General Electric GE-625, GE-635 Integrated Data Store Data Management System Reference Manual

COMPUTERS
GE-625/635
Integrated
Data Store
GE-625/635
INTEGRATED
DATA STORE
REFERENCE MANUAL
August 1965
GENERAL @@ ELECTRIC
COMPUTER DEPARTMENT
PREFACE
This manual presents a general description of the IDS system and discusses its features and
capabilities. A description of the source language and detailed programming information is also
presented.
IDS is a mass memory oriented system initially implemented for use with the DS-20 Disc Storage
Unit. Detailed information on this equipment appears in the DS-20 Disc Storage Unit reference
manual, CPB-345. Future releases of the IDS system will be applicable to the DS-15 and DS-25
mass storage units. Manuals on these units will be published shortly.
Since IDS is used to extend the functions of the COBOL language, the reader should have a working
knowledge of that language before using this manual.
Comments on this publication may be addressed to Technical Publications, Computer Department,
General Electric Company, P. O. Box 2961, Phoenix, Arizona, 85002.
(с) 1965 by General Electric Company
GE-600 SERIES Ds
Ta
NTS
CONTE]
Page
1. INTRODUCTION 1
2. INTEGRATED SYSTEMS DESIGN
Comparison of Systems Design Approaches . . . . . . . . . . . . . ... 3
Conventional File Organization . . . . . . . . .. . . ... ...... 3
IDS File Organization . . 1.42 444 4 4 4 4 4 4 4 LL 1 11 LL 10 о
Associating Records intoChains . . . . . . . . . . . .. ......... 6
3. IDS ENVIRONMENT
DS-20 aaa eee aaa eee aaa aaa g
Read/Write System . . . . . . . . . 10
Reference Code . . . . . . . . . . . . .. . 2..2... . . 2... ... 12
Input/Output Controller . . . . . . . . . . . 2 eee. 12
Data Buffers ... 2.201024 4 4 21 4 4 4 4 4 4 4 LVL LL 110 12
Priority Control .. .. 2.420444 44 44 4 4 4 V4 4 1 LL 11110 12
Record Unpacking . . . 014 4 44 4 4 4 4 V4 4 LL LL LL 110 13
File Protection . .. . 1424 4 4 4244 4 4 4 4 A LL LL LL 10 13
4, DATA ORGANIZATION
Data Records and Fields . . . . . 2 42 241 4 4 LL 4 LL 1111110 15
Record Classes . . . 224 4 4 4 44 42 4 4 ea ooo a a ar 17
IDS Chains. 2... 242424 44 44 4 4 4 4 4 4 4 4 aaa errar 17
Multiple Chains . . . 11 422 4 4 4 4 4 4 4 4 ea en ore 18
Chain Processing , . . . 1 2 4 4 4 4 4 4 LL 1 1 LL LL LL 11 LV 1120 20
Linking a New Detail Record . . . . . 1.214424 4 4 1 4 1111120 20
Master Record Selection . . . . 2.114 4 4 e. e ae... e. 21
Chain Ordering . . . 22244414 4 4 4 4 V4 LL 4 LL LL LL 11110 21
Prime Chain 2.244241 44 4 4 4 4 LL 1 LL LL LL LL LAS 21
Data Structure ... 222 4 44 224 4 4 4 4 1 LL ae. a e ea 22
Summary of Data Structures . . . . . . . . . . . . . 2... .... 25
5. IDS PROGRAMMING LANGUAGE
Source Language ... 222124 4 4 4 4 4 0 4 4 4 A LL LL LL LL A0 27
Introduction . . . 22441 4 4 4 4 4 11 1 4 1 1 4 1 111 a ee 27
Identification Division. . . . . . . . . . . . . . 27
GE-600 SERIES =
111
Environment Division
Data Division . . . . « e e «e e e e e. e e e ee ere ree.
File Description . . . . . e «e e e + e ee eere re re erre.
IDS File Description Entry
Record Description . . . . . 2 2 e e eee eee e e
IDS Record Description Entry
TYPE 2... 122221040444 04 604 4 4 404464448845
RETRIEVAL
PLACE . . Hr
PAGE-RANGE
INTERVAL
AUTHORITY
Chain Definition . . . . . «e «+ e «e e e e e e eee ee.
Complete Chain Definition Entry Skeleton
LINKED-PRIOR
RANDOMIZE . . 2. 2 + 4 4 4 4 4 4 44040600 406684648800
CHAIN-ORDER
DUPLICATES
SORT KEY
SELECT . . 2.2.2 2 2 4 e e e 4 6 6 4 0 0 0 0 e 6 60 000800
MATCH-KEY
SYNONYM. .... 2. 2 2 2 2 4 4 6 0 4 4 60 6 0 0 0 0 00 + 000
LINKED MASTER .......+0 0 + + + 6 + + 5 + 6 4 0 655
Sample Coding
Procedure Division
STORE .. . 2.2.2 222 4 64 4 4 4 6 ке
RETRIEVE .... 222 4 4 6 4 6 4 4004 604 0 0000066000
IDS Imperative Statements
MOVE
HEAD
MODIFY
DELETE
СО . . . ee ee ee ee ee eee ee ee ee ee ee eee
PERFORM ..... +. 2 4 2 2 4 4 4 +6 0 6 0 0 0 0 0 4080 080
IDS Conditional Statements
CLOSE IDS
OPEN IDS
«+ + + + | | ® | + | + |. . ФФ в ое Ф + в + т.е.‘ 2
. 8 + | # ©, © | #8 | | = |. ее + = + + >
+. + + + + + + es 2 8 + 4 ee 2 + + es =
о eo хо с 1 #8 ® #8 + я 8 т ес. © & + © 8 |. ео. Ф.Ф. A 8
о а + а в че в и 8 6 # « oe © ноя фт + ¢ #2 +s 0
« & + + « ® в оч о « #8 = s&s 8 8 + « + = 2 + 2 es +s 2 + >
# ee #4 #2 & #1 8 1 & = gg 3
+ че |. + | ee 8 а © 9х | |. |. |, ФФ в Фо ee + 2 + eo + + =
« & о о 8 # ® & в о я. ее а ве т Фея Ч & 2 + = + *
4 + + . + + 4 + + e &« & в я &£ zz + + 2 += eo
о s&s 4 8 + + + + + + + + о + © + в e ое зв ое E +s A + UE
о о «о ео 8 e + ® + + + +8 + + Де | © +2 |. s+ ¢ + + es + ©
a + + + о |, + + + + + в и о т в 6 + ee + +s |. E + 2 + +
о о 6 + 4 ® 8 8 + + + 8 ee о € в о че в о [т ее Фе 4
+ + + + + |. 8 |, |, + € A Ф.Ф т... +
в о + + + 8 а о |. s + + + 8 ¢ & Фе | + ee oo + N
о в e + и о 1 0 #8 + « #2 a + = s s+ « + +2 ee s+ + +s + + 2 2 =
e so & + . ©. & e в ©. Що о ee $ в Фо + + о = . ео ee .о + + + os
Re e AE ea se + + = ss +s + 4 0
се 8 e & e + © 9 + о чи фо #8 в ХФ Ф.Ф. + 2»
в о о 4 о в оф. хо = ® |. ве 8 о ге че & a ev AA
о 6 о + + ® с . в а в о ео в ож оо в ФФ ое U
6. OPERATIONAL CHARACTERISTICS
GECOS System Control . . . . . +. «vv 0 6 604 4 4 08€ 4 6 888065
Object Program Execution
Assignment of IDS Buffers
File Unit Initialization
а о с а ХФ ® + + во 1 + в о). ое. ФФ а #1 о а Фе фе 8
e eo 8 8 во e о о До + я о ¢ + в Фе во + s+ Фо Фе ФФ *
а зе тж ее #8 1 1 1 Що |“ @« + о в о в эф Фо = + +r + ФФ ео в ®
GE-600 SERIES
iv
65
69
67
67
IDS
APPENDIX
A - IDS Record Formats . .. .. . . . + + + + e e » e e e + e « e... 69
B - Reserved Words . . . . . . 24 2 4 4 4 4 +4 . . 6 4 eo oo 4 0 0 0 6 71
C - IDS Error Conditions . . . . . « » « . . e e + e e e. e e e e e . e 73
ILLUSTRATIONS
Figure Page
1 Conventional Record Formats . . . . . RE 4 4 4 4 440 4
2 DSU Layout--Conventional . . . .. 224 0 4 4 4 0 4 0 4 4 4 4 4 4 4 4 0 0 4
3 Record ACCESS . . . . . . . 4 4 6 4 4 4 0 0 4 4 4 4 4 4 0 4 4 4 4 4 4 0 a 0 5
4 IDS Record Formats . . . . 2.22 4 4 4 4 4 04 4 e e e e e e e e e ae 5
о Chaining Example ... 442 4 4 4 44 4 4 0 4 4 0 0 4 4 a e ea e e. 7
6 Disc Surface Configuration . . . . . . . . . . . se e seao re. 9
7 Conventional Disc Records .... 444 4 4 4 4 44 e a e e e ee. 10
8 IDS Page .. 222244 44 . 4 0 4 0 0 0 4 0 0 e e e e en 11
