial Port I/O
Introduction
What Is MATLAB’s Serial Port Interface?
MATLAB’s serial port interface provides direct access to peripheral devices
such as modems, printers, and scientific instruments that you connect to your
computer’s serial port. This interface is established through a serial port object.
The serial port object supports functions and properties that allow you to:
• Configure serial port communications
• Use serial port control pins
• Write and read data
• Use events and callbacks
• Record information to disk
If you want to communicate with PC-compatible data acquisition hardware
such as multifunction I/O boards, you need the Data Acquisition Toolbox. If you
want to communicate with GPIB- or VISA-compatible instruments, you need
the Instrument Control Toolbox.
For more information about these products, visit the MathWorks Web site at
http://www.mathworks.com/products.
Supported Serial Port Interface Standards
Over the years, several serial port interface standards have been developed.
These standards include RS-232, RS-422, and RS-485 – all of which are
supported by MATLAB’s serial port object. Of these, the most widely used
interface standard for connecting computers to peripheral devices is RS-232.
In this guide, it is assumed you are using the RS-232 standard, which is
discussed in “Overview of the Serial Port” on page 8-4. Refer to your computer
and device documentation to see which interface standard you can use.
Supported Platforms
MATLAB’s serial port interface is supported for Microsoft Windows 95,
Windows 98, Windows 2000, Windows NT, Linux, and Sun Solaris platforms.
8-2
Using the Examples with Your Device
Many of the examples in this section reflect specific peripheral devices
connected to a PC serial port – in particular a Tektronix TDS 210 two-channel
oscilloscope connected to the COM1 port. Therefore, many of the string
commands are specific to this instrument.
If your peripheral device is connected to a different serial port, or if it accepts
different commands, you should modify the examples accordingly.
8-3
8
Serial Port I/O
Overview of the Serial Port
This section provides an overview of the serial port. Topics include:
• What is Serial Communication?
• The Serial Port Interface Standard
• Connecting Two Devices with a Serial Cable
• Serial Port Signals and Pin Assignments
• Serial Data Format
• Finding Serial Port Information for Your Platform
For many serial port applications, you can communicate with your device
without detailed knowledge of how the serial port works. If your application is
straightforward, or if you are already familiar with the topics mentioned above,
you may want to begin with “The Serial Port Session” on page 8-19 to see how
to use your serial port device with MATLAB.
What Is Serial Communication?
Serial communication is the most common low-level protocol for
communicating between two or more devices. Normally, one device is a
computer, while the other device can be a modem, a printer, another computer,
or a scientific instrument such as an oscilloscope or a function generator.
As the name suggests, the serial port sends and receives bytes of information
in a serial fashion – one bit at a time. These bytes are transmitted using either
a binary (numerical) format or a text format.
The Serial Port Interface Standard
The serial port interface for connecting two devices is specified by the TIA/
EIA-232C standard published by the Telecommunications Industry
Association.
The original serial port interface standard was given by RS-232, which stands
for Recommended Standard number 232. The term “RS-232” is still in popular
use, and is used in this guide when referring to a serial communication port
that follows the TIA/EIA-232 standard. RS-232 defines these serial port
characteristics:
• The maximum bit transfer rate and cable length
8-4
Overview of the Serial Port
• The names, electrical characteristics, and functions of signals
• The mechanical connections and pin assignments
Primary communication is accomplished using three pins: the Transmit Data
pin, the Receive Data pin, and the Ground pin. Other pins are available for
data flow control, but are not required.
Other standards such as RS-485 define additional functionality such as higher
bit transfer rates, longer cable lengths, and connections to as many as 256
devices.
Connecting Two Devices with a Serial Cable
The RS-232 standard defines the two devices connected with a serial cable as
the Data Terminal Equipment (DTE) and Data Circuit-Terminating
Equipment (DCE). This terminology reflects the RS-232 origin as a standard
for communication between a computer terminal and a modem.
Throughout this guide, your computer is considered a DTE, while peripheral
devices such as modems and printers are considered DCE’s. Note that many
scientific instruments function as DTE’s.
Since RS-232 mainly involves connecting a DTE to a DCE, the pin assignments
are defined such that straight-through cabling is used, where pin 1 is connected
to pin 1, pin 2 is connected to pin 2, and so on. A DTE to DCE serial connection
using the transmit data (TD) pin and the receive data (RD) pin is shown below.
Refer to “Serial Port Signals and Pin Assignments” on page 8-6 for more
information about serial port pins.
Computer
Device
TD
RD
Pin 3
Pin 3
Pin 2
Pin 2
DCE
DTE
RD
TD
8-5
8
Serial Port I/O
If you connect two DTE’s or two DCE’s using a straight serial cable, then the
TD pin on each device are connected to each other, and the RD pin on each
device are connected to each other. Therefore, to connect two like devices, you
must use a null modem cable. As shown below, null modem cables cross the
transmit and receive lines in the cable.
Computer
Computer
TD
TD
Pin 3
Pin 3
Pin 2
Pin 2
DTE
DTE
RD
RD
Note You can connect multiple RS-422 or RS-485 devices to a serial port. If
you have an RS-232/RS-485 adaptor, then you can use MATLAB’s serial port
object with these devices.
Serial Port Signals and Pin Assignments
Serial ports consist of two signal types: data signals and control signals. To
support these signal types, as well as the signal ground, the RS-232 standard
defines a 25-pin connection. However, most PC’s and UNIX platforms use a
9-pin connection. In fact, only three pins are required for serial port
communications: one for receiving data, one for transmitting data, and one for
the signal ground.
8-6
Overview of the Serial Port
The pin assignment scheme for a 9-pin male connector on a DTE is given below.
1
2
6
3
7
4
8
5
9
The pins and signals associated with the 9-pin connector are described below.
Refer to the RS-232 standard for a description of the signals and pin
assignments used for a 25-pin connector.
Table 8-1: Serial Port Pin and Signal Assignments
Pin
Label
Signal Name
Signal Type
1
CD
Carrier Detect
Control
2
RD
Received Data
Data
3
TD
Transmitted Data
Data
4
DTR
Data Terminal Ready
Control
5
GND
Signal Ground
Ground
6
DSR
Data Set Ready
Control
7
RTS
Request to Send
Control
8
CTS
Clear to Send
Control
9
RI
Ring Indicator
Control
The term “data set” is synonymous with “modem” or “device,” while the term
“data terminal” is synonymous with “computer.”
8-7
8
Serial Port I/O
Note The serial port pin and signal assignments are with respect to the DTE.
For example, data is transmitted from the TD pin of the DTE to the RD pin of
the DCE.
Signal States
Signals can be in either an active state or an inactive state. An active state
corresponds to the binary value 1, while an inactive state corresponds to the
binary value 0. An active signal state is often described as logic 1, on, true, or
a mark. An inactive signal state is often described as logic 0, off, false, or a
space.
For data signals, the “on” state occurs when the received signal voltage is more
negative than -3 volts, while the “off” state occurs for voltages more positive
than 3 volts. For control signals, the “on” state occurs when the received signal
voltage is more positive than 3 volts, while the “off” state occurs for voltages
more negative than -3 volts. The voltage between -3 volts and +3 volts is
considered a transition region, and the signal state is undefined.
To bring the signal to the “on” state, the controlling device unasserts (or lowers)
the value for data pins and asserts (or raises) the value for control pins.
Conversely, to bring the signal to the “off” state, the controlling device asserts
the value for data pins and unasserts the value for control pins.
8-8
Overview of the Serial Port
The “on” and “off” states for a data signal and for a control signal are shown
below.
Data Signal
Control Signal
Signal value (Volts)
6
off
on
3
0
-3
on
off
-6
The Data Pins
Most serial port devices support full-duplex communication meaning that they
can send and receive data at the same time. Therefore, separate pins are used
for transmitting and receiving data. For these devices, the TD, RD, and GND
pins are used. However, some types of serial port devices support only one-way
or half-duplex communications. For these devices, only the TD and GND pins
are used. In this guide, it is assumed that a full-duplex serial port is connected
to your device.
The TD pin carries data transmitted by a DTE to a DCE. The RD pin carries
data that is received by a DTE from a DCE.
The Control Pins
9-pin serial ports provide several control pins that:
• Signal the presence of connected devices
• Control the flow of data
The control pins include RTS and CTS, DTR and DSR, CD, and RI.
8-9
8
Serial Port I/O
The RTS and CTS Pins. The RTS and CTS pins are used to signal whether the
devices are ready to send or receive data. This type of data flow control – called
hardware handshaking – is used to prevent data loss during transmission.
When enabled for both the DTE and DCE, hardware handshaking using RTS
and CTS follows these steps:
1 The DTE asserts the RTS pin to instruct the DCE that it is ready to receive
data.
2 The DCE asserts the CTS pin indicating that it is clear to send data over the
TD pin. If data can no longer be sent, the CTS pin is unasserted.
3 The data is transmitted to the DTE over the TD pin. If data can no longer be
accepted, the RTS pin is unasserted by the DTE and the data transmission
is stopped.
To enable hardware handshaking in MATLAB, refer to “Controlling the Flow
of Data: Handshaking” on page 8-59.
The DTR and DSR Pins. Many devices use the DSR and DTR pins to signal if they
are connected and powered. Signaling the presence of connected devices using
DTR and DSR follows these steps:
1 The DTE asserts the DTR pin to request that the DCE connect to the
communication line.
2 The DCE asserts the DSR pin to indicate it’s connected.
3 DCE unasserts the DSR pin when it’s disconnected from the communication
line.
The DTR and DSR pins were originally designed to provide an alternative
method of hardware handshaking. However, the RTS and CTS pins are usually
used in this way, and not the DSR and DTR pins. However, you should refer to
your device documentation to determine its specific pin behavior.
The CD and RI Pins. The CD and RI pins are typically used to indicate the
presence of certain signals during modem-modem connections.
CD is used by a modem to signal that it has made a connection with another
modem, or has detected a carrier tone. CD is asserted when the DCE is
8-10
Overview of the Serial Port
receiving a signal of a suitable frequency. CD is unasserted if the DCE is not
receiving a suitable signal.
RI is used to indicate the presence of an audible ringing signal. RI is asserted
when the DCE is receiving a ringing signal. RI is unasserted when the DCE is
not receiving a ringing signal (for example, it’s between rings).
Serial Data Format
The serial data format includes one start bit, between five and eight data bits,
and one stop bit. A parity bit and an additional stop bit may be included in the
format as well. The diagram below illustrates the serial data format.
Start bit
Data bits
Parity bit
Stop bits
The format for serial port data is often expressed using the following notation
number of data bits - parity type - number of stop bits
For example, 8-N-1 is interpreted as eight data bits, no parity bit, and one stop
bit, while 7-E-2 is interpreted as seven data bits, even parity, and two stop bits.
The data bits are often referred to as a character since these bits usually
represent an ASCII character. The remaining bits are called framing bits since
they frame the data bits.
Bytes Versus Values
The collection of bits that comprise the serial data format is called a byte. At
first, this term may seem inaccurate since a byte is 8 bits and the serial data
format can range between 7 bits and 12 bits. However, when serial data is
stored on your computer, the framing bits are stripped away, and only the data
bits are retained. Moreover, eight data bits are always used regardless of the
number of data bits specified for transmission, with the unused bits assigned a
value of 0.
8-11
8
Serial Port I/O
When reading or writing data, you may need to specify a value, which can
consist of one or more bytes. For example, if you read one value from a device
using the int32 format, then that value consists of four bytes. For more
information about reading and writing values, refer to “Writing and Reading
Data” on page 8-29.
Synchronous and Asynchronous Communication
The RS-232 standard supports two types of communication protocols:
synchronous and asynchronous.
Using the synchronous protocol, all transmitted bits are synchronized to a
common clock signal. The two devices initially synchronize themselves to each
other, and then continually send characters to stay synchronized. Even when
actual data is not really being sent, a constant flow of bits allows each device
to know where the other is at any given time. That is, each bit that is sent is
either actual data or an idle character. Synchronous communications allows
faster data transfer rates than asynchronous methods, because additional bits
to mark the beginning and end of each data byte are not required.
Using the asynchronous protocol, each device uses its own internal clock
resulting in bytes that are transferred at arbitrary times. So, instead of using
time as a way to synchronize the bits, the data format is used.
In particular, the data transmission is synchronized using the start bit of the
word, while one or more stop bits indicate the end of the word. The requirement
to send these additional bits causes asynchronous communications to be
slightly slower than synchronous. However, it has the advantage that the
processor does not have to deal with the additional idle characters. Most serial
ports operate asynchronously.
Note When used in this guide, the terms “synchronous” and “asynchronous”
refer to whether read or write operations block access to the MATLAB
command line. Refer to “Controlling Access to the MATLAB Command Line”
on page 8-29 for more information.