9 Input/Output Controller .. 1.042204 44 4 4 4 4 4 4 4 4 4 0 4 4 1 000 13
10 IDS Record .. . . . . e eme me e. noe ae ae oe ea e. 15
11 Chain Association . . . . 2.211 4 1244 24 4 4 4 4 e eee ec en. 16
12 IDS Chain .. 22204202 2 2 0 4 4 4 4 aa o e e eee e. 18
13 Master Record of Two Chains . . . . . . . . . . e e e e ee e «e e e e. 19
14 Chain Processing . . . 144 14444 4 40 4 4 4 0 ea ee 20
15 IDS Shorthand . . . . . . . . 2 ©... ee ee ee ee eee 22
16 Purchase Order Data Structure . . . 1 212 1 4 4424 4 4 4 4 4 00 23
17 Chain Network ... 2.2 2224 24 4 4 4 4 4 oe eee ae e ea 25
18 Legal IDS Structures . . . . 2.244 14 42444 4 44 4 4 44 0 4 44 0 .. 26
19 Illegal IDS Structure . . . . 2.2 222414444441 4 4 1 4 4 44140 26
20 IDS Compilation Process . . . . 2.222044 4 44 4 4 4444 4 4 4 4 e e. 66
GE-600 SERIES os
1. INTRODUCTION
Integrated Data Store (IDS) is a new information-oriented method of integrating the operating
functions of a business. Its use sharply reduces the high systems and programming costs associ-
ated with implementing an integratedbusiness system. As a general-purpose system, it uses mass
random access storage as an extension of memory and provides an efficient data organization
technique. In addition, a simple but effective language is used to operate the system.
IDS provides a convenient method of describing complex information structures through the
meaningful association of the data record contents. Once the data is described, the system
automatically structures it to suit the hardware requirements of the mass storage device. The
task of organizing data records for meaningful association is handled by the IDS system. This
record association is achieved through the use of chains, which provide cross-reference linkages
between records. These record chains are the integrating force of IDS.
The IDS language provides a simplified means for record processing in the environment nf mass
a — gm wm em He em A - их “я ful
Fand access storage.
Present procedural languages offer programming convenience in field and sequential record
processing. However, they are inadequate for processing records in the random environment of
mass storage.
Language statements such as those for read/write operations produce serial rather than random
actions. The burden of organizing data records and designing the logic involved in processing and
maintaining these records is placed upon the programmer. These same functions are performed
automatically by the IDS software system.
GE-600 SERIES =
2. INTEGRATED SYSTEMS DESIGN
The design of integrated data systems is highly complex. The conventional system is organized
function-by-function, each with a bulky file. Mandatory cross-referencing to show interaction
between functional operations becomes highly intricate and in many cases necessitates carrying
redundant information within the files. Many computer runs are necessary to process these files.
Then, the results must be coordinated and correlated. IDS provides an answer to these problems:
a single file organized around the data rather than the function.
The concept of Integrated Data Store significantly simplifies the design of integrated systems by
permitting the establishment of meaningful associations between data records without any re-
dundancy. A ‘comparison between the conventional approach and the IDS approach to systems
design in the following discussion illustrates the benefits of the latter.
COMPARISON OF SYSTEMS DESIGN APPROACHES
Conventional File Organization
When designing an information system, the designer must resolve many questions. Some of
these are:
¢ What information must be stored in the system ?
e How should this information be stored (volume, sequence, accessibility, etc.) ?
e What cross references must exist between information in the various files?
e How will information be retrieved in response to information processing requirements ?
If, in a conventional approach to file organization, the problem being studied is the design of a
purchasing information system, it may be necessary to plan three duplicate files of purchase
orders:
1. A vendor file with related open purchase orders for general processing.
2. A purchase order file for expediting purposes.
3. A file by inventory item number for production inquiries.
Figure 1 illustrates a conventional approach to file organization using disc storage.
GE-600 SERIES =
RESTA AA AA
VENDOR Vendor Vendor Name ::Purchasel: “Purchase:
RECORD No. 34692 & Address |. Order pi} © Order i
опа ДОННА
(Linkage)
PURCHASE Date Item Qty. Del. Item
ORDER — — @> Placed No. 16 Date No. 2
RECORD
(Linkage)
Ÿ
INVENTORY |Part No.| Qty. / Purchase;
ITEM — ——»| AAA On = Order :
RECORD Hand > #122A |;
} |
Figure 1. Conventional Record Formats
Note that the series of records shown in Figure 1 contains a considerable amount of redundant
data (shaded fields), These records are tvpically stored in separate areas of the file unit with
trailers to separate overflow areas, as shown in Figure 2. (Overflow areas are an extension of
the normal file area. They store the records with the duplicate disc addresses often created when
using randomizing techniques for record storage.)
The systems design and programming effort involved in designing the file layout record content
and data associations, to take full advantage of the hardware capabilities, is both complex and
lengthy.
VENDOR
PURCHASE
ON LON FUN
ORDER DDR Prrehace trder 1<7A
FILE
INVENTORY
ITEM Inventory Item AAA Inventory Item BBB
FILE
MERE OW Vendor Records
AREAS Purchase Order Records
Inventory ltem Records
m
Figure 2. DSU Layout--Conventional
GE-600 SERIES —
The processing and maintenance of records stored, as already shown, is both cumbersome and
time-consuming. Related information is packed in different sections of the file unit, and records
must be individually accessed in each location to complete the maintenance function (Figure 3).
(Access 1
Eg PA Sr =;
(Access 2)
TT
(Access 3)
INVENTORY
ITEM RECORD
———— es
Figure 3. Record Access
If there can be several hundred or several thousand open purchase orders at any one time and
the activity (changes, new orders, receipts. etc) is high, then maintaining these files require
considerable processing time.
IDS File Organization
IDS was designed to relieve the systems programming and storage problems inherent in the
conventional approach to data organization and the subsequent processing of this data.
The IDS record construction comparable to the examples of conventional file organization appear
in Figure 4.
VENDOR a Vendor Vendor Name Link
RECORD No . 434692 & Address Address
ORDER a Purchase Date Link Link
dd
RECORD Order #122A Placed Address A ress
TRC QE
vi Item Qty. Del. Link Link
RECORD No. | 15 Date Address Address
TORY Part No. Qty. Link
— — — — >
RECORD AAA On Hand Address
Figure 4. IDS Record Formats
GE-600 SERIES =
Note there is no redundant information in the above records. Therefore, the purchase order
record, to convey its full meaning, must be properly associated with (1) the vendor record which
describes the vendor with whom the order was placed (2) the item records which describe the
quantity ordered and the delivery date, and (3) the inventory records which described the items
being purchased.
ASSOCIATING RECORDS INTO CHAINS
The chaining feature is the fundamental structuring tool of IDS. Chains are made up of all the
information about a particular function, as in the vendor record in Figure 5. A chain must
contain one master record (the vendor record) and can contain any number of detail records
(other data directly related to the vendor). All the interlocking relationships of pertinent infor-
mation are cross-referenced by the chains. Chains are linked together automatically. A master
record in one chain may be a detail record in another. There is no redundancy in the storing,
processing, or maintaining of data. Figure 5 illustrates the chaining network of logical records
as they might appear in an information system.
With an information system structured in this manner, file interrogation and processing is
conveniently simplified.
In examining the records and chaining associations in Figure 5 note that there are four types
of records: |
Vendor
Purchase order
Item
Inventory
To provide meaning to the system, these logical records are associated in chains:
All the purchase order records--for a vendor
All the item records--for a purchase order
All the order item records--for an inventory item
Within each chain there can be a variable number of records. These records can be linked into
any number of chains. The information system organized in this manner permits interrogation
by all functions of the business with records and data stored only once in IDS.
In the IDS system, it is possible to store related records within the same block on the disc storage
unit. During each disc access a block of information is transferred to memory. Therefore, with
proper data organization, the probability is high that, with a single access to the disc for a
particular record, most of the related information will also be available in the block stored in
memory.
With this feature in mind and by using Figure 5, it is easy to show how the purchasing function
of the business can obtain all information pertinent to vendor status. For example, if information
is requested for vendor number 34692, the processing sequence of Figure 5 is:
1. Retrieve vendor record (vendor code is 34692).
The basic vendor data is now available in memory for processing. (In each vendor
record there is a chain link to the first purchase order record.)
&E-600 SERIES =
d
Y | me TY)
AL
Order Order ON
HIZ2A #20 7А
Purchase
1 Order 4 \
Chain
J | Po N N
SS Np о — \
7 rder
#147A an
{ В Purchase )
; Order
Inventory Item #1 Item Item #3 Inventory
Item CCC Qty. 10 Chain Qty. 25 Item BBB
\ J Item #2 В )
a Qty. 20 N
N_ Item #1 Inventory Item #1 4
Qty. 15 On Order Qty. 5 <
Chain -
| A |
EA | | `` A
Liavellloly
Item AAA
Figure 5. Chaining Example
Retrieve next purchase order record in purchase order chain.
Process data of purchase order number 122A. (In each purchase order record there is
a chain link to the first item record in the item chain and a link to the next purchase
order record in the purchase order chain.)
Retrieve next item record in order item chain.
Process data of item number 1. (The last item record contains a chain link back to the
purchase order chain.)
Processing of purchase order 147A with its item records 1, 2 and 3 occurs in the same
manner.
Following processing of purchase order number 207A, the purchase order chain links
back to the vendor record, which completes the cycle for this vendor.
Composite vendor information is stored and retrieved through these chains, which provide a
meaningful association of related data. Redundancy is eliminated; for example, vendor number
is carried only in the vendor record, which is then associated with its purchase orders through
GE-600 SERIES
IDS
-7-
Flexibility of the IDS system organization can best be illustrated by viewing a second method of
processing the records in Figure 5. The production control functions of the business might,
for example, inquire as to the inventory status, both on hand and scheduled on order, of inventory
item AAA, The processing sequence is as follows:
1. Retrieve inventory item (inventory item number is AAA).
Process the inventory on hand balance. (Each inventory record contains a chain link to
the first item in the on order chain.)
2. Retrieve next item record in the inventory on order chain.
Process item number 1 (quantity 5). (NOTE: The item record is in two chains--the
inventory on order chain and the purchase order item chain.)
3. Repeat step 2 until the last item record in the inventory on order chain is processed
and the chain terminates back, at inventory item AAA.
When the inventory status indicates the need for vendor expediting of an order item, this can be
accomplished simultaneously with the above processing. This is possible because the items in
the inventory on order chain are also chained into the order item chain and in turn into the
purchase order chain. By traversing these chains, all necessary purchase order and vendor
data can be obtained.
GE-600 SERIES —
3. IDS ENVIRONMENT
The reader must have a general knowledge of a disc storage unit (DSU) to understand the over-all
organization and concept of IDS. IDS implementation for the Compatibles/600 computers utilizes
the DS-20 Disc Storage Unit. The following discussion presents some basic information on the
DSU and on IDS record organization.
DS-20
Each DS-20 contains a maximum of 16 circular discs. Both sides of each disc are used, giving
32 recording surfaces. Information is recorded as magnetized spots on concentric tracks.
Each disc surface is divided into an inner zone and an outer zone. There are 128 circular tracks
in each zone, making a total of 256 tracks on each side of each disc. The 128 outer tracks are
+ + ri + A 3 el =
divided into 16 sectors, and the 128 inner tracks are divided into 8 sectors. Figure 6 shows the
format and the manner in which sertors are numbered on disc surfaces.
+
The sector (sometimes called a “frame” or “record”) is the basic, physically addressable unit
on a disc and consists of 240 characters for the GE-600 Series.
"ROTATION
y
12 6
16 Discs
7
8
Inner Zone ——7 >< DD Outer Zone
128 tracks 128 tracks
8 sectors/track 16 sectors/track
Figure 6. Disc Surface Configuration
GE-600 SERIES —
Read/Write System
Each disc is served by a separate movable positioning arm. Each arm contains eight read/write
heads, four for the upper surface of the disc and four for the lower.
1 1 1 ух ы + ra “a » T fay ada fa BENE Ea CEA A + = 2
essable within cach individual file unit. Up to 32 contiguous sectors
can be read or written with one instruction from the central processor. With the arm in one
position, it is possible to transfer a total of 96 sectors.
Sectors are continuously addr
Pos wr Nr NSE ENA MASAMI Y
For each sector on the disc for information storage, there is permanently recorded a sector
(record) address. Whena searchis made to locate the correct sector for a read or write operation,
the physical address and the address contained in the instruction are compared. When the two
addresses are confirmed electronically as identical, the addressed sector is read or written.
The disc address tells where a record is physically located in the file.
Logical records are those records designated by the systems designer/programmer--for example,
inventory record or payroll record. They contain data fields pertaining to a specific segment of
the application system. Logical records are composed of data and chain fields. The total com-
ined length of a logical record may be any length up to the capacity of a data block (IDS page)
which is typically 8-16 sectors.