How Are the Bits Transmitted?
By definition, serial data is transmitted one bit at a time. The order in which
the bits are transmitted is given below:
8-12
Overview of the Serial Port
1 The start bit is transmitted with a value of 0.
2 The data bits are transmitted. The first data bit corresponds to the least
significant bit (LSB), while the last data bit corresponds to the most
significant bit (MSB).
3 The parity bit (if defined) is transmitted.
4 One or two stop bits are transmitted, each with a value of 1.
The number of bits transferred per second is given by the baud rate. The
transferred bits include the start bit, the data bits, the parity bit (if defined),
and the stop bits.
Start and Stop Bits
As described in “Synchronous and Asynchronous Communication” on page
8-12, most serial ports operate asynchronously. This means that the
transmitted byte must be identified by start and stop bits. The start bit
indicates when the data byte is about to begin and the stop bit(s) indicates
when the data byte has been transferred. The process of identifying bytes with
the serial data format follows these steps:
1 When a serial port pin is idle (not transmitting data), then it is in an “on”
state.
2 When data is about to be transmitted, the serial port pin switches to an “off”
state due to the start bit.
3 The serial port pin switches back to an “on” state due to the stop bit(s). This
indicates the end of the byte.
Data Bits
The data bits transferred through a serial port may represent device
commands, sensor readings, error messages, and so on. The data can be
transferred as either binary data or ASCII data.
Most serial ports use between five and eight data bits. Binary data is typically
transmitted as eight bits. Text-based data is transmitted as either seven bits
or eight bits. If the data is based on the ASCII character set, then a minimum
of seven bits is required since there are 27 or 128 distinct characters. If an
8-13
8
Serial Port I/O
eighth bit is used, it must have a value of 0. If the data is based on the extended
ASCII character set, then eight bits must be used since there are 28 or 256
distinct characters.
The Parity Bit
The parity bit provides simple error (parity) checking for the transmitted data.
The types of parity checking are given below.
Table 8-2: Parity Types
Parity Type
Description
Even
The data bits plus the parity bit result in an even number
of 1’s.
Mark
The parity bit is always 1.
Odd
The data bits plus the parity bit result in an odd number
of 1’s.
Space
The parity bit is always 0.
Mark and space parity checking are seldom used since they offer minimal error
detection. You may choose to not use parity checking at all.
The parity checking process follows these steps:
1 The transmitting device sets the parity bit to 0 or to 1 depending on the data
bit values and the type of parity checking selected.
2 The receiving device checks if the parity bit is consistent with the
transmitted data. If it is, then the data bits are accepted. If it is not, then an
error is returned.
Note Parity checking can detect only 1-bit errors. Multiple-bit errors can
appear as valid data.
For example, suppose the data bits 01110001 are transmitted to your
computer. If even parity is selected, then the parity bit is set to 0 by the
8-14
Overview of the Serial Port
transmitting device to produce an even number of 1’s. If odd parity is selected,
then the parity bit is set to 1 by the transmitting device to produce an odd
number of 1’s.
Finding Serial Port Information for Your Platform
The ways to find serial port information for Windows and UNIX platforms are
described below.
Note Your operating system provides default values for all serial port
settings. However, these settings are overridden by your MATLAB code, and
will have no effect on your serial port application.
Windows Platform
You can easily access serial port information through the Windows Control
Panel. You can invoke the Control Panel with the Start button (Start ->
Settings -> Control Panel).
For Windows NT, you access the serial ports by selecting the Ports icon within
the Control Panel. The resulting Ports dialog box is shown below.
8-15
8
Serial Port I/O
To obtain information on the possible settings for COM1, select this port under
the Ports list box and then select Settings.
You can access serial port information for the Windows 95, Windows 98, and
Windows 2000 operating systems with the System Properties dialog box,
which is available through the Control Panel.
UNIX Platform
To find serial port information for UNIX platforms, you need to know the serial
port names. These names may vary between different operating systems.
On Linux, serial port devices are typically named ttyS0, ttyS1, and so on. You
can use the setserial command to display or configure serial port information.
For example, to display which ports are available
setserial -bg /dev/ttyS*
/dev/ttyS0 at 0x03f8 (irq = 4) is a 16550A
/dev/ttyS1 at 0x02f8 (irq = 3) is a 16550A
To display detailed information about ttyS0
setserial -ag /dev/ttyS0
/dev/ttyS0, Line 0, UART: 16550A, Port: 0x03f8, IRQ: 4
Baud_base: 115200, close_delay: 50, divisor: 0
closing_wait: 3000, closing_wait2: infinte
Flags: spd_normal skip_test session_lockout
8-16
Overview of the Serial Port
Note If the setserial -ag command does not work, make sure that you
have read and write permission for the port.
For all supported UNIX platforms, you can use the stty command to display
or configure serial port information. For example, to display serial port
properties for ttyS0
stty -a < /dev/ttyS0
To configure the baud rate to 4800 bits per second
stty speed 4800 < /dev/ttyS0 > /dev/ttyS0
Selected Bibliography
1 TIA/EIA-232-F, Interface Between Data Terminal Equipment and Data
Circuit-Terminating Equipment Employing Serial Binary Data Interchange.
2 Jan Axelson, Serial Port Complete, Lakeview Research, Madison, WI, 1998.
3 Instrument Communication Handbook, IOTech, Inc., Cleveland, OH, 1991.
4 TDS 200-Series Two Channel Digital Oscilloscope Programmer Manual,
Tektronix, Inc., Wilsonville, OR.
5 Courier High Speed Modems User’s Manual, U.S. Robotics, Inc., Skokie, IL,
1994.
6 Getting Started with Your AT Serial Hardware and Software for Windows
98/95, National Instruments, Inc., Austin, TX, 1998.
8-17
8
Serial Port I/O
Getting Started with Serial I/O
To get you started with MATLAB’s serial port interface, this section provides
the following information:
• “Example: Getting Started” illustrates some basic serial port commands.
• “The Serial Port Session” describes the steps you use to perform any serial
port task from beginning to end.
• “Configuring and Returning Properties” describes how you display serial
port property names and property values, and how you assign values to
properties.
Example: Getting Started
If you have a device connected to the serial port COM1 and configured for a
baud rate of 4800, you can execute the following complete example.
s = serial('COM1');
set(s,'BaudRate',4800);
fopen(s);
fprintf(s,'*IDN?')
out = fscanf(s);
fclose(s)
delete(s)
clear s
The *IDN? command queries the device for identification information, which is
returned to out. If your device does not support this command, or if it is
connected to a different serial port, you should modify the above example
accordingly.
Note *IDN? is one of the commands supported by the Standard Commands
for Programmable Instruments (SCPI) language, which is used by many
modern devices. Refer to your device documentation to see if it supports the
SCPI language.
8-18
Getting Started with Serial I/O
The Serial Port Session
The serial port session comprises all the steps you are likely to take when
communicating with a device connected to a serial port. These steps are:
1 Create a serial port object – You create a serial port object for a specific
serial port using the serial creation function.
You can also configure properties during object creation. In particular, you
may want to configure properties associated with serial port
communications such as the baud rate, the number of data bits, and so on.
2 Connect to the device – You connect the serial port object to the device
using the fopen function.
After the object is connected, you can alter device settings by configuring
property values, read data, and write data.
3 Configure properties – To establish the desired serial port object behavior,
you assign values to properties using the set function or dot notation.
In practice, you can configure many of the properties at any time including
during, or just after, object creation. Conversely, depending on your device
settings and the requirements of your serial port application, you may be
able to accept the default property values and skip this step.
4 Write and read data – You can now write data to the device using the
fprintf or fwrite function, and read data from the device using the fgetl,
fgets, fread, fscanf, or readasync function.
The serial port object behaves according to the previously configured or
default property values.
5 Disconnect and clean up – When you no longer need the serial port object,
you should disconnect it from the device using the fclose function, remove
it from memory using the delete function, and remove it from the MATLAB
workspace using the clear command.
The serial port session is reinforced in many of the serial port documentation
examples. Refer to “Example: Getting Started” on page 8-18 to see a basic
example that uses the steps shown above.
8-19
8
Serial Port I/O
Configuring and Returning Properties
You establish the desired serial port object behavior by configuring property
values. You can display or configure property values using the set function, the
get function, or dot notation.
Displaying Property Names and Property Values
Once the serial port object is created, you can use the set function to display
all the configurable properties to the command line. Additionally, if a property
has a finite set of string values, then set also displays these values.
s = serial('COM1');
set(s)
ByteOrder: [ {littleEndian} | bigEndian ]
BytesAvailableFcn
BytesAvailableFcnCount
BytesAvailableFcnMode: [ {terminator} | byte ]
ErrorFcn
InputBufferSize
Name
OutputBufferSize
OutputEmptyFcn
RecordDetail: [ {compact} | verbose ]
RecordMode: [ {overwrite} | append | index ]
RecordName
Tag
Timeout
TimerFcn
TimerPeriod
UserData
SERIAL specific properties:
BaudRate
BreakInterruptFcn
DataBits
DataTerminalReady: [ {on} | off ]
FlowControl: [ {none} | hardware | software ]
Parity: [ {none} | odd | even | mark | space ]
PinStatusFcn
Port
ReadAsyncMode: [ {continuous} | manual ]
8-20
Getting Started with Serial I/O
RequestToSend: [ {on} | off ]
StopBits
Terminator
You can use the get function to display one or more properties and their
current values to the command line. To display all properties and their current
values
get(s)
ByteOrder = littleEndian
BytesAvailable = 0
BytesAvailableFcn =
BytesAvailableFcnCount = 48
BytesAvailableFcnMode = terminator
BytesToOutput = 0
ErrorFcn =
InputBufferSize = 512
Name = Serial-COM1
OutputBufferSize = 512
OutputEmptyFcn =
RecordDetail = compact
RecordMode = overwrite
RecordName = record.txt
RecordStatus = off
Status = closed
Tag =
Timeout = 10
TimerFcn =
TimerPeriod = 1
TransferStatus = idle
Type = serial
UserData = []
ValuesReceived = 0
ValuesSent = 0
SERIAL specific properties:
BaudRate = 9600
BreakInterruptFcn =
DataBits = 8
DataTerminalReady = on
FlowControl = none
8-21
8
Serial Port I/O
Parity = none
PinStatus = [1x1 struct]
PinStatusFcn =
Port = COM1
ReadAsyncMode = continuous
RequestToSend = on
StopBits = 1
Terminator = LF
To display the current value for one property, you supply the property name to
get.
get(s,'OutputBufferSize')
ans =
512
To display the current values for multiple properties, you must include the
property names as elements of a cell array.
get(s,{'Parity','TransferStatus'})
ans =
'none'
'idle'
You can also use the dot notation to display a single property value.
s.Parity
ans =
none
Configuring Property Values
You can configure property values using the set function
set(s,'BaudRate',4800);
or the dot notation.
s.BaudRate = 4800;
To configure values for multiple properties, you can supply multiple property
name/property value pairs to set.
set(s,'DataBits',7,'Name','Test1-serial')
8-22
Getting Started with Serial I/O
Note that you can configure only one property value at a time using the dot
notation.
In practice, you can configure many of the properties at any time while the
serial port object exists – including during object creation. However, some
properties are not configurable while the object is connected to the device or
when recording information to disk. Refer to “Property Reference” on page 8-70
for information about when a property is configurable.
Specifying Property Names
Serial port property names are presented using mixed case. While this makes
property names easier to read, you can use any case you want when specifying
property names. Additionally, you need use only enough letters to identify the
property name uniquely, so you can abbreviate most property names. For
example, you can configure the BaudRate property any of these ways.
set(s,'BaudRate',4800)
set(s,'baudrate',4800)
set(s,'BAUD',4800)
When you include property names in an M-file, you should use the full property
name. This practice can prevent problems with future releases of MATLAB if
a shortened name is no longer unique because of the addition of new properties.
Default Property Values
Whenever you do not explicitly define a value for a property, then the default
value is used. All configurable properties have default values.
Note Your operating system provides default values for all serial port
settings such as the baud rate. However, these settings are overridden by your
MATLAB code, and will have no effect on your serial port application.
If a property has a finite set of string values, then the default value is enclosed
by {}. For example, the default value for the Parity property is none.
set(s,'Parity')
[ {none} | odd | even | mark | space ]
You can find the default value for any property in the property reference pages.
8-23
8
Serial Port I/O
Creating a Serial Port Object
You create a serial port object with the serial function. serial requires the
name of the serial port connected to your device as an input argument.
Additionally, you can configure property values during object creation. For
example, to create a serial port object associated with the serial port COM1
s = serial('COM1');
The serial port object s now exists in the MATLAB workspace. You can display
the class of s with the whos command.
whos s
Name
s
Size
1x1
Bytes
512
Class
serial object
Grand total is 11 elements using 512 bytes
Once the serial port object is created, the properties listed below are
automatically assigned values. These general purpose properties provide
descriptive information about the serial port object based on the object type and
the serial port.
Table 8-3: Descriptive General Purpose Properties
Property Name
Description
Name
Specify a descriptive name for the serial port object
Port
Indicate the platform-specific serial port name
Type
Indicate the object type
You can display the values of these properties for s with the get function.
get(s,{'Name','Port','Type'})
ans =
'Serial-COM1'
'COM1'
8-24
'serial'
Creating a Serial Port Object
Configuring Properties During Object Creation
You can configure serial port properties during object creation. serial accepts
property names and property values in the same format as the set function.
For example, you can specify property name/property value pairs.
s = serial('COM1','BaudRate',4800,'Parity','even');
If you specify an invalid property name, the object is not created. However, if
you specify an invalid value for some properties (for example, BaudRate is set
to 50), the object may be created but you will not be informed of the invalid
value until you connect the object to the device with the fopen function.
The Serial Port Object Display
The serial port object provides you with a convenient display that summarizes
important configuration and state information. You can invoke the display
summary these three ways:
• Type the serial port object variable name at the command line.
• Exclude the semicolon when creating a serial port object.
• Exclude the semicolon when configuring properties using the dot notation.
The display summary for the serial port object s is given below.
Serial Port Object : Serial-COM1
Communication Settings
Port:
COM1
BaudRate:
9600
Terminator:
'LF'
Communication State
Status:
RecordStatus:
closed
off
Read/Write State
TransferStatus:
BytesAvailable:
ValuesReceived:
ValuesSent:
idle
0
0
0
8-25
8
Serial Port I/O
Creating an Array of Serial Port Objects
In MATLAB, you can create an array from existing variables by concatenating
those variables together. The same is true for serial port objects. For example,
suppose you create the serial port objects s1 and s2
s1 = serial('COM1');
s2 = serial('COM2');
You can now create a serial port object array consisting of s1 and s2 using the
usual MATLAB syntax. To create the row array x
x = [s1 s2]
Instrument Object Array
Index:
1
2
Type:
serial
serial
Status:
closed
closed
Name:
Serial-COM1
Serial-COM2
To create the column array y
y = [s1;s2];
Note that you cannot create a matrix of serial port objects. For example, you
cannot create the matrix
z = [s1 s2;s1 s2];
??? Error using ==> serial/vertcat
Only a row or column vector of instrument objects can be created.
Depending on your application, you may want to pass an array of serial port
objects to a function. For example, to configure the baud rate and parity for s1
and s2 using one call to set
set(x,'BaudRate',19200,'Parity','even')
Refer to the serial port function reference to see which functions accept a serial
port object array as an input.
8-26
Connecting to the Device
Connecting to the Device
Before you can use the serial port object to write or read data, you must connect
it to your device via the serial port specified in the serial function. You connect
a serial port object to the device with the fopen function.
fopen(s)
Some properties are read-only while the serial port object is connected and
must be configured before using fopen. Examples include the
InputBufferSize and the OutputBufferSize properties. Refer to “Property
Reference” on page 8-70 to determine when you can configure a property.
Note You can create any number of serial port objects. However, you can
connect only one serial port object to a given serial port at a time.
You can examine the Status property to verify that the serial port object is
connected to the device.
s.Status
ans =
open
As illustrated below, the connection between the serial port object and the
device is complete, and you can write and read data.
MATLAB
Serial Port I/O Hardware
s=serial('COM1');
fopen(s)
COM1
Instrument
01.00
8-27
8
Serial Port I/O
Configuring Communication Settings
Before you can write or read data, both the serial port object and the device
must have identical communication settings. Configuring serial port
communications involves specifying values for properties that control the baud
rate and the serial data format. These properties are given below.
Table 8-4: Communication Properties
Property Name
Description
BaudRate
Specify the rate at which bits are transmitted
DataBits
Specify the number of data bits to transmit
Parity
Specify the type of parity checking
StopBits
Specify the number of bits used to indicate the end of a
byte
Terminator
Specify the terminator character
Note If the serial port object and the device communication settings are not
identical, then you cannot successfully read or write data.
Refer to your device documentation for an explanation of its supported
communication settings.
8-28
Writing and Reading Data
Writing and Reading Data
For many serial port applications, there are three important questions that you
should consider when writing or reading data:
• Will the read or write function block access to the MATLAB command line?
• Is the data to be transferred binary (numerical) or text?
• Under what conditions will the read or write operation complete?
For write operations, these questions are answered in “Writing Data” on page
8-31. For read operations, these questions are answered in “Reading Data” on
page 8-36.
Example: Introduction to Writing and Reading Data
Suppose you want to return identification information for a Tektronix TDS 210
two-channel oscilloscope connected to the serial port COM1. This requires
writing the *IDN? command to the instrument using the fprintf function, and
then reading back the result of that command using the fscanf function.
s = serial('COM1');
fopen(s)
fprintf(s,'*IDN?')
out = fscanf(s)
The resulting identification information is shown below.
out =
TEKTRONIX,TDS 210,0,CF:91.1CT FV:v1.16 TDS2CM:CMV:v1.04
End the serial port session.
fclose(s)
delete(s)
clear s
Controlling Access to the MATLAB Command Line
You control access to the MATLAB command line by specifying whether a read
or write operation is synchronous or asynchronous.
A synchronous operation blocks access to the command line until the read or
write function completes execution. An asynchronous operation does not block
8-29
8
Serial Port I/O
access to the command line, and you can issue additional commands while the
read or write function executes in the background.
The terms “synchronous” and “asynchronous” are often used to describe how
the serial port operates at the hardware level. The RS-232 standard supports
an asynchronous communication protocol. Using this protocol, each device uses
its own internal clock. The data transmission is synchronized using the start
bit of the bytes, while one or more stop bits indicate the end of the byte. Refer
to “Serial Data Format” on page 8-11 for more information on start bits and
stop bits. The RS-232 standard also supports a synchronous mode where all
transmitted bits are synchronized to a common clock signal.
At the hardware level, most serial ports operate asynchronously. However,
using the default behavior for many of the read and write functions, you can
mimic the operation of a synchronous serial port.
Note When used in this guide, the terms “synchronous” and “asynchronous”
refer to whether read or write operations block access to the MATLAB
command line. In other words, these terms describe how the software behaves,
and not how the hardware behaves.
The two main advantages of writing or reading data asynchronously are:
• You can issue another command while the write or read function is
executing.
• You can use all supported callback properties (see “Events and Callbacks” on
page 8-48).
8-30
Writing and Reading Data
For example, since serial ports have separate read and write pins, you can
simultaneously read and write data. This is illustrated below.
MATLAB
Serial Port I/O Hardware
Write
s=serial('COM1');
fopen(s)
Instrument
Write
COM1
Read
Read
Writing Data
This section describes writing data to your serial port device in three parts:
• “The Output Buffer and Data Flow” describes the flow of data from MATLAB
to the device.
• “Writing Text Data” describes how to write text data (string commands) to
the device.
• “Writing Binary Data” describes how to write binary (numerical) data to the
device.
The functions associated with writing data are given below.
Table 8-5: Functions Associated with Writing Data
Function Name
Description
fprintf
Write text to the device
fwrite
Write binary data to the device
stopasync
Stop asynchronous read and write operations
8-31
8
Serial Port I/O
The properties associated with writing data are given below.
Table 8-6: Properties Associated with Writing Data
Property Name
Description
BytesToOutput
Indicate the number of bytes currently in the output
buffer
OutputBufferSize
Specify the size of the output buffer in bytes
Timeout
Specify the waiting time to complete a read or write
operation
TransferStatus
Indicate if an asynchronous read or write operation is
in progress
ValuesSent
Indicate the total number of values written to the
device
The Output Buffer and Data Flow
The output buffer is computer memory allocated by the serial port object to
store data that is to be written to the device. When writing data to your device,
the data flow follows these two steps:
1 The data specified by the write function is sent to the output buffer.
2 The data in the output buffer is sent to the device.
The OutputBufferSize property specifies the maximum number of bytes that
you can store in the output buffer. The BytesToOutput property indicates the
number of bytes currently in the output buffer. The default values for these
properties are given below.
s = serial('COM1');
get(s,{'OutputBufferSize','BytesToOutput'})
ans =
[512]
[0]
If you attempt to write more data than can fit in the output buffer, an error is
returned and no data is written.
8-32
Writing and Reading Data
For example, suppose you write the string command *IDN? to the TDS 210
oscilloscope using the fprintf function. As shown below, the string is first
written to the output buffer as six values.
MATLAB
s=serial('COM1');
fopen(s)
fprintf(s,'*IDN?')
Output Buffer
*IDN?
...
6 values
6 bytes
Bytes used during write
Bytes unused during write
The *IDN? command consists of six values since the terminator is
automatically written. Moreover, the default data format for the fprintf
function specifies that one value corresponds to one byte. For more information
about bytes and values, refer to “Bytes Versus Values” on page 8-11. fprintf
and the terminator are discussed in “Writing Text Data” on page 8-34.
8-33
8
Serial Port I/O
As shown below, after the string is written to the output buffer, it is then
written to the device via the serial port.
Output Buffer
Serial Port I/O Hardware
Instrument
*IDN?
...
COM1
6 values
6 bytes
Bytes used during write
Bytes unused during write
Writing Text Data
You use the fprintf function to write text data to the device. For many devices,
writing text data means writing string commands that change device settings,
prepare the device to return data or status information, and so on.
For example, the Display:Contrast command changes the display contrast of
the oscilloscope.
s = serial('COM1');
fopen(s)
fprintf(s,'Display:Contrast 45')
By default, fprintf writes data using the %s\n format since many serial port
devices accept only text-based commands. However, you can specify many
other formats as described in the fprintf reference pages.
You can verify the number of values sent to the device with the ValuesSent
property.
s.ValuesSent
ans =
20
8-34
Writing and Reading Data
Note that the ValuesSent property value includes the terminator since each
occurrence of \n in the command sent to the device is replaced with the
Terminator property value.
s.Terminator
ans =
LF
The default value of Terminator is the line feed character. The terminator
required by your device will be described in its documentation.
Synchronous Versus Asynchronous Write Operations. By default, fprintf operates
synchronously and will block the MATLAB command line until execution
completes. To write text data asynchronously to the device, you must specify
async as the last input argument to fprintf.
fprintf(s,'Display:Contrast 45','async')
Asynchronous operations do not block access to the MATLAB command line.
Additionally, while an asynchronous write operation is in progress, you can:
• Execute an asynchronous read operation since serial ports have separate
pins for reading and writing
• Make use of all supported callback properties
You can determine which asynchronous operations are in progress with the
TransferStatus property. If no asynchronous operations are in progress, then
TransferStatus is idle.
s.TransferStatus
ans =
idle
Rules for Completing a Write Operation with fprintf. A synchronous or asynchronous
write operation using fprintf completes when:
• The specified data is written.
• The time specified by the Timeout property passes.
Additionally, you can stop an asynchronous write operation with the
stopasync function.
8-35
8
Serial Port I/O
Writing Binary Data
You use the fwrite function to write binary data to the device. Writing binary
data means writing numerical values. A typical application for writing binary
data involves writing calibration data to an instrument such as an arbitrary
waveform generator.
Note Some serial port devices accept only text-based commands. These
commands may use the SCPI language or some other vendor-specific
language. Therefore, you may need to use the fprintf function for all write
operations.
By default, fwrite translates values using the uchar precision. However, you
can specify many other precisions as described in the reference pages for this
function.
By default, fwrite operates synchronously. To write binary data
asynchronously to the device, you must specify async as the last input
argument to fwrite. For more information about synchronous and
asynchronous write operations, refer to the “Writing Text Data” on page 8-34.
For a description of the rules used by fwrite to complete a write operation,
refer to its reference pages.
Reading Data
This section describes reading data from your serial port device in three parts:
• “The Input Buffer and Data Flow” describes the flow of data from the device
to MATLAB.
• “Reading Text Data” describes how to read from the device, and format the
data as text.
• “Reading Binary Data” describes how to read binary (numerical) data from
the device.
8-36
Writing and Reading Data
The functions associated with reading data are given below.
Table 8-7: Functions Associated with Reading Data
Function Name
Description
fgetl
Read one line of text from the device and discard the
terminator
fgets
Read one line of text from the device and include the
terminator
fread
Read binary data from the device
fscanf
Read data from the device, and format as text
readasync
Read data asynchronously from the device
stopasync
Stop asynchronous read and write operations
The properties associated with reading data are given below.