Conventional disc organizations (not IDS) specify one or more logical records to be contained in a
sector, or one or more full or partial sectors to be equal to a logical record. Examples of
conventional disc records are shown in Figure 7.
-———— DISC SECTOR (240 CHARACTERS) —— — —]
RECORD
#1 y
MORE THAN ONE
SECTOR PER
RECORD
LESS THAN ONE
RECORD SECTOR PER
#1 RECORD
RECORD RECORD ONE OR MORE RECORDS
A 47 COMPLETELY FILLING
ONE SECTOR
Figure 7. Conventional Disc Records
GE-600 SERIES —
-10-
Note that where equality, or multiples thereof, does not exist (normally the case), inefficiencies
in disc storage utilization result.
IDS Pages
In the IDS system a page (or block) of records consists of a fixed number of disc sectors as
assicned by the program implementor. A page may contain any combination of logical record
77-07
TA a] ts 1 A + 1 +
es linked into their respective chains, Each type of record has its own specific length.
Related record types are associated and linked according to their data content and may be stored
within the same page. Space is fully utilized by the automatic packing of these records within
the page. During processing, whole pages are read into memory. Therefore, with proper data
organization, the probability is high that much of the related information will be available in a
single file access. Figure 8 is an example of an IDS page.
=
Page Header Record | Personnel Record |
Inventory Record inventory Rec-
ord I Vendor Record Order Item
Record Order Item Record | Vendor
Record
EE
GC ии ии и”,
Figure 8. IDS Page
ZZ
The organization of the “page” facilitates file maintenance and processing, as well as making it
possible to accommodate a variable number of record types without duplication of information.
Every page begins with a unique Page “Header” record. This record contains several control
fields used by the system, as follows:
1. Reference address of the page (page number)
2. Space available in the page for additional records
3. /O control indicating whether page has been altered since retrieval
4
Chain field indicating address of the first record of a chain of calculated records, all
of which are randomized to this page
5. Line numbers available for assignment within the page
Records within the page contain the following control fields in addition to their user-specified
data fields:
1. Line number
2. Record type
3. Record length
GE-600 SERIES —
-11-
Reference Code
The reference code is a relative addressing code permanently assigned to each record. Tt can be
considered to be a logical address in contrast to an address that defines where the record is
physically located in the file. Once a record is assigned a reference code, it maintains that
reference code regardless of expansion, contraction, segmentation, or other basic reorganization
of the record or file. Thus, the reference code can be used for direct record retrieval on a
long-term basis.
The reference code consists of the page number and the specific line number of a data record
contained within the page as explained below:
1. The page number portion, a sequential number permanently assigned to each page, which
defines what proportion of the way through the IDS page environment the page is stored.
Each record stored within a page shares the same page number as part of its reference
code. Page numbers are converted to actual disc addresses by the IDS mapping routine
at execution time.
2. The line number portion, is a numbering system of records within each page. As
records are stored in a page, an available line number (line numbers are reusedas
records are deleted) is assigned to the record and is combined with the page number
to complete the reference code.
INPUT/OUTPUT CONTROLLER
The Input/Output Controller of the IDS controls the mass storage device. Its major function is
to control the flow of pages of records in and out of memory in response to commands to store,
retrieve, modify and delete specific data records. It also controls the page processing function
within memory. Figure 9 illustrates the functions of the Input/Output Controller.
Data Buffers
To minimize the mass storage seek and transfer time, an inventory of data pages is maintained
in memory. These pages are stored in numerous buffers in memory. The number of buffers
depends on the amount of space available after the IDS subroutines and the problem-solving
routines have been loaded.
The greater the number of data pages stored in core memory, the greater the possibility that the
one needed next will already be there. To further improve the possibility of finding the page
desired in memory, the Input/Output Controller keeps track of the sequence of page utilization
(record activity) and holds the most recently active pages in memory. Pages which are infrequently
accessed are retired from memory as others are called in. The Input/Output Controller notes
which pages have been modified and writes only the modified pages back to mass storage.
Priority Control
Each time a new page is brought into memory, its page number buffer location is placed at the
head of a page list. If a page already in memory is used again, it moves to the head of the list.
Thus, this list tends to hold the most frequently used pages at the top of the list and the pages
with little or no recent use at the bottom of the list. Page space is maintained to allow for the
input of a new page. Following input, the page at the bottom of the new page list is automatically
written back to the disc storage file (providing there has been activity updating that page).
GE-600 SERIES —
-12-
CORE MEMORY
WORKING STORAGE
EN
>]
IDS
PAGE <
BUFFERS
IDS PAGE
Figure 9. Input/Output Controller
Record Unpacking
Once the Input/Output Controller finds a place for the page in memory, it locates the record
called for in the page. The fields from the record will be unpacked into Working-Storage if a
MOVE TO WORKING -STORAGE command is specified.
File Protection
The automatic control provided by the Input/Output Controller for bringing records into memory,
writing from memory to disc storages and making the record available for the programmer in
the Working-Storage area, eliminates to a large extent the possibility of erroneous updating of
record fields. File integrity is therefore maintained through the elimination of record maintenance
by the programmer and by restricting the programmer to the updating of record fields in the
Working-Storage area only. The modify function, which can address data fields, provides the only
way of changing an actual record.
GE-600 SERIES —
-13-
Data organization refers to the interrecord relationships established within the IDS. The record
is the basic unit of data. Record association is achieved through chains which provide cross-
reference linkages between records. These chains provide the integrating force which is implied
in the name Integrated Data Store.
As shown below, the IDS record contains a set of “data” fields which collectively describe the
event, activity, status, or plan that the record represents. IDS augments these records with
additional fields called “identification” and “chain” fields, as shown in Figure 10.
PA A
N о J A
iudentiiílccar vu X Valid Fivius
Field Chain Fields
\\
Figure 10. IDS Record
The chain fields contain the reference codes of other IDS records. They point from one record
to the next, creating a circular association of records (Figure 11).
These associations are automatically processed according to the data descriptions and the pro-
cedural commands executed. The arrows in Figure 11 indicate the linking actually carried out
through the storing of the reference code of one record in the body of the prior record.
DATA RECORDS AND FIELDS
[ast
he
The records of the IDS are fixed-format, fixed-length records in the COBOL tradition; that is, the
length and format of a specific type of record, such as payroll or inventory record, are fixed by
the specifications of the systems designer. Records of many different types, each with its own
length and format may be used in the system. In order that control may be maintained, each
record has the same identification fields at the very beginning. These fields are (1) line number
portion of the reference code, (2) record type (such as inventory, payroll, etc.), and (3) record
length. The rest of the record consists of data and chain fields to suit the application requirements.
GE-600 SERIES —
-15-
Z—
Next Next
Record In Record In
4176 Chain X #849 Chain X
147 176
Chain X
Next
Record In
#147 Chain X
849
Figure 11. Chain Association
Records may have any number of data fields, each defined as some number of decimal, alphabetic
or alphanumeric characters. Fields may vary in size from one character up to many characters,
as for a drawing or part number or an employee’s name. These fields will be specified by the
systems designer.
Chain fields contain the reference code for addressing and are defined for each chain in which a
record participates. Experience in IDS systems indicates that the average record is in only two
chains. An occasional pivotal record in the information integration may be in six or eight chains.
There is no upper limit on the number of data or chain fields except that provided by the maximum
page size. Average record size has been 30-40 characters in total length, with an occasional record
of 120-180 characters.