Table 8-8: Properties Associated with Reading Data
Property Name
Description
BytesAvailable
Indicate the number of bytes available in the input
buffer
InputBufferSize
Specify the size of the input buffer in bytes
ReadAsyncMode
Specify whether an asynchronous read operation is
continuous or manual
Timeout
Specify the waiting time to complete a read or write
operation
TransferStatus
Indicate if an asynchronous read or write operation
is in progress
ValuesReceived
Indicate the total number of values read from the
device
8-37
8
Serial Port I/O
The Input Buffer and Data Flow
The input buffer is computer memory allocated by the serial port object to store
data that is to be read from the device. When reading data from your device,
the data flow follows these two steps:
1 The data read from the device is stored in the input buffer.
2 The data in the input buffer is returned to the MATLAB variable specified
by the read function.
The InputBufferSize property specifies the maximum number of bytes that
you can store in the input buffer. The BytesAvailable property indicates the
number of bytes currently available to be read from the input buffer. The
default values for these properties are given below.
s = serial('COM1');
get(s,{'InputBufferSize','BytesAvailable'})
ans =
[512]
[0]
If you attempt to read more data than can fit in the input buffer, an error is
returned and no data is read.
For example, suppose you use the fscanf function to read the text-based
response of the *IDN? command previously written to the TDS 210 oscilloscope.
As shown below, the text data is first read into to the input buffer via the serial
port.
Instrument
Serial Port I/O Hardware
Input Buffer
data
COM1
Bytes used during read
Bytes unused during read
8-38
...
Writing and Reading Data
Note that for a given read operation, you may not know the number of bytes
returned by the device. Therefore, you may need to preset the
InputBufferSize property to a sufficiently large value before connecting the
serial port object.
As shown below, after the data is stored in the input buffer, it is then
transferred to the output variable specified by fscanf.
Input Buffer
MATLAB
data
...
out=fscanf(s)
Bytes used during read
Bytes unused during read
Reading Text Data
You use the fgetl, fgets, and fscanf functions to read data from the device,
and format the data as text.
For example, suppose you want to return identification information for the
oscilloscope. This requires writing the *IDN? command to the instrument, and
then reading back the result of that command.
s = serial('COM1');
fopen(s)
fprintf(s,'*IDN?')
out = fscanf(s)
out =
TEKTRONIX,TDS 210,0,CF:91.1CT FV:v1.16 TDS2CM:CMV:v1.04
By default, fscanf reads data using the %c format since the data returned by
many serial port devices is text based. However, you can specify many other
formats as described in the fscanf reference pages.
8-39
8
Serial Port I/O
You can verify the number of values read from the device – including the
terminator – with the ValuesReceived property.
s.ValuesReceived
ans =
56
Synchronous Versus Asynchronous Read Operations. You specify whether read
operations are synchronous or asynchronous with the ReadAsyncMode property.
You can configure ReadAsyncMode to continuous or manual.
If ReadAsyncMode is continuous (the default value), the serial port object
continuously queries the device to determine if data is available to be read. If
data is available, it is asynchronously stored in the input buffer. To transfer the
data from the input buffer to MATLAB, you use one of the synchronous
(blocking) read functions such as fgetl or fscanf. If data is available in the
input buffer, these functions will return quickly.
s.ReadAsyncMode = 'continuous';
fprintf(s,'*IDN?')
s.BytesAvailable
ans =
56
out = fscanf(s);
If ReadAsyncMode is manual, the serial port object does not continuously query
the device to determine if data is available to be read. To read data
asynchronously, you use the readasync function.You then use one of the
synchronous read functions to transfer data from the input buffer to MATLAB.
s.ReadAsyncMode = 'manual';
fprintf(s,'*IDN?')
s.BytesAvailable
ans =
0
readasync(s)
s.BytesAvailable
ans =
56
out = fscanf(s);
8-40
Writing and Reading Data
Asynchronous operations do not block access to the MATLAB command line.
Additionally, while an asynchronous read operation is in progress, you can:
• Execute an asynchronous write operation since serial ports have separate
pins for reading and writing
• Make use of all supported callback properties
You can determine which asynchronous operations are in progress with the
TransferStatus property. If no asynchronous operations are in progress, then
TransferStatus is idle.
s.TransferStatus
ans =
idle
Rules for Completing a Read Operation with fscanf. A read operation with fscanf
blocks access to the MATLAB command line until:
• The terminator specified by the Terminator property is read.
• The time specified by the Timeout property passes.
• The specified number of values specified is read.
• The input buffer is filled.
Reading Binary Data
You use the fread function to read binary data from the device. Reading binary
data means that you return numerical values to MATLAB.
For example, suppose you want to return the cursor and display settings for the
oscilloscope. This requires writing the CURSOR? and DISPLAY? commands to the
instrument, and then reading back the results of those commands.
s = serial('COM1');
fopen(s)
fprintf(s,'CURSOR?')
fprintf(s,'DISPLAY?')
Since the default value for the ReadAsyncMode property is continuous, data is
asynchronously returned to the input buffer as soon as it is available from the
device. You can verify the number of values read with the BytesAvailable
property.
8-41
8
Serial Port I/O
s.BytesAvailable
ans =
69
You can return the data to MATLAB using any of the synchronous read
functions. However, if you use fgetl, fgets, or fscanf, then you must issue the
function twice since there are two terminators stored in the input buffer. If you
use fread, then you can return all the data to MATLAB in one function call.
out = fread(s,69);
By default, fread returns numerical values in double precision arrays.
However, you can specify many other precisions as described in the fread
reference pages. You can convert the numerical data to text using MATLAB’s
char function.
val = char(out)'
val =
HBARS;CH1;SECONDS;-1.0E-3;1.0E-3;VOLTS;-6.56E-1;6.24E-1
YT;DOTS;0;45
For more information about synchronous and asynchronous read operations,
refer to “Reading Text Data” on page 8-39. For a description of the rules used
by fread to complete a read operation, refer to its reference pages.
Example: Writing and Reading Text Data
This example illustrates how to communicate with a serial port instrument by
writing and reading text data.
The instrument is a Tektronix TDS 210 two-channel oscilloscope connected to
the serial port COM1. Therefore, many of the commands given below are
specific to this instrument. A sine wave is input into channel 2 of the
oscilloscope, and your job is to measure the peak-to-peak voltage of the input
signal.
1. Create a serial port object – Create the serial port object s associated with
serial port COM1.
s = serial('COM1');
2. Connect to the device – Connect s to the oscilloscope. Since the default
value for the ReadAsyncMode property is continuous, data is asynchronously
returned to the input buffer as soon as it is available from the instrument.
8-42
Writing and Reading Data
fopen(s)
3. Write and read data – Write the *IDN? command to the instrument using
fprintf, and then read back the result of the command using fscanf.
fprintf(s,'*IDN?')
idn = fscanf(s)
idn =
TEKTRONIX,TDS 210,0,CF:91.1CT FV:v1.16 TDS2CM:CMV:v1.04
You need to determine the measurement source. Possible measurement
sources include channel 1 and channel 2 of the oscilloscope.
fprintf(s,'MEASUREMENT:IMMED:SOURCE?')
source = fscanf(s)
source =
CH1
The scope is configured to return a measurement from channel 1. Since the
input signal is connected to channel 2, you must configure the instrument to
return a measurement from this channel.
fprintf(s,'MEASUREMENT:IMMED:SOURCE CH2')
fprintf(s,'MEASUREMENT:IMMED:SOURCE?')
source = fscanf(s)
source =
CH2
You can now configure the scope to return the peak-to-peak voltage, and then
request the value of this measurement.
fprintf(s,'MEASUREMENT:MEAS1:TYPE PK2PK')
fprintf(s,'MEASUREMENT:MEAS1:VALUE?')
Transfer data from the input buffer to MATLAB using fscanf.
ptop = fscanf(s,'%g')
ptop =
2.0199999809E0
4. Disconnect and clean up – When you no longer need s, you should
disconnect it from the instrument, and remove it from memory and from the
MATLAB workspace.
fclose(s)
8-43
8
Serial Port I/O
delete(s)
clear s
Example: Parsing Input Data Using strread
This example illustrates how to use the strread function to parse and format
data that you read from a device. strread is particularly useful when you want
to parse a string into one or more variables, where each variable has its own
specified format.
The instrument is a Tektronix TDS 210 two-channel oscilloscope connected to
the serial port COM1.
1. Create a serial port object – Create the serial port object s associated with
serial port COM1.
s = serial('COM1');
2. Connect to the device – Connect s to the oscilloscope. Since the default
value for the ReadAsyncMode property is continuous, data is asynchronously
returned to the input buffer as soon as it is available from the instrument.
fopen(s)
3. Write and read data – Write the RS232? command to the instrument using
fprintf, and then read back the result of the command using fscanf. RS232?
queries the RS-232 settings and returns the baud rate, the software flow
control setting, the hardware flow control setting, the parity type, and the
terminator.
fprintf(s,'RS232?')
data = fscanf(s)
data =
9600;0;0;NONE;LF
Use the strread function to parse and format the data variable into five new
variables.
[br,sfc,hfc,par,tm] = strread(data,'%d%d%d%s%s','delimiter',';')
br =
9600
sfc =
0
8-44
Writing and Reading Data
hfc =
0
par =
'NONE'
tm =
'LF'
4. Disconnect and clean up – When you no longer need s, you should
disconnect it from the instrument, and remove it from memory and from the
MATLAB workspace.
fclose(s)
delete(s)
clear s
Example: Reading Binary Data
This example illustrates how you can download the TDS 210 oscilloscope
screen display to MATLAB. The screen display data is transferred and saved
to disk using the Windows bitmap format. This data provides a permanent
record of your work, and is an easy way to document important signal and
scope parameters.
Since the amount of data transferred is expected to be fairly large, it is
asynchronously returned to the input buffer as soon as it is available from the
instrument. This allows you to perform other tasks as the transfer progresses.
Additionally, the scope is configured to its highest baud rate of 19,200.
1. Create a serial port object – Create the serial port object s associated with
serial port COM1.
s = serial('COM1');
2. Configure property values – Configure the input buffer to accept a
reasonably large number of bytes, and configure the baud rate to the highest
value supported by the scope.
s.InputBufferSize = 50000;
s.BaudRate = 19200;
3. Connect to the device – Connect s to the oscilloscope. Since the default
value for the ReadAsyncMode property is continuous, data is asynchronously
returned to the input buffer as soon as it is available from the instrument.
8-45
8
Serial Port I/O
fopen(s)
4. Write and read data – Configure the scope to transfer the screen display as
a bitmap.
fprintf(s,'HARDCOPY:PORT RS232')
fprintf(s,'HARDCOPY:FORMAT BMP')
fprintf(s,'HARDCOPY START')
Wait until all the data is sent to the input buffer, and then transfer the data to
the MATLAB workspace as unsigned 8-bit integers.
out = fread(s,s.BytesAvailable,'uint8');
5. Disconnect and clean up – When you no longer need s, you should
disconnect it from the instrument, and remove it from memory and from the
MATLAB workspace.
fclose(s)
delete(s)
clear s
Viewing the Bitmap Data
To view the bitmap data, you should follow these steps:
1 Open a disk file.
2 Write the data to the disk file.
3 Close the disk file.
4 Read the data into MATLAB using the imread function.
5 Scale and display the data using the imagesc function.
Note that the file I/O versions of the fopen, fwrite, and fclose functions are
used.
fid = fopen('test1.bmp','w');
fwrite(fid,out,'uint8');
fclose(fid)
a = imread('test1.bmp','bmp');
imagesc(a)
8-46
Writing and Reading Data
Since the scope returns the screen display data using only two colors, an
appropriate colormap is selected.
mymap = [0 0 0; 1 1 1];
colormap(mymap)
The resulting bitmap image is shown below.
8-47
8
Serial Port I/O
Events and Callbacks
You can enhance the power and flexibility of your serial port application by
using events. An event occurs after a condition is met and may result in one or
more callbacks.
While the serial port object is connected to the device, you can use events to
display a message, display data, analyze data, and so on. Callbacks are
controlled through callback properties and callback functions. All event types
have an associated callback property. Callback functions are M-file functions
that you construct to suit your specific application needs.
You execute a callback when a particular event occurs by specifying the name
of the M-file callback function as the value for the associated callback property.
Example: Introduction to Events and Callbacks
This example uses the M-file callback function instrcallback to display a
message to the command line when a bytes-available event occurs. The event
is generated when the terminator is read.
s = serial('COM1');
fopen(s)
s.BytesAvailableFcnMode = 'terminator';
s.BytesAvailableFcn = @instrcallback;
fprintf(s,'*IDN?')
out = fscanf(s);
The resulting display from instrcallback is shown below.
BytesAvailable event occurred at 17:01:29 for the object:
Serial-COM1.
End the serial port session.
fclose(s)
delete(s)
clear s
You can use the type command to display instrcallback at the command line.
8-48
Events and Callbacks
Event Types and Callback Properties
The serial port event types and associated callback properties are described
below.
Table 8-9: Event Types and Callback Properties
Event Type
Associated Properties
Break interrupt
BreakInterruptFcn
Bytes available
BytesAvailableFcn
BytesAvailableFcnCount
BytesAvailableFcnMode
Error
ErrorFcn
Output empty
OutputEmptyFcn
Pin status
PinStatusFcn
Timer
TimerFcn
TimerPeriod
Break-Interrupt Event. A break-interrupt event is generated immediately after a
break interrupt is generated by the serial port. The serial port generates a
break interrupt when the received data has been in an inactive state longer
than the transmission time for one character.
This event executes the callback function specified for the BreakInterruptFcn
property. It can be generated for both synchronous and asynchronous read and
write operations.
Bytes-Available Event. A bytes-available event is generated immediately after a
predetermined number of bytes are available in the input buffer or a
terminator is read, as determined by the BytesAvailableFcnMode property.
If BytesAvailableFcnMode is byte, the bytes-available event executes the
callback function specified for the BytesAvailableFcn property every time the
number of bytes specified by BytesAvailableFcnCount is stored in the input
8-49
8
Serial Port I/O
buffer. If BytesAvailableFcnMode is terminator, then the callback function
executes every time the character specified by the Terminator property is read.
This event can be generated only during an asynchronous read operation.
Error Event. An error event is generated immediately after an error occurs.
This event executes the callback function specified for the ErrorFcn property.
It can be generated only during an asynchronous read or write operation.
An error event is generated when a timeout occurs. A timeout occurs if a read
or write operation does not successfully complete within the time specified by
the Timeout property. An error event is not generated for configuration errors
such as setting an invalid property value.
Output-Empty Event. An output-empty event is generated immediately after the
output buffer is empty.
This event executes the callback function specified for the OutputEmptyFcn
property. It can be generated only during an asynchronous write operation.
Pin Status Event. A pin status event is generated immediately after the state (pin
value) changes for the CD, CTS, DSR, or RI pins. Refer to “Serial Port Signals
and Pin Assignments” on page 8-6 for a description of these pins.
This event executes the callback function specified for the PinStatusFcn
property. It can be generated for both synchronous and asynchronous read and
write operations.
Timer Event. A timer event is generated when the time specified by the
TimerPeriod property passes. Time is measured relative to when the serial
port object is connected to the device.
This event executes the callback function specified for the TimerFcn property.
Note that some timer events may not be processed if your system is
significantly slowed or if the TimerPeriod value is too small.
Storing Event Information
You can store event information in an callback function or in a record file.
Event information is stored in a callback function using two fields: Type and
Data. The Type field contains the event type, while the Data field contains
event-specific information. As described in “Creating and Executing Callback
8-50
Events and Callbacks
Functions” on page 8-52, these two fields are associated with a structure that
you define in the callback function header. Refer to “Debugging: Recording
Information to Disk” on page 8-62 to learn about recording data and event
information to a record file.
The event types and the values for the Type and Data fields are given below.
Table 8-10: Event Information
Event Type
Field
Field Value
Break interrupt
Type
BreakInterrupt
Data.AbsTime
day-month-year hour:minute:second
Type
BytesAvailable
Data.AbsTime
day-month-year hour:minute:second
Type
Error
Data.AbsTime
day-month-year hour:minute:second
Data.Message
An error string
Type
OutputEmpty
Data.AbsTime
day-month-year hour:minute:second
Type
PinStatus
Data.AbsTime
day-month-year hour:minute:second
Data.Pin
CarrierDetect, ClearToSend,
DataSetReady, or RingIndicator
Data.PinValue
on or off
Type
Timer
Data.AbsTime
day-month-year hour:minute:second
Bytes available
Error
Output empty
Pin status
Timer
The Data field values are described below.
8-51
8
Serial Port I/O
The AbsTime Field. AbsTime is defined for all events, and indicates the absolute
time the event occurred. The absolute time is returned using the clock format.
day-month-year hour:minute:second
The Pin Field. Pin is used by the pin status event to indicate if the CD, CTS, DSR,
or RI pins changed state. Refer to “Serial Port Signals and Pin Assignments”
on page 8-6 for a description of these pins.
The PinValue Field. PinValue is used by the pin status event to indicate the state
of the CD, CTS, DSR, or RI pins. Possible values are on or off.
The Message Field. Message is used by the error event to store the descriptive
message that is generated when an error occurs.
Creating and Executing Callback Functions
You can specify the callback function to be executed when a specific event type
occurs by including the name of the M-file as the value for the associated
callback property. You can specify the callback function as a function handle or
as a string cell array element. Function handles are described in the
function_handle reference pages.
For example, to execute the callback function mycallback every time the
terminator is read from your device
s.BytesAvailableFcnMode = 'terminator';
s.BytesAvailableFcn = @mycallback;
Alternatively, you can specify the callback function as a cell array.
s.BytesAvailableFcn = {'mycallback'};
M-file callback functions require at least two input arguments. The first
argument is the serial port object. The second argument is a variable that
captures the event information given in Table 8-10, Event Information, on page
8-51. This event information pertains only to the event that caused the callback
function to execute. The function header for mycallback is shown below.
function mycallback(obj,event)
8-52
Events and Callbacks
You pass additional parameters to the callback function by including both the
callback function and the parameters as elements of a cell array. For example,
to pass the MATLAB variable time to mycallback
time = datestr(now,0);
s.BytesAvailableFcnMode = 'terminator';
s.BytesAvailableFcn = {@mycallback,time};
Alternatively, you can specify the callback function as a string in the cell array.
s.BytesAvailableFcn = {'mycallback',time};
The corresponding function header is
function mycallback(obj,event,time)
If you pass additional parameters to the callback function, then they must be
included in the function header after the two required arguments.
Note You can also specify the callback function as a string. In this case, the
callback is evaluated in the MATLAB workspace and no requirements are
made on the input arguments of the callback function.
Enabling Callback Functions After They Error
If an error occurs while an callback function is executing, then:
• The callback function is automatically disabled.
• A warning is displayed at the command line, indicating that the callback
function is disabled.
If you want to enable the same callback function, then you must disconnect the
object with the fclose function. If you want to use a different callback function,
then the callback will be enabled when you configure the callback property to
the new value.
8-53
8
Serial Port I/O
Example: Using Events and Callbacks
This example uses the M-file callback function instrcallback to display
event-related information to the command line when a bytes-available event or
an output-empty event occurs.
1. Create a serial port object – Create the serial port object s associated with
serial port COM1.
s = serial('COM1');
2. Connect to the device – Connect s to the Tektronix TDS 210 oscilloscope.
Since the default value for the ReadAsyncMode property is continuous, data is
asynchronously returned to the input buffer as soon as it is available from the
instrument.
fopen(s)
3. Configure properties – Configure s to execute the callback function
instrcallback when a bytes-available event or an output-empty event occurs.
Since instrcallback requires the serial port object and event information to
be passed as input arguments, the callback function is specified as a function
handle.
s.BytesAvailableFcnMode = 'terminator';
s.BytesAvailableFcn = @instrcallback;
s.OutputEmptyFcn = @instrcallback;
4. Write and read data – Write the RS232? command asynchronously to the
oscilloscope. This command queries the RS-232 settings and returns the baud
rate, the software flow control setting, the hardware flow control setting, the
parity type, and the terminator.
fprintf(s,'RS232?','async')
instrcallback is called after the RS232? command is sent, and when the
terminator is read. The resulting displays are shown below.
OutputEmpty event occurred at 17:37:21 for the object:
Serial-COM1.
BytesAvailable event occurred at 17:37:21 for the object:
Serial-COM1.
8-54
Events and Callbacks
Read the data from the input buffer.
out = fscanf(s)
out =
9600;0;0;NONE;LF
5. Disconnect and clean up – When you no longer need s, you should
disconnect it from the instrument, and remove it from memory and from the
MATLAB workspace.
fclose(s)
delete(s)
clear s
8-55
8
Serial Port I/O
Using Control Pins
As described in “Serial Port Signals and Pin Assignments” on page 8-6, 9-pin
serial ports include six control pins. These control pins allow you to:
• Signal the presence of connected devices
• Control the flow of data
The properties associated with the serial port control pins are given below.
Table 8-11: Control Pin Properties
Property Name
Description
DataTerminalReady
Specify the state of the DTR pin
FlowControl
Specify the data flow control method to use
PinStatus
Indicate the state of the CD, CTS, DSR, and RI pins
RequestToSend
Specify the state of the RTS pin
Signaling the Presence of Connected Devices
DTE’s and DCE’s often use the CD, DSR, RI, and DTR pins to indicate whether
a connection is established between serial port devices. Once the connection is
established, you can begin to write or read data.
You can monitor the state of the CD, DSR, and RI pins with the PinStatus
property. You can specify or monitor the state of the DTR pin with the
DataTerminalReady property.
The following example illustrates how these pins are used when two modems
are connected to each other.
Example: Connecting Two Modems
This example connects two modems to each other via the same computer, and
illustrates how you can monitor the communication status for the
computer-modem connections, and for the modem-modem connection. The first
modem is connected to COM1, while the second modem is connected to COM2.
8-56
Using Control Pins
1. Create the serial port objects – After the modems are powered on, the serial
port object s1 is created for the first modem, and the serial port object s2 is
created for the second modem.
s1 = serial('COM1');
s2 = serial('COM2');
2. Connect to the devices – s1 and s2 are connected to the modems. Since the
default value for the ReadAsyncMode property is continuous, data is
asynchronously returned to the input buffers as soon as it is available from the
modems.
fopen(s1)
fopen(s2)
Since the default value of the DataTerminalReady property is on, the computer
(data terminal) is now ready to exchange data with the modems. You can verify
that the modems (data sets) are ready to communicate with the computer by
examining the value of the Data Set Ready pin using the PinStatus property.
s1.Pinstatus
ans =
CarrierDetect:
ClearToSend:
DataSetReady:
RingIndicator:
'off'
'on'
'on'
'off'
The value of the DataSetReady field is on since both modems were powered on
before they were connected to the objects.
3. Configure properties – Both modems are configured for a baud rate of 2400
bits per second and a carriage return (CR) terminator.
s1.BaudRate =
s1.Terminator
s2.BaudRate =
s2.Terminator
2400;
= 'CR';
2400;
= 'CR';
4. Write and read data – Write the atd command to the first modem. This
command puts the modem “off the hook,” which is equivalent to manually
lifting a phone receiver.
fprintf(s1,'atd')
8-57
8
Serial Port I/O
Write the ata command to the second modem. This command puts the modem
in “answer mode,” which forces it to connect to the first modem.
fprintf(s2,'ata')
After the two modems negotiate their connection, you can verify the connection
status by examining the value of the Carrier Detect pin using the PinStatus
property.
s1.PinStatus
ans =
CarrierDetect:
ClearToSend:
DataSetReady:
RingIndicator:
'on'
'on'
'on'
'off'
You can also verify the modem-modem connection by reading the descriptive
message returned by the second modem.
s2.BytesAvailable
ans =
25
out = fread(s2,25);
char(out)'
ans =
ata
CONNECT 2400/NONE
Now break the connection between the two modems by configuring the
DataTerminalReady property to off. You can verify that the modems are
disconnected by examining the Carrier Detect pin value.
s1.DataTerminalReady = 'off';
s1.PinStatus
ans =
CarrierDetect: 'off'
ClearToSend: 'on'
DataSetReady: 'on'
RingIndicator: 'off'
5. Disconnect and clean up – Disconnect the objects from the modems, and
remove the objects from memory and from the MATLAB workspace.
8-58
Using Control Pins
fclose([s1 s2])
delete([s1 s2])
clear s1 s2
Controlling the Flow of Data: Handshaking
Data flow control or handshaking is a method used for communicating between
a DCE and a DTE to prevent data loss during transmission. For example,
suppose your computer can receive only a limited amount of data before it must
be processed. As this limit is reached, a handshaking signal is transmitted to
the DCE to stop sending data. When the computer can accept more data,
another handshaking signal is transmitted to the DCE to resume sending data.
If supported by your device, you can control data flow using one of these
methods:
• Hardware handshaking
• Software handshaking
Note Although you may be able to configure your device for both hardware
handshaking and software handshaking at the same time, MATLAB does not
support this behavior.
You can specify the data flow control method with the FlowControl property.
If FlowControl is hardware, then hardware handshaking is used to control
data flow. If FlowControl is software, then software handshaking is used to
control data flow. If FlowControl is none, then no handshaking is used.
Hardware Handshaking
Hardware handshaking uses specific serial port pins to control data flow. In
most cases, these are the RTS and CTS pins. Hardware handshaking using
these pins is described in “The RTS and CTS Pins” on page 8-10.