IDS records are stored only once inIDS. Conventional approaches to file organization often require
records, or certain fields in the records, to be repeated in several files. With the ability to link
records into any number of chains (as required by the system), the same data fields are available
for all chains processed. This technique of eliminating redundant data has four important
advantages:
1. Additional space required for duplicate records is eliminated, resulting in a reduction in
the total capacity required.
2. The work of data maintenance is greatly reduced, as there is only one record to retrieve
and modify.
3. The possibility that one of the copies of a record will not be properly modified is
eliminated. Since there is only one copy of a record, any incorrect information will be
quickly recognized and corrected.
4, All reports drawn from the file will be consistent, since there is only one set of facts
(records).
GE-600 SERIES —
-16-
Record Classes
trieve. The designation of the data records as to class is at the option of the systems design
and is based on the storage and retrieval requirements of these data records.
IDS record processing requires that there be some aspect of every record which makes it unique,
i of i e
e
or different from any other record. All records are unique by virtue of their reference code
Some records are also unique because they contain one or more data fields--such as a drawing
+ =
AL a
number, order number, and pay number--where no duplicate values are allowed.
CALCULATED RECORDS. Any record within the system can be classified as a “calculated”
record. Its storage and retrieval are based upon the contents of one or more data fields. The
contents of these fields are externally specified numbers--such as employee numbers, part
numbers, or order numbers. The contents of these fields are processed through a randomizing
procedure which determines a page number foraninitial storage location. The record is stored at
this page or one very close to it. A purchase order or an inventory item record is typically de-
fined as this type of record. The subsequent retrieval of this record follows this same basic
procedure.
SECONDARY RECORDS. “Secondary” records are a class made unique by virtue of their chain
relationship to a specified type of master record and a sort control field used to sequence that
chain. The item records associated with a purchase order (master) record are a good example of
secondary order records (refer to Figure 5). These records are stored and retrieved by first
locating the mirchase order record and thon stepping through the order item chain tu lucate the
tell record vy comparison ot its item number field.
PRIMARY RECORDS. Records designated in the data description as “primary” are unique only
as a result of their reference codes. Generally primary records are used in place of calculated
records where the external assignment of identification fields, such as part number or order num -
ber, is not required. In these cases, an internally generated number (the reference code) is
assigned and used as the identification field.
IDS CHAINS
An IDS chain is illustrated in Figure 12. Its characteristics are:
® One master record and any number of detail records
e Links records together in an endless loop
e Associates related records in meaningful sequences
r Sig s to class for storage and retrieval purposes, they must
also be defined as to their relationship within a chain--master or detail records. These record
r ecifi
This master-detail relationship of records is analogous to the conventional header-trailer file
sequence of records. The master record of the chain describes the fixed information (header
type) pertaining to the variable number of detail records in the chain. The detail records in the
chain describe the variable information pertaining to the master record.
GE-600 SERIES —
-17-
MASTER
Detail 1 Detail n
Detail 2
Figure 12. IDS Chain
MULTIPLE CHAINS
Chains exist for two separate but closely related reasons. First is the case where the source
documentation shows that a portion of the information appears in multiples--for example, a per-
sonnel record with a variable number of deductions and work expe riences.
This type of information is easily structured by building a personnel master record. Two chains
are created containing the personnel record as the master record. As many deduction records as
necessary are linked into the deduction chain as details. Work experience for the employee in-
volved is handledina like manner. Both chains are now linked to the same master record as shown
in Figure 13.
The second case for chain structuring of information involves the association of several related
source documents. Relating all the purchase orders for a given vendor is an example. Figure 5
(page 7) illustrates this example. The purchase order chain associates all of the purchase order
records with their vendor record.
IDS chains have several structural aspects which should be emphasized:
A chain type is named with a symbolic name. There will be as many chains of the chain type
as there are master records for that chain type.
Each chain has only one master record. The record type of the master record is the same
for all chains of the same type.
A chain is created whenever its master record is stored in the IDS. The chain is destroyed
when the master record is deleted.
GE-600 SERIES _
-18-
A chain may contain more than one type of detail record. Any number of detail records may
be in a chain.
Detail records cannot be storedunlessa master record exists which qualifies as the particular
master according to the specified master selection rule for that chain.
Whenever a master record is deleted, all of its detail records are automatically deleted (and
their detail records too).
Work
Exper-
ience #1
Deduction
#n
Work Experience
Chain
Deduction
Chain
Personnel
Record
Work Deduct i
Exper- MASTER e Lo ion
ience %2
or . 7
peer ta | a |
Figure 13. Master Record of Two Chains
All records in a chain are associated in an endless loop with the last detail chained back to
the master.
The master record of the chain stores the reference code of the first detail in the chain.
A record (detail or master) may be defined to be in as many chains as are required. It
may be defined as master in one chain and detail in another.
A record cannot be defined as a detail to itself directly or indirectly.
As records are stored in the system, they are automatically linked into their defined chains.
When a record is deleted, the chains in which it is a detail record are automatically patched
to relink around the deleted record.
GE-600 SERIES —
-19-
CHAIN PROCESSING
IDS offers complete flexibility inthe retrieval of records within a chain by providing three methods
of chain processing. These methods are illustrated in Figure 14.
MASTER
NL >
И
A NN
N |P |M N |P IM
e |r ta el|r ia
Detail #1 х |+ |5 Detail #n x (ils
о | © о | Е
t r е t r e
M
N Ё a
e . 5
O
t |r|£
Figure 14. Chain Processing
1. Chain NEXT--The definition of a record in a chain automatically provides the record
with a chain-next field. This is the manner in which all chains are constructed. Each
record contains a chain-next field which contains the reference code of the next record
in the chain.
Chain PRIOR (optional)--IDS provides a chain-prior field for all records in a chain when
the chain is specified by the systems designer as prior processable. This field contains
the reference code of the prior record in the chain. This permits the chain to be pro-
cessed efficiently in a backward direction.
Chain MASTER (optional)--IDS provides a chain-master field for all detail records in a
chain when specified in the data description. This field contains the reference code of
the master record of the chain. Retrieval of the master record is much faster with this
ability to address the master record directly from any detail in the chain. Processing
need not access all the detail records in the process of seeking the master.
LINKING A NEW DETAIL RECORD
In order to insert a new detail record in a chain, two steps are required:
The appropriate master and its chain must be selected.
The record must be inserted in that chain according to the chain ordering rules.
GE-600 SERIES —
-20-
MASTER RECORD SELECTION
There are two rules under which the master record is selected for a new detail record. These are:
e Select Unique Master--This rule uses the record retrieval criteria, established in the
data definitinn far the masctear пал ta retr iaova the narticular master record indi cated
WALL LA чл” В АДА МАЗИ de JJ A Piu” ALLA Vu A de Nr e Ad “ty LAS de Vr E de de A Y Uv LAS pa Ce de LL LE A 64 A + Ed CA th d de Ww A Ad SA AA
by the data values currently stored in the match control field of Working-Storage.
e Select Current Master--This rule uses the last record processed, of the master record
tira na tha maaoctar nannanvd rf tha max: datai]l
Ly MY , AD Lid Ta А с LE À А GUAL LA WL LAU LIU VY MU LCALA.
CHAIN ORDERING
All chains in the Integrated Data Store system are ordered in one of six methods specified by the
systems designer with the chain-order clause in the IDS language.
The chain-order clause must be used in each Master Chain Definition entry.
The six options of the chain-order clause are:
1. Sorted Within Type --With this option the records of the chain are maintained in sequence
within record type, independent of other types.