If FlowControl is hardware, then the RTS and CTS pins are automatically
managed by the DTE and DCE. You can return the CTS pin value with the
PinStatus property. You can configure or return the RTS pin value with the
RequestToSend property.
8-59
8
Serial Port I/O
Note Some devices also use the DTR and DSR pins for handshaking.
However, these pins are typically used to indicate that the system is ready for
communication, and are not used to control data transmission. In MATLAB,
hardware handshaking always uses the RTS and CTS pins.
If your device does not use hardware handshaking in the standard way, then
you may need to manually configure the RequestToSend property. In this case,
you should configure FlowControl to none. If FlowControl is hardware, then
the RequestToSend value that you specify may not be honored. Refer to the
device documentation to determine its specific pin behavior.
Software Handshaking
Software handshaking uses specific ASCII characters to control data flow.
These characters, known as Xon and Xoff (or XON and XOFF), are described
below.
Table 8-12: Software Handshaking Characters
Character
Decimal
Value
Description
Xon
17
Resume data transmission
Xoff
19
Pause data transmission
When using software handshaking, the control characters are sent over the
transmission line the same way as regular data. Therefore you need only the
TD, RD, and GND pins.
The main disadvantage of software handshaking is that you cannot write the
Xon or Xoff characters while numerical data is being written to the device. This
is because numerical data may contain a 17 or 19, which makes it impossible
to distinguish between the control characters and the data. However, you can
write Xon or Xoff while data is being asynchronously read from the device since
you are using both the TD and RD pins, respectively.
8-60
Using Control Pins
Example: Using Software Handshaking
Suppose you want to use software flow control with the example described in
“Example: Reading Binary Data” on page 8-45. To do this, you must configure
the oscilloscope and serial port object for software flow control.
fprintf(s,'RS232:SOFTF ON')
s.FlowControl = 'software';
To pause data transfer, you write the numerical value 19 to the device.
fwrite(s,19)
To resume data transfer, you write the numerical value 17 to the device.
fwrite(s,17)
8-61
8
Serial Port I/O
Debugging: Recording Information to Disk
While the serial port object is connected to the device, you can record this
information to a disk file:
• The number of values written to the device, the number of values read from
the device, and the data type of the values
• Data written to the device, and data read from the device
• Event information
Recording information to disk provides a permanent record of your serial port
session, and is an easy way to debug your application.
You record information to a disk file with the record function. The properties
associated with recording information to disk are given below.
Table 8-13: Recording Properties
Property Name
Description
RecordDetail
Specify the amount of information saved to a record
file
RecordMode
Specify whether data and event information is
saved to one record file or to multiple record files
RecordName
Specify the name of the record file
RecordStatus
Indicate if data and event information are saved to
a record file
Example: Introduction to Recording Information
This example records the number of values written to and read from the device,
and stores the information to the file myfile.txt
s = serial('COM1');
fopen(s)
s.RecordName = 'myfile.txt';
record(s)
fprintf(s,'*IDN?')
idn = fscanf(s);
8-62
Debugging: Recording Information to Disk
fprintf(s,'RS232?')
rs232 = fscanf(s);
End the serial port session.
fclose(s)
delete(s)
clear s
You can use the type command to display myfile.txt at the command line.
Creating Multiple Record Files
When you initiate recording with the record function, the RecordMode property
determines if a new record file is created or if new information is appended to
an existing record file.
You can configure RecordMode to overwrite, append, or index. If RecordMode
is overwrite, then the record file is overwritten each time recording is
initiated. If RecordMode is append, then the new information is appended to the
file specified by RecordName. If RecordMode is index, a different disk file is
created each time recording is initiated. The rules for specifying a record
filename are discussed in the next section.
Specifying a Filename
You specify the name of the record file with the RecordName property. You can
specify any value for RecordName – including a directory path – provided the
filename is supported by your operating system. Additionally, if RecordMode is
index, then the filename follows these rules:
• Indexed filenames are identified by a number. This number precedes the
filename extension and is increased by 1 for successive record files.
• If no number is specified as part of the initial filename, then the first record
file does not have a number associated with it. For example, if RecordName is
myfile.txt, then myfile.txt is the name of the first record file,
myfile01.txt is the name of the second record file, and so on.
• RecordName is updated after the record file is closed.
• If the specified filename already exists, then the existing file is overwritten.
8-63
8
Serial Port I/O
The Record File Format
The record file is an ASCII file that contains a record of one or more serial port
sessions. You specify the amount of information saved to a record file with the
RecordDetail property.
RecordDetail can be compact or verbose. A compact record file contains the
number of values written to the device, the number of values read from the
device, the data type of the values, and event information. A verbose record file
contains the preceding information as well as the data transferred to and from
the device.
Binary data with precision given by uchar, schar, (u)int8, (u)int16 or (u)int32
is recorded using hexadecimal format. For example, if the decimal value 255 is
read from the instrument as a 16-bit integer, the hexadecimal value 00FF is
saved in the record file. Single- and double-precision floating-point numbers
are recorded as decimal values using the %g format, and as hexadecimal values
using the format specified by the IEEE Standard 754-1985 for Binary
Floating-Point Arithmetic.
The IEEE floating-point format includes three components: the sign bit, the
exponent field, and the significand field. Single-precision floating-point values
consist of 32 bits. The value is given by
value = ( – 1 )
sign
(2
exp-127
) ( 1.significand )
Double-precision floating-point values consist of 64 bits. The value is given by
value = ( – 1 )
sign
(2
exp-1023
) ( 1.significand )
The floating-point format component, and the associated single-precision and
double-precision bits are given below.
Component
Single-Precision Bits
Double-Precision Bits
sign
1
1
exp
2-9
2-12
significand
10-32
13-64
Bit 1 is the left-most bit as stored in the record file.
8-64
Debugging: Recording Information to Disk
Example: Recording Information to Disk
This example illustrates how to record information transferred between a
serial port object and a Tektronix TDS 210 oscilloscope. Additionally, the
structure of the resulting record file is presented.
1. Create the serial port object – Create the serial port object s associated
with the serial port COM1.
s = serial('COM1');
2. Connect to the device – Connect s to the oscilloscope. Since the default
value for the ReadAsyncMode property is continuous, data is asynchronously
returned the input buffer as soon as it is available from the instrument.
fopen(s)
3. Configure property values – Configure s to record information to multiple
disk files using the verbose format. Recording is then initiated with the first
disk file defined as WaveForm1.txt.
s.RecordMode = 'index';
s.RecordDetail = 'verbose';
s.RecordName = 'WaveForm1.txt';
record(s)
4. Write and read data – The commands written to the instrument, and the
data read from the instrument are recorded in the record file. Refer to
“Example: Writing and Reading Text Data” on page 8-42 for an explanation of
the oscilloscope commands.
fprintf(s,'*IDN?')
idn = fscanf(s);
fprintf(s,'MEASUREMENT:IMMED:SOURCE CH2')
fprintf(s,'MEASUREMENT:IMMED:SOURCE?')
source = fscanf(s);
Read the peak-to-peak voltage with the fread function. Note that the data
returned by fread is recorded using hex format.
fprintf(s,'MEASUREMENT:MEAS1:TYPE PK2PK')
fprintf(s,'MEASUREMENT:MEAS1:VALUE?')
ptop = fread(s,s.BytesAvailable);
Convert the peak-to-peak voltage to a character array.
8-65
8
Serial Port I/O
char(ptop)'
ans =
2.0199999809E0
The recording state is toggled from on to off. Since the RecordMode value is
index, the record filename is automatically updated.
record(s)
s.RecordStatus
ans =
off
s.RecordName
ans =
WaveForm2.txt
5. Disconnect and clean up – When you no longer need s, you should
disconnect it from the instrument, and remove it from memory and from the
MATLAB workspace.
fclose(s)
delete(s)
clear s
The Record File Contents
The contents of the WaveForm1.txt record file are shown below. Since the
RecordDetail property was verbose, the number of values, commands, and
data were recorded. Note that data returned by the fread function is in hex
format.
type WaveForm1.txt
Legend:
* - An event occurred.
> - A write operation occurred.
< - A read operation occurred.
1
2
3
4
8-66
Recording on 22-Jan-2000 at 11:21:21.575. Binary data in...
> 6 ascii values.
*IDN?
< 56 ascii values.
TEKTRONIX,TDS 210,0,CF:91.1CT FV:v1.16 TDS2CM:CMV:v1.04
> 29 ascii values.
Debugging: Recording Information to Disk
5
6
7
8
9
10
MEASUREMENT:IMMED:SOURCE CH2
> 26 ascii values.
MEASUREMENT:IMMED:SOURCE?
< 4 ascii values.
CH2
> 27 ascii values.
MEASUREMENT:MEAS1:TYPE PK2PK
> 25 ascii values.
MEASUREMENT:MEAS1:VALUE?
< 15 uchar values.
32 2e 30 31 39 39 39 39 39 38 30 39 45 30 0a
Recording off.
8-67
8
Serial Port I/O
Saving and Loading
You can save serial port objects to a MAT-file just as you would any workspace
variable – using the save command. For example, suppose you create the serial
port object s associated with the serial port COM1, configure several property
values, and perform a write and read operation.
s = serial('COM1');
s.BaudRate = 19200;
s.Tag = 'My serial object';
fopen(s)
fprintf(s, '*IDN?')
out = fscanf(s);
To save the serial port object and the data read from the device to the MAT-file
myserial.mat
save myserial s out
Note You can save data and event information as text to a disk file with the
record function.
You can recreate s and out in the workspace using the load command.
load myserial
Values for read-only properties are restored to their default values upon
loading. For example, the Status property is restored to closed. Therefore, to
use s, you must connect it to the device with the fopen function. To determine
if a property is read-only, examine its reference pages.
Using Serial Port Objects on Different Platforms
If you save a serial port object from one platform, and then load that object on
a different platform having different serial port names, then you will need to
modify the Port property value. For example, suppose you create the serial port
object s associated with the serial port COM1 on a Windows platform. If you
want to save s for eventual use on a Linux platform, you should configure Port
to an appropriate value such as ttyS0 after the object is loaded.
8-68
Disconnecting and Cleaning Up
Disconnecting and Cleaning Up
When you no longer need your serial port object, you should disconnect it from
the device, and clean up your MATLAB environment by removing the object
from memory and from the workspace. These are the steps you take to end a
serial port session.
Disconnecting a Serial Port Object
When you no longer need to communicate with the device, you should
disconnect it from the serial port object with the fclose function.
fclose(s)
You can examine the Status property to verify that the serial port object and
the device are disconnected.
s.Status
ans =
closed
After fclose is issued, the serial port associated with s is available. You can
now connect another serial port object to it using fopen.
Cleaning Up the MATLAB Environment
When you no longer need the serial port object, you should remove it from
memory with the delete function.
delete(s)
Before using delete, you must disconnect the serial port object from the device
with the fclose function.
A deleted serial port object is invalid, which means that you cannot connect it
to the device. In this case, you should remove the object from the MATLAB
workspace. To remove serial port objects and other variables from the
MATLAB workspace, use the clear command.
clear s
If you use clear on a serial port object that is still connected to a device, the
object is removed from the workspace but remains connected to the device. You
can restore cleared objects to MATLAB with the instrfind function.
8-69
8
Serial Port I/O
Property Reference
This section includes information to help you learn about serial port properties.
The information is organized using these topics:
• “The Property Reference Page Format” describes the organization of the
property reference pages.
• “Serial Port Object Properties” summarizes the properties using several
categories based on how they are used. Following this section, the properties
are listed alphabetically and described in detail.
The Property Reference Page Format
Each serial port property description contains some or all of this information:
• The property name
• A description of the property
• The property characteristics, including:
- Read only – the condition under which the property is read only
A property can be read only always, never, while the serial port object is
open, or while the serial port object is recording. You can configure a
property value using the set function or dot notation. You can return the
current property value using the get function or dot notation.
- Data type – the property data type
This is the data type you use when specifying a property value.
• Valid property values including the default value
When property values are given by a predefined list, the default value is
usually indicated by {}.
• An example using the property
• Related properties and functions
8-70
Property Reference
Serial Port Object Properties
The serial port object properties are briefly described below, and organized into
categories based on how they are used. Following this section, the properties
are listed alphabetically and described in detail.
Communications Properties
BaudRate
Specify the rate at which bits are transmitted
DataBits
Specify the number of data bits to transmit
Parity
Specify the type of parity checking
StopBits
Specify the number of bits used to indicate the end of a
byte
Terminator
Specify the terminator character
Write Properties
BytesToOutput
Indicate the number of bytes currently in the output
buffer
OutputBufferSize
Specify the size of the output buffer in bytes
Timeout
Specify the waiting time to complete a read or write
operation
TransferStatus
Indicate if an asynchronous read or write operation
is in progress
ValuesSent
Indicate the total number of values written to the
device
8-71
8
Serial Port I/O
Read Properties
BytesAvailable
Indicate the number of bytes available in the input
buffer
InputBufferSize
Specify the size of the input buffer in bytes
ReadAsyncMode
Specify whether an asynchronous read operation is
continuous or manual
Timeout
Specify the waiting time to complete a read or write
operation
TransferStatus
Indicate if an asynchronous read or write operation is
in progress
ValuesReceived
Indicate the total number of values read from the
device
Callback Properties
8-72
BreakInterrupt
Fcn
Specify the M-file callback function to execute when a
break-interrupt event occurs
BytesAvailable
Fcn
Specify the M-file callback function to execute when a
specified number of bytes is available in the input
buffer, or a terminator is read
BytesAvailable
FcnCount
Specify the number of bytes that must be available in
the input buffer to generate a bytes-available event
BytesAvailable
FcnMode
Specify if the bytes-available event is generated after
a specified number of bytes is available in the input
buffer, or after a terminator is read
ErrorFcn
Specify the M-file callback function to execute when
an error event occurs
Property Reference
Callback Properties (Continued)
OutputEmptyFcn
Specify the M-file callback function to execute when
the output buffer is empty
PinStatusFcn
Specify the M-file callback function to execute when
the CD, CTS, DSR, or RI pins change state
TimerFcn
Specify the M-file callback function to execute when a
predefined period of time passes
TimerPeriod
Specify the period of time between timer events
Control Pin Properties
DataTerminal
Ready
Specify the state of the DTR pin
FlowControl
Specify the data flow control method to use
PinStatus
Indicate the state of the CD, CTS, DSR, and RI pins
RequestToSend
Specify the state of the RTS pin
Recording Properties
RecordDetail
Specify the amount of information saved to a record
file
RecordMode
Specify whether data and event information are saved
to one record file or to multiple record files
RecordName
Specify the name of the record file
RecordStatus
Indicate if data and event information are saved to a
record file
8-73
8
Serial Port I/O
General Purpose Properties
8-74
ByteOrder
Specify the order in which the device stores bytes
Name
Specify a descriptive name for the serial port object
Port
Indicate the platform-specific serial port name
Status
Indicate if the serial port object is connected to the
device
Tag
Specify a label to associate with a serial port object
Type
Indicate the object type
UserData
Specify data that you want to associate with a serial
port object
Index
Symbols
%val 3-5, 3-8
allocating memory 3-25
DIGITAL Visual Fortran 3-8
A
ActiveX
automation client 7-4
automation server 7-4, 7-16
callback 7-8
client support 7-6
collections 7-10
concepts 7-3
control containment 7-4
controller 7-16
Count property 7-10
event handler function 7-8
events 7-3
Execute method 7-16
integration 7-3
interface 7-3
custom 7-4
standard 7-4
invoking event 7-8
Item method 7-10
launching server 7-21
limitations of MATLAB support 7-13
MATLAB as automation client 7-13
methods 7-3
object model 7-3
ProgID 7-6
properties 7-3
server 7-16
use in the MATLAB Engine 4-4
add-in
MATLAB for Visual Studio 1-29
API
access methods 1-4
memory management 1-37
argument checking 2-9
argument passing, from Java methods
data conversion 5-56
built-in data types 5-57
conversions you can perform 5-57
Java objects 5-57
argument passing, to Java methods
data conversion 5-46
built-in arrays 5-48
built-in data types 5-48
Java object arrays 5-52
Java object cell arrays 5-53
Java objects 5-50
objects of Object class 5-51
string arrays 5-50
string data types 5-49
effect of dimension on 5-53
argument type, Java
effect on method dispatching 5-54
array access methods 3-7
mat 6-3
arrays
cell 1-8
empty 1-9
hybrid 2-40
logical 1-9
MATLAB 1-6
multidimensional 1-8
persistent 2-38
serial port object 8-26
sparse 2-29
temporary 2-38, 3-37
arrays, Java
I-1
Index
accessing elements of 5-36
assigning
the empty matrix 5-41
values to 5-39
with single subscripts 5-40
comparison with MATLAB arrays 5-30
concatenation of 5-43
creating a copy 5-45
creating a reference 5-44
creating in MATLAB 5-34
creating with javaArray 5-34
dimensionality of 5-30
dimensions 5-33
indexing 5-32
with colon operator 5-37
with single subscripts 5-36, 5-38
linear arrays 5-41
passed by reference 5-49
representing in MATLAB 5-30
sizing 5-32
subscripted deletion 5-42
using the end subscript 5-39
ASCII file mode 6-6
ASCII flat file 6-4
automation
client 7-13
controller 7-16
server 7-16
bccengmatopts.bat 1-18
bccopts.bat 1-17
binary data
reading from a device 8-41
writing to a device 8-36
binary file mode 6-6
BLAS and LAPACK functions 2-40
building MEX files for 2-44
example of 2-46
handling complex numbers 2-42
passing arguments 2-41
specifying the function name 2-41
BSTR 7-17
buffer
input, serial port object 8-38
output, serial port object 8-32
C
C example
convec.c 2-21
doubleelem.c 2-24
findnz.c 2-26
fulltosparse.c 2-29
phonebook.c 2-17
revord.c 2-11
sincall.c 2-33
timestwo.c 2-8
timestwoalt.c 2-10
xtimesy.c 2-14
B
bcc53engmatopts.bat 1-18
bcc53opts.bat 1-17
bcc54engmatopts.bat 1-18
bcc54opts.bat 1-17
bcc55engmatopts.bat 1-18
bcc55opts.bat 1-17
I-2
C language
data types 1-9
debugging 2-47
MEX-files 2-1
C language example
basic 2-8
calling MATLAB functions 2-33
Index
calling user-defined functions 2-33
handling 8-, 16-, 32-bit data 2-23
handling arrays 2-25
handling complex data 2-21
handling sparse arrays 2-29
passing multiple values 2-13
persistent array 2-39
strings 2-11
callback 7-8
serial port object 8-48
functions 8-52
properties 8-49
caller workspace 2-37
cat
using with Java arrays 5-43
using with Java objects 5-12
cell 1-9
using with Java objects 5-60
cell arrays 1-8, 2-16
converting from Java object 5-60
char 1-9
overloading toChar in Java 5-58
class
using in Java 5-17
classes, Java 5-5
built-in 5-5
defining 5-5
identifying using which 5-25
importing 5-8
loading into workspace 5-7
making available to MATLAB 5-6
sources for 5-5
third-party 5-5
user-defined 5-5
classpath.txt
finding and editing 5-7
using with Java archive files 5-6
using with Java classes 5-6
using with Java packages 5-6
client (DDE) 7-23
collections 7-10
colon
using in Java array access 5-37
using in Java array assignment 5-41
COM (component object model) 7-3
object model 7-4
commands. See individual commands.
Compaq Visual Fortran compiler
debugging 3-39
compiler
changing on UNIX 1-12
debugging
Compaq Visual Fortran 3-39
Microsoft 2-48
Watcom 2-49
preconfigured options file 1-17–1-19
selecting on Windows 1-14
supported 1-11
compiling
engine application
UNIX 4-18–4-19
Windows 4-19
MAT-file application
UNIX 6-29
Windows 6-30
complex data
in Fortran 3-22
compopts.bat 1-23
computational routine 2-3, 3-3
concatenation
of Java arrays 5-43
of Java objects 5-12
configuration 1-11
problems 1-35
I-3
Index
UNIX 1-12
Windows 1-14, 1-16
control pins
serial port object, using 8-56
convec.c 2-21
convec.f 3-22
conversation (DDE) 7-23
conversion, data
in Java method arguments 5-46
copying a Java array 5-45
Count property 7-10
cxxopts.sh 1-19
D
data access
within Java objects 5-16
data bits 8-13
data format
serial port 8-11
data storage 1-6
data type 2-7
C language 1-9
cell arrays 1-8
checking 2-9
complex double-precision nonsparse matrix
1-7
empty arrays 1-9
Fortran language 1-9
logical arrays 1-9
MAT-file 6-6
MATLAB 1-9
MATLAB string 1-7
multidimensional arrays 1-8
numeric matrix 1-7
objects 1-8
sparse arrays 2-29
I-4
sparse matrices 1-7
structures 1-8
data, MATLAB 1-6
exporting from 6-4–6-5
importing to 6-3–6-4
dblmat.f 3-25
dbmex 3-38
DCE 8-5
DCOM (distributed component object model) 7-22
using MATLAB as a server 7-22
DDE (dynamic data exchange) 7-23
accessing MATLAB as server 7-25
advisory links 7-32
client 7-23
conversation 7-23
hot link 7-32
item 7-24
MATLAB
requesting data from 7-28
sending commands to 7-27
sending data to 7-29
using as client 7-31
name hierarchy 7-26
notifying when data changes 7-32
server 7-23
service name 7-24
topic 7-24, 7-26–7-30
engine 7-27
system 7-26
warm link 7-32
Windows clipboard formats 7-24–7-25
ddeadv 7-31
ddeexec 7-31
ddeinit 7-31
ddepoke 7-31
ddereq 7-31
ddeterm 7-31
Index
ddeunadv 7-31
debugging C language MEX-files 2-47
UNIX 2-47
Windows 2-48
debugging Fortran language MEX-files
UNIX 3-38
Windows 3-39
DEC Alpha
declaring pointers 3-5
df50engmatopts.bat 1-19
df50opts.bat 1-18
df60engmatopts.bat 1-19
df60opts.bat 1-18
diary 6-4
diary file 6-4
DIGITAL Visual Fortran compiler
debugging 3-39
directory
eng_mat 1-47
mex 1-47
mx 1-47
refbook 1-46
directory organization
MAT-file application 6-8
Microsoft Windows 1-44
UNIX 1-42
directory path
convention 1-2
display
serial port object 8-25
display function
overloading toString in Java 5-26
distributed component object model. See DCOM.
dll extension 1-4
DLLs 1-11
locating 1-32
documenting MEX-file 2-37, 3-36
double 1-9
overloading toDouble in Java 5-58
doubleelem.c 2-24
DTE 8-5
dynamic data exchange. See DDE.