2. Dorted--With this option the various records ot the chain are maintained in a single
sequence regardless of the number of record types in the chain. With this option the
control fields of the various records must be of identical size.
NOTE: When either of the sorted options are specified, details are added to the chain
based upon the content of the defined sort control fields of the detail records.
3. First--This option causes the detail to be added as the first detail record in the chain
relative to the master record.
4. Last--This option causes the detail to be added as the last detail record in the chain
relative to the master record.
о. Before--This option causes the insertion of the detail record just before the current
record in the chain. This option may be used only in conjunction with the “Current
Master” selection rule.
6. After--This option causes the insertion of the detail record just after the current record
of the chain. This option may be used only in conjunction with the “Current Master”
selection rule.
PRIME CHAIN
Access time in present disc-type random access memories varies greatly, since it depends on the
position of the desired record relative to the record last accessed. The IDS organization of
records acknowledges this factor of hardware design and stores new detail records as close as
possible to the master record of the chain. When a detail record is specified as a detail in several
GE-600 SERIES —
-21-
chains, a prime chain may be chosen and defined by the systems designer preparing the data
description. Selection of a prime chain shouldbe based upon an estimate of the most active chain.
Thereafter, when an IDS page is retrieved which contains the master record of a prime chain, it is
highly probable that the detail records of that chain will also be contained in that page or a page
closely associated with it. The prime chain is the chain used to retrieve a secondary record by
the RETRIEVE command, unless specified otherwise by the data description.
DATA STRUCTURE
A special graphic technique called “IDS shorthand” has been developed to display records and their
master-detail (chain) relationships.
Its use is particularly important in developing an over-all view when planning an information
system. This technique (see Figure 15) uses a block shape to designate a record type--employee
(record type 1) and deduction (record type 2)--and an arrow connecting two blocks to designate a
chain. The arrow points from the master to the detail, as shown in Figure 15,
(IDS Shorthand)
Employee
(master record)
Deduction Chain
Y
Deduction
(detaíl record)
(Expanded Version)
Employee
(master) a
Deduction #1 , Deduction #n
, (Deduction) ;
(detail) CHAIN (detail)
Deduction #2
> (detail)
Figure 15. IDS Shorthand
GE-600 SERIES -
-22-
In the foregoing example of IDS shorthand, the vertical block-arrow-block sequence carries the
following message:
i. There are a number of records in the system of the master type (one for each employee.
2. Each of these records is the master of a chain of the specified type (deduction).
3. There are a number of records of the detail type (deductions 1, 2, 3, 4, etc.) in each
such chain.
The purchase order data structure (Figure 16) shows how a vendor record and a particular
order record from the example shown in Figure 17is normally structured in the IDS system.
IDS SHORTHAND
Vendor
Record
VENDOR 34692
Purchase Order Chain
Pc SEEN A AIL JESS 2 TEES A A awe
ORDER 147A y
С — |! Purchase Inventory
Order Item
Record Record
ITEM 1
ITEM 2
ITEM 3 Order Item Chain
Inventory on Order Chain
LT
Order
Item
Record
Figure 16. Purchase Order Data Structure
The purchase order contains four groups of information.
1. Information about the vendor--such as his name, address, and vendor code.
2. Information about the order--such as the order number, due date, mode of transportation,
and dollar value.
3. Information about theorder item--such as delivery date, quantity, unit price, and extended
dollar value.
4, Information about the inventory item--such as its identification and description.
GE-600 SERIES —
-93-
The data structure in Figure 16 shows all four groups and their chain associations with only four
blocks and three arrows. To expand this structure, four different record types would be designed
to carry the information contained in the four groups:
Vendor record--There would be a vendor record for every vendor with whom the business is
concerned:
1. It would be the master record of a purchase order chain
2. Thus, the vendor record is only a master.
Purchase order record--There would be an order record for each order currently stored in
the system:
1. It would be a detail in the purchase order chain
2. Each order, in turn, would be the master of an order item chain
3. Thus, the purchase order record is both a master and a detail.
Order item record--There would be an order item record for each item on each order.
1. It would be a detail in the inventory on order chain.
2. It would be a detail in the order item chain
3. Thus, the order item record is a detail in two chains.
Inventory item record--There would be an inventory recordfor each inventory item currently
stored in the system.
1. It would be the master record of the inventory on order chain.
The expanded data structure for the above records is shown in Figure 17.
Figure 16 showed the same data structure in IDS shorthand.
&E-600 SERIES —
-24-
—
CY)
all
(
7
Inventory
Item CCC
\
Figure 17. Chain Network
SUMMARY OF DATA STRUCTURES
Order Order ON
#122A #207A
— Purchase Ya
Ä | Order A |
| Chain J NN
Order
#147A
N | Purchase /
Order
Item #1 Item Item #3 Inventory
Qty. 10 Chain Qty. 25 Item BBB
J Ж Item #2 В
Inventory |)
JF т
teem 7 On Order rem A et
NT Chain tren |
| |
A AN J
Inventory |
Item AAA
By using IDS shorthand, very complex data structures may be presented in a condensed and under-
standable form. Thus, the following examples (Figures 18 and 19) show a quick summary of data
structures which are legal and illegal within IDS.
NOTE: A circular definition, as shown in Figure 19, is not allowed.
GE-600 SERIES-
IDS
5
E
|
a
Figure 18. Legal IDS Structures
GE-600 SERIES
|
NV
Figure 19. Illegal IDS Structure
5. IDS PROGRAMMING LANGUAGE
SOURCE LANGUAGE
Introduction
The source language of IDS is an extension of COBOL as implemented for the GE-600 Series
product line. Therefore, all formats and language specifications of COBOL must be adhered to
when preparing a source program. In a few cases, certain elements of the COBOL language are
not applicable to the IDS usage. These exceptions are mentioned in the following sections on the
IDS language.
identification Division
The purpose and usage of the Identification Division are identical with those defined for COBOL,
with no speciai function for {s.
Environment Division
All portions of the Environment Division, except the File-Control paragraph of the Input-Output
Section, are used as defined by COBOL.
To define the file name which represents the IDS data file, a unique version of the SELECT
sentence is required as shown below:
NOTES:
Ca
FILE-CONTROL. SELECT IDS file-name
ASSIGN TO file-code-1.
The SELECT IDS sentence must be used only once to identify the IDS data file.
Other optional phrases of the SELECT sentence as specified for COBOL should not be
used with the SELECT IDS sentence.
Thao filo
11.18
Ada mot be a He nhanrnantan NAPA Ann ol sting nf Hwun lattare (A 7\ Ar a lattar
1 ii VUC па CT LW ra Dio e i
u a U-CN Cter Wora con ing OI TWO 1€T1€EIS \a,...., 4; OF ¿CTE
and a digit (0,....,9). Each file code must be unique with respect to other file codes in
the program. At execution time, the object program is submitted to the General Com-
prehensive Operating Supervisor (GECOS) with “file cards” specifying the peripheral
device for each file. The file code in the file card must be the same as that assigned in
the source program. GECOS associates each of the object program’s files with its
peripheral device by matching the file codes. (See the GE-525/635 Comprehensive
Operating Supervisor Reference Manual, CPB-1002.)
n
vu
GE-600 SERIES —
-27-
IDS FILE DESCRIPTICN
ENTRY
4. When multiple file units are used, an additional restriction is placed on the file code, if
an IDS data file is physically stored on two or more file units. In this case, the allocator
of GECOS assigns the file code specified by the file card to the first unit and a file code
which is one greater to the second unit. etc. For example, if an IDS file has been
physically stored on two file units and the file code of AA has been assigned, the first
unit must be referenced using file code AA and the second unit, using file code AB. The
user must be sure that any other files referencedby his program do not use a file code
which may conflict with this procedure.
Data Division
The description of the IDS data file is contained in a special section of the Data Division called the
IDS Section. This section must physically follow the Working-Storage Section, if present, and
precede the Constant Section.
The IDS Section contains a File Description, Record Description, and Chain Definition as required
to describe the complete data file.
FILE DESCRIPTION. The File Description entry provides information regarding the physical
characteristics of the IDS data file. The entry is used only for documentation purposes, since the
Input/Output Controller module of IDS expects to find these same characteristics stored as a
special Environment record within the disc storage unit. (See Appendix A.) As this implies, the
disc storage unit must be preconditioned with the Page Header records and the Environment
record prior to the execution of any IDS program.
The entry consists of a level indicator, a file name, and a series of clauses which define the
physical characteristics of the IDS file. The mnemonic level indicator MD is used to identify
the start of the File Description entry and to distinguish it from the Record Description entries
which will follow. The format for the complete IDS File Description entry follows.
IDS File Description Entry
FUNCTION: To document information concerning the physical structure of the IDS file.
FORMAT: MD file-name [: PAGE CONTAINS integer-1 CHARACTERS |
[: FILE CONTAINS integer-2 PAGES |.
NOTES:
1. The PAGE size (integer-1) specified may be any value up to a maximum of 4096 characters
as determined by the particular IDS application. However, the most efficient use of the
storage capacity of Mass Storage Device involved should be considered when establishing
the page size. When the DS-20 is the device involved, any size other than 240, 480, 960,
1920, or 3840 characters will result in wasted space on the device due to the addressing
characteristics of the DS-20.
2. The FILE clause expresses the total physical storage requirements of the IDS file. This
value must be equivalent to orless than the capacity which has been reserved for the file
by the allocation procedure of GECOS. When storage is allocated by GECOS, it is reserved
in increments of 23,040 characters. This increment of storage is referred to as a “link.”
see the GECOS manual for a discussion of the allocation of permanent random dise or
drum files. The maximum number of pages possible within the IDS page numbering system
is 262,144 (218),
GE-600 SERIES es
-98-
RECORD DESCRIPTION. In general, the format and usage of Record Descriptions required for
IDS are the same as those for COBOL Record Descriptions. The exceptions to the normal usage
are described below.
The Record Description entries perform three functions:
1. Provide information to IDS regarding the format of each logical record type as it will
exist within a page on the external storage device.
2. Define the internal Working -Storage areas which areusedto pass data between the user's
program and the mass storage device.
3. Provide parameters which will affect the procedures used to store and retrieve data
records.
The external format of an IDS record consists of control fields and data fields, as shown in
Appendix A, Only the data fields are defined by the Record Descriptions supplied by the IDS
Section, The first character of the set of data fields of a record is always the first character
following the last chain field, This means that computer word orientation is never applied to
external data formats, SYNCHRONIZED, as used in the normal COBOL sense, does not apply
to the IDS record formats; if used, it will not affect external formats, In defining the external
format of the record, only the level 02 Record Description entries are considered by the IDS
Translator, The SIZE and CLASS clauses are the significant elements of the description; how-
ever, any of the standard COBOL: clauses may be used with the following exceptions: (1) The
OCCURS, RENAMES and editing clauses are not applicable to IDS and should not be used at the
UZ level within the [DS Section (2) The COPY clause as specified by COBOL should not be used.
A special version of the COPY clause for use with IDS is defined in this manual in the discussion
of the unique IDS Record Description clauses,
In addition to defining the external format of the IDS record, the Record Description entries
define the Working-Storage areas which will serve as the communication interface between the
user’s routine and the IDS data file. As a record is retrieved from the storage device, the user can
make various data fields of the record available only by causing the record to be moved to
Working -Storage. The same process is used when a record is stored. The user must first have
initialized Working-Storage with the data fields of the record to be stored.
The IDS Translator causes unique Working-Storage areas to be established for each level 02
entry processed in the IDS Section. These areas are always created so that the first character
of the area will begin in character position zero of a computer word. The area created for a given
02 entry may contain subfields which are defined by any number of lower level entries; these
entries may be separately referenced by the user’s COBOL procedure. The IDS operations,
however, can operate only on units of data represented by level 02 Record Description entries.
Therefore, any field that is to serve as a control field or any field that may be modified by
IDS must be defined as a level 02 entry.
GE-600 SERIES —
-29-
Lower level entries (03-49) may be used to define subfields of the 02 entry with no restrictions.
Any legal COBOL clause may be used, as long as it does not contradict the description provided
for the 02 entry. For a further clarification of the concept of levels of data description, see the
GE-625/635 COBOL Reference Manual, CPB-1007.
Two of the standard COBOL record description clauses allowable at the 02 level do not cause
the generation of Working Storage areas. These clauses are the REDEFINES and FILLER clauses.
The REDEFINES clause may be used for its normal purpose of redefinition of an area previously
defined. This enables the COBOL procedural statements to reference the Working Storage area
by either of its definitions. The field-oriented functions of IDS, (MOVE, MODIFY), however, will
respond only to the original definition of the field.
The use of FILLER as a data-name at the 02 level causes the IDS Translator to ignore the entry
after taking note of the size of the area involved. No Working Storage is created and that portion
of the external format is not available to the user’s program. This technique can be used to
advantage when the use of core memory is critical.
An additional restriction imposed by the IDS Translator prohibits the use of qualification of those
data names at the 02 level of Record Description. Qualification of lower level entries up to the
02 level is permissible. If the same data name occurs as a 02 entry for different record types,
the same Working Storage area will be shared by the various records involved.
The third function of the Record Descriptions is to define certain special IDS characteristics of
each record type of the data. These special characteristics are defined at the 01 level and are
described in detail on the following pages.
GE-600 SERIES —
-30-
IDS RECORD DESCRIPTICN
ENTRY
IDS Record Description Entry
FUNCTION: To specify the additional characteristics unique to IDS, which may be used to define
data records.
01 data-name-1 COPY FROM LIBRARY
FORMAT Option 2.
01 data-name-1 TYPE IS integer-1
rn \
data-name-2 FIELD
‘RETRIEVAL VIA \ data-name-3 )
CHAIN
CALC
N J
‘PLACE NEAR data-name-4 CHAN |
¡"PAGE -RANGE IS integer-2 TO integer-3 |
INTERVAL IS integer-4 PAGES |
AUTHORITY IS integer-5 |
NOTES:
1. Eachof the above clauses is applicable only at the record (01) level.
2. Data-name-1 must be unique since qualification by file name is not meaningful.
3. All format considerations are as specified for COBOL.
4. The option 1 entry is used only when the Record Description entry is to be copied from a
library source. This library source, which is actually a file assigned to file code “,L”,
is searched for the level 01 entry identified by data-name-1.
When the 01 entry is found, the Record Description entry and all of its associated entries
(level 02-49 and 98) will be copied from the library. This option enables the user to
include only those portions of the total data description required for the current appli-
cation. The user must be sure, however, that all records which are to be referenced by
the procedure either directly or indirectly have been defined.
GE-600 SERIES —
-31-
TYPE
Type
FUNCTION: To define the Record Type code to be used for reference purposes for each record
tvpe within IDS.