dynamic memory allocation
in Fortran 3-25
mxCalloc 2-11
dynamically linked subroutines 1-3
E
empty arrays 1-9
empty matrix
conversion to Java NULL 5-54
in Java array assignment 5-41
empty string
conversion to Java object 5-54
end
use with Java arrays 5-39
eng_mat directory 1-47, 4-6
engClose 4-3, 4-4
engdemo.c 4-6
engEvalString 4-4
engGetArray 4-3
engGetMatrix 4-4
engGetVisible 4-4
engine
compiling 4-17
linking 4-17
UNIX 4-18
windows 4-19
engine application
Windows 4-19
engine example
calling MATLAB
from C program 4-6
I-5
Index
from Fortran program 4-11
engine functions 4-3–4-4
engine library 4-2
communicating with MATLAB
UNIX 4-4
Windows 4-4
engOpen 4-3, 4-4
engOpenSingleUse 4-4
engopts.sh 1-19
engOutputBuffer 4-4
engPutArray 4-3
engPutMatrix 4-4
engSetVisible 4-4
engwindemo.c 4-6, 6-24
event handler
function 7-8
writing 7-8
events
serial port object 8-48
storing information 8-50
types 8-49
examples, Java programming
communicating through a serial port 5-68
creating and using a phone book 5-73
finding an internet protocol address 5-66
reading a URL 5-63
exception
floating-point 4-17, 6-28
exceptions, Java
handling 5-28
explore example 1-9
extension
MEX-file 1-3
F
-f option 1-17
I-6
fengdemo.f 4-11
fieldnames
using with Java objects 5-15
file mode
ASCII 6-6
binary 6-6
files
flat 6-4
linking multiple 2-37, 3-36
findnz.c 2-26
floating-point exceptions
Borland C++ Compiler on Windows 4-17, 6-28
engine applications 4-17
masking 4-17, 6-28
MAT-file applications 6-28
fopen 6-4, 6-5
Fortran
case in 3-6
data types 1-9
pointers
concept 3-5, 3-17
declaring 3-5
Fortran examples
convec.f 3-22
dblmat.f 3-25
fulltosparse.f 3-28
matsq.f 3-17
passstr.f 3-15
revord.f 3-12
sincall.f 3-32
timestwo.f 3-10
xtimesy.f 3-19
Fortran language examples
calling MATLAB functions 3-32
handling complex data 3-22
handling sparse matrices 3-28
passing arrays of strings 3-14
Index
passing matrices 3-17
passing multiple values 3-19
passing scalar 2-7, 3-10
passing strings 3-12
Fortran language MEX-files 3-3
components 3-3
fread 6-4
fulltosparse.c 2-29
fulltosparse.f 3-28
function handles
serial port object callback 8-52
fwrite 6-5
import
using with Java classes 5-8
include directory 6-9
indexing Java arrays
using single colon subscripting 5-38
using single subscripting 5-36
internet protocol address
Java example 5-66
ir 1-7, 2-29, 3-28
isa
using with Java objects 5-18
isjava
using with Java objects 5-18
Item method 7-10
G
-g option 2-47
gateway routine 2-3, 3-3, 3-6
accessing mxArray data 2-3
gccopts.sh 1-19
GetFullMatrix 7-17
H
handshaking
serial port object 8-59
help 2-37, 3-36
help files 2-37, 3-36
hybrid array
persistent 2-40
temporary 2-40
hybrid arrays 2-40
I
IDE
building MEX-files 1-20
IEEE routines 1-4
J
Java
API class packages 5-3
archive (JAR) files 5-6
development kit 5-5
Java Virtual Machine (JVM) 5-3
packages 5-6
Java, MATLAB interface to
arguments passed to Java methods 5-46
arguments returned from Java methods 5-56
arrays, working with 5-29
benefits of 5-3
classes, using 5-5
examples 5-62
methods, invoking 5-19
objects, creating and using 5-10
overview 5-3
javaArray function 5-34
jc 1-7, 2-29, 3-28
I-7
Index
L
LAPACK and BLAS functions 2-40
building MEX files for 2-44
example of 2-46
handling complex numbers 2-42
passing arguments 2-41
specifying the function name 2-41
lccengmatopts.bat 1-18
lccopts.bat 1-17
library path
setting on UNIX 4-18, 6-29
linking DLLs to MEX-files 1-28
linking multiple files 2-37, 3-36
load 6-4, 6-6
using with Java objects 5-14
loading
serial port objects 8-68
locating DLLs 1-32
logical arrays 1-9
M
mat.h 6-9
matClose 6-7
matDeleteArray 6-7
matDeleteMatrix 6-8
matdemo1.f 6-21
matdemo2.f 6-24
MAT-file
C language
reading 6-16
compiling 6-28
data types 6-6
examples 6-9
Fortran language
creating 6-21
reading 6-24
I-8
linking 6-28
overview 6-2
subroutines 6-6
UNIX libraries 6-9
using 6-3
Windows libraries 6-9
MAT-file application
UNIX 6-29
Windows 6-30
MAT-file example
creating
C language 6-11
Fortran language 6-21
reading
C language 6-16
Fortran language 6-24
MAT-functions 6-7–6-8
matGetArray 6-7
matGetArrayHeader 6-7
matGetDir 6-7, 6-8
matGetFp 6-7
matGetMatrix 6-8
matGetNextArray 6-7
matGetNextArrayHeader 6-7
matGetNextMatrix 6-8
matGetString 6-8
MATLAB
ActiveX interface 7-10
arrays 1-6
as DCOM server client 7-12
data 1-6
data file format 6-3
data storage 1-6
data type 1-9
engine 4-2
exporting data 6-3–6-5
importing data 6-3–6-4
Index
MAT-file 6-6
reading arrays from 6-6
saving arrays to 6-6
moving data between platforms 6-5–6-6
stand-alone applications 6-2, 6-3
string 1-7
using as a computation engine 4-2
variables 1-6
MATLAB add-in for Visual Studio
configuring on Windows 95 and 98 1-30
configuring on Windows ME 1-30
MATLAB Automation Server
supported methods 7-17
MATLAB for Visual Studio add-in 1-29
matOpen 6-7
matopts.sh 1-19
matPutArray 6-7
matPutArrayAsGlobal 6-7
matPutMatrix 6-8
matPutString 6-8
matrix
complex double-precision nonsparse 1-7
numeric 1-7
sparse 1-7, 3-28
matrix.h 6-9
matsq.f 3-17
memory
allocation 2-11
leak 1-40, 2-38
temporary 3-37
memory management 1-37, 2-38, 3-37
API 1-37
compatibility 1-37
routines 1-4
special considerations 2-38
methods
using with Java methods 5-24
methods, Java
calling syntax 5-19
converting input arguments 5-46
displaying 5-24
displaying information about 5-22
finding the defining class 5-25
overloading 5-54
passing data to 5-46
static 5-21
undefined 5-27
methodsview 5-22
output fields 5-24
mex
<ENV_VAR>#<val> 1-22
-g 2-47, 3-38
mex build script 1-20, 2-9
default options file, UNIX 1-22
default options file, Windows 1-23
switches 1-21
<name>#<def> 1-22
<name>=<def> 1-22
@<rsp_file> 1-21
-argcheck 1-21
-c 1-21
-D<name>[#<def>] 1-21
-f <file> 1-21
-fortran 3-36
-g 1-21
-h[elp] 1-21
-I<pathname> 1-21
-inline 1-21
-l<file> 1-21
-L<pathname> 1-21
-O 1-22
-outdir <name> 1-22
-output <name> 1-22
-setup 1-14, 1-22
I-9
Index
-U<name> 1-22
-v 1-22
-V4 1-22
mex directory 1-47
mex.bat 2-9
mex.m 2-9
mex.sh 2-9
mexAtExit 2-39
register a function 2-39
mexaxp extension 1-3
mexCallMATLAB 2-33, 2-35, 2-38, 3-32, 3-34, 3-35
mexErrMsgTxt 2-38, 3-7
mexEvalString 2-37, 3-36
MEX-file 1-3
advanced topics 2-37
Fortran 3-36
applications of 1-3
arguments 2-5
C language 2-1
calling 1-4
compiling 2-9
Microsoft Visual C++ 1-28
UNIX 1-12, 1-24–1-26
Windows 1-16, 1-26–1-29
components 2-3
computation error 1-37
configuration problem 1-35
creating C language 2-3, 2-9
creating Fortran language 3-3
custom building 1-20
debugging C language 2-47
debugging Fortran language 3-38
DLL linking 1-28
documenting 2-37, 3-36
dynamically allocated memory 2-38
examples 2-7, 3-9
extensions 1-3
I-10
load error 1-35
overview 1-3
passing cell arrays 2-16
passing structures 2-16
problems 1-33–1-37
segmentation error 1-36
syntax errors 1-35
temporary array 2-38
using 1-3
versioning 1-28
mexFunction 2-3, 3-3, 3-6
altered name 3-39
parameters 2-3, 3-3
mexGetArray 2-37
mexGetMatrix 3-36
mexglx extension 1-3
mexhp7 extension 1-3
mexhpux extension 1-3
mexMakeArrayPersistent 2-38
mexMakeMemoryPersistent 2-38
mexopts.bat 1-23
mexopts.sh 1-19
mexPutArray 2-37
mexPutMatrix 3-36
mexrs6 extension 1-3
mexSetTrapFlag 2-38
mexsg extension 1-3
mexsol extension 1-4
mexversion.rc 1-28
M-file
creating data 6-3
Microsoft compiler
debugging 2-48
Microsoft Windows
directory organization 1-44
msvc50engmatopts.bat 1-18
msvc50opts.bat 1-18
Index
msvc60engmatopts.bat 1-18
mxSetImagData 1-40, 1-41
msvc60opts.bat 1-18
mxSetIr 1-41
multidimensional arrays 1-8
mx directory 1-47
mxArray 1-6, 3-7
accessing data 2-3
contents 1-6
improperly destroying 1-38
ir 1-8
jc 1-8
nzmax 1-8
pi 1-8
pr 1-8
temporary with improper data 1-40
type 1-6
mxCalloc 2-11, 2-38, 3-7
mxCopyComplex16ToPtr 3-22
mxCopyPtrToComplex16 3-22
mxCopyPtrToReal8 3-8, 3-19
mxCreateFull 3-7, 3-17
mxCreateNumericArray 2-23
mxCreateSparse 3-7
mxCreateString 2-13, 3-7
mxDestroyArray 1-37, 2-40, 3-37
mxFree 1-38
mxGetCell 2-16
mxGetData 2-16, 2-23, 2-26
mxGetField 2-16
mxGetImagData 2-23, 2-26
mxGetPi 2-21, 3-17
mxGetPr 2-16, 2-21, 3-17
mxGetScalar 2-10, 2-16
mxMalloc 2-11, 2-38
mxRealloc 2-11, 2-38
mxSetCell 1-39, 2-40
mxSetData 1-40, 1-41, 2-40
mxSetField 1-39
mxSetJc 1-41
mxSetPi 1-40, 1-41
mxSetPr 1-40, 2-40
mxUNKNOWN_CLASS 2-35, 3-35
N
ndims
using with Java arrays 5-33
nlhs 2-3, 2-5, 3-3, 3-6
nrhs 2-3, 2-5, 3-3, 3-6
null modem cable 8-6
numeric matrix 1-7
nzmax 1-8, 3-28
O
objects 1-8
serial port 8-24
objects, Java
accessing data within 5-16
concatenating 5-12
constructing 5-10
converting to MATLAB cell array 5-60
converting to MATLAB structures 5-59
identifying fieldnames 5-15
information about 5-17
class name 5-18
class type 5-18
passing by reference 5-12
saving and loading 5-14
options file
creating new 1-20
modifying 1-20
preconfigured 1-17
I-11
Index
specifying 1-17
when to specify 1-17
overloading Java methods 5-54
P
parity bit 8-14
passing data to Java methods 5-46
passstr.f 3-15
persistent arrays
exempting from cleanup 2-38
phonebook.c 2-17
pi 1-7
plhs 2-3, 2-5, 3-3, 3-6
pointer 3-5
Fortran language MEX-file 3-17
pr 1-7
prhs 2-3, 2-5, 3-3, 3-6
properties
serial port object 8-71
protocol
DCOM 7-22
PutFullMatrix 7-19
R
read/write failures, checking for 6-11
reading
binary data from a device 8-41
text data from a device 8-39
record file
serial port object
creating multiple files 8-63
filename 8-63
format 8-64
refbook directory 1-46
references
I-12
to Java arrays 5-44
revord.c 2-11
revord.f 3-12
routine
computational 2-3
gateway 2-3, 3-3
mex 1-4
mx 1-4
RS-232 standard 8-4
S
save 6-5, 6-6
using with Java objects 5-14
saving
serial port objects 8-68
serial port
data format 8-11
devices,connecting 8-5
object creation 8-24
RS-232 standard 8-4
session 8-19
signal and pin assignments 8-6
serial port object
array creation 8-26
callback properties 8-49
configuring communications 8-28
connecting to device 8-27
disconnecting 8-69
display 8-25
event types 8-49
handshaking 8-59
input buffer 8-38
output buffer 8-32
properties 8-71
reading binary data 8-41
reading text data 8-39
Index
recording information to disk 8-62
using control pins 8-56
using events and callbacks 8-48
writing and reading data 8-29
writing binary data 8-36
writing text data 8-34
serializable interface 5-14
service name 7-23, 7-24
session
serial port 8-19
shared libraries directory
UNIX 6-9
Windows 6-9
sincall.c 2-33
sincall.f 3-32
T
temporary arrays 2-38
automatic cleanup 2-38
destroying 1-41
temporary memory
cleaning up 1-41
text data
reading from a device 8-39
writing to a device 8-34
timestwo.c 2-8
timestwo.f 3-10
timestwoalt.c 2-10
troubleshooting
MEX-file creation 1-33
size
using with Java arrays 5-32
sparse 1-9
sparse arrays 2-29
sparse matrices 1-7
start bit 8-13
static data, Java
accessing 5-17
assigning 5-17
static methods, Java 5-21
stop bit 8-13
storing data 1-6
string 1-7
struct 1-9
using with Java objects 5-59
structures, MATLAB 1-8, 2-16
converting from Java object 5-59
subroutines
dynamically linked 1-3
system configuration 1-11
U
uint8 1-9
UNIX
directory organization 1-42
URL
Java example 5-63
V
variable scope 2-37
variables 1-6
versioning MEX-files 1-28
Visual Basic
MATLAB DDE server example 7-29
W
wat11copts.bat 1-18
wat11engmatopts.bat 1-19
Watcom compiler
debugging 2-49
I-13
Index
watcopts.bat 1-18
watengmatopts.bat 1-19
which
using with Java methods 5-25
Windows
ActiveX 7-16
automation 7-16
directory organization 1-44
mex -setup 1-14
selecting compiler 1-14
Windows clipboard format
Metafilepict 7-25
text 7-24
XLTable 7-25
workspace
caller 2-37, 3-36
MEX-file function 2-37, 3-36
write/read failures, checking for 6-11
writing
binary data to a device 8-36
text data to a device 8-34
writing event handlers 7-8
X
xtimesy.c 2-14
xtimesy.f 3-19
Y
yprime.c 1-12, 1-16
yprimef.F 1-12
yprimef.f 1-16
yprimefg.F 1-12
yprimefg.f 1-16
I-14