FORMAT: TYPE IS integer-1
NOTES:
1. This clause is required for each level 01 entry.
2. Integer-1 may be any value from 1 to 999.
GE-600 SERIES -
-39-
RETRIEVAL
Retrieval
; De
RETRIEVE data-name” form of th
Division)
FORMAT:
NOTES :
``
data-name-2 FIELD
RETRIEVAL VIA < data-name-3
CALC
This clause is required for each level 01 entry.
The first of the three variations of this clause (RETRIEVAL VIA data-name-2) provides
for records whose reference codes enable them to be retrieved directly. Data-name-2
refers to a field which is defined for this record. The user must supply the IDS generated
reference code to retrieve a record defined in this manner. This is done by initializing
the contents of the data-name-2 field in Working -Storage. In this case, data-name-2
is the field that is equivalent to the reference code, because it is not contained in the
data record as defined but instead is contained as the record’s 24-bit reference code.
(See Appendix A for the reference code format.) During the retrieval process, the
supplied contents of data-name-2 are converted to a 24-bit binary integer and the record
corresponding to that value is retrieved. These special considerations associated with
the field names require that the field be defined as an 8-digit numeric value (PICTURE IS
9(8) SYNCHRONIZED LEFT.)
The second of the three variations of this clause (RETRIEVAL VIA data-name-3 CHAIN)
provides for the retrieval of a record which is a detail in the chain named. The master
of the chain must first be retrieved and then, by searching the chain, the specific record
may be found. Data-name-3 must refer to a defined chain name.
The third of the three variations of this clause (RETRIEVAL VIA CALC CHAIN) pro-
vides for the retrieval of the record through the use of its defined control field to pro-
duce a random number. This random number is equated to a specific page of the data
file. The Page Header record is retrieved and the program finds the specific record by
searching the CALC chain whose master is the Page Header record.
These three RETRIEVAL procedures provide a basis for classification of each record
as one of the following:
Primary - Retrieved directly via reference code
Secondary - Retrieved via its chain association
Calculated - Randomized to the page containing the chain which leads to the record.
Subsequent discussions of IDS will refer to records using these terms.
GE-600 SERIES —
-33-
PLACE
Place
FUNCTION: To specify a “special case” procedure to be used when this record is to be stored
in the mass storage device.
FORMAT: | Lace NEAR data-name-4 cham)
NOTES:
1. Data-name-4 must be a defined chain name and this record must be a detail in that chain.
2. When the record has been defined as a primary or secondary record by the RETRIEVAL
clause, the record will be stored physically near its logical position in the chain named.
3. This clause is not meaningful when the record has been defined as a calculated record
by the RETRIEVAL clause.
4,
When this clause is omitted, primary records will simply be stored in a convenient
location. A secondary record will be stored near its logical insert point in the chain
defined by the RETRIEVAL clause.
GE-600 SERIES —
-34-
PAGE-RANGE
Page-Range
FUNCTION: To provide a method of partitioning the various records of IDS within the total
environment.
FORMAT: |.PAGE-RANGE IS integer-2 TO integer -3 |
u J
NOTES:
1. Integer-2 and integer-3 represent the first and last page numbers of a series of pages
within which records of this type are tobe stored. Integer-2 may be of greater magnitude
than integer-3. In this case, the range is defined as beginning with the integer-2 page
and extending to the last page of the total environment: continuing with page 1 and
extending through the integer-3 page.
2. The page numbers specified must fall within the total number of pages specified for the
file by the FILE clause of the File Description entry.
3. This clause supersedes the storage procedures specified by the PLACE clause.
GE-600 SERIES —
-35-
INTERVAL
Interval
FUNCTION: To provide a method to ensure a uniform distribution of a given type of records
across the total TDS environment.
FORMAT: |INTERVAL IS integer-4 PAGES |
NOTES:
1. Integer-4 represents the number of pages which will be skipped when a record is stored
relative to the previous record processed of the same type.
2. This clause supersedes the storage procedures specified by the PLACE clause.
3. This clause is only applicable to primary or secondary records as defined by the
RETRIEVAL clause.
4. This clause will normally be used only for the initial loading of the file unit.
GE-600 SERIES =
-36-
AUTHORITY
Authority
FUNCTION: To provide a method to safeguard the data contained in a record from unauthorized
reference or modification.
FORMAT | AUTHORITY IS integer-5 |
| ot
NOTES:
1. Integer-5 may be any value not exceeding 4095(,,). The value supplied is used as a
“lock” for the data contained in any record of this type. Whenever this record is
referred to during execution, a “key” must have been supplied which matches the lock.
The key is supplied by the OPEN statement as defined in the description of the Procedure
Division.
GE-600 SERIES —
-37-
COMPLETE CHAIN DEFINITION
ENTRY SKELETON
CHAIN DEFINITION. As has been mentioned before, a record normally belongs to at least
one and possibly many chains. A Chain Definition entry must exist for each such chain. The
Chain Definition entries for a given record must immediately follow the Record Description
entries for that record.
The Chain Definition entry consists of a level indicator (98), a chain name, and a series of clauses
which define the characteristics of the chain. The complete Chain Definition entry skeleton is
shown below followed by a detailed description of the clauses.
Complete Chain Definition Entry Skeleton
FUNCTION: To specify the status of the record as either a master or a detail of a chain, and to
specify the characteristics of the chain. ,
FORMAT Option 1.
98 data-name-1 CHAIN MASTER
(SORTED WITHIN TYPE )
SORTED
FIRST
¡CHAIN -ORDER IS >
— LAST
BEFORE
AFTER
[;LINKED TO PRIOR ] .
FORMAT Option 2.
data-name-1
98 CHAIN DETAIL
CALC
| RANDOMIZE ON date-name-2 [ranvourze.…] |
fr
T
ARE FIRST
DUPLICATES ARE LAST
NOT ALLOWED
GE-000 SERIES =
- 38
COMPLETE CHAIN DEFINITION
ENTRY SKELETON (cont.)
| (ASCENDING \ I f ASCENDING \-1 |
€ KEY IS data-name-3 : > |
| \ DESCENDING || DESCENDING J J |
= „”
-
[ASCENDING RANGE KEY IS data-name-3 |
UNIQUE |
SELECT MASTER
CURRENT |
:MA TCH-KEY IS data-name-4 L maton ey. J]
—
‘SYNONYM data-name-5 EQUALS data-name-4 |
ht
LINKED TO MASTER |
hs
NOTES:
1. Level-number must be 98.
Z. Lata-name-1 must be unique, since as a chain name it may not be quaiiliea.
3. Either option 1 or option 2 must be used for each chain in which the currently defined
record is a part. The clauses shown for option 1 may be used only when the record is the
master of a chain. Only one master record may be defined per chain. The clauses shown
for option 2 may only be used when the record is a detail of a chain. Any number of record
types may be defined as detail records of a chain.
4. An option 2 entry must be used whenever the record has been defined as a calculated
record by the RETRIEVAL clause. In this case, the record must be defined as a detail
of the CALC chain. (See Note 4 of the RETRIEVAL clause in the description of the
Record Description entry.)
GE-600 SERIES —
- 39-
LINKED-PRIOR
Linked-Prior
FUNCTION: To provide an extra chain field for each record of the chain which points to the prior
record in the chain.
FORMAT: [ :LINKED TO PRIOR |]
NOTES:
1. This clause givesthe file designer the option of specifying an extra chain field which points
to the prior record in the chain. This is desirable for several reasons and also has
several disadvantages.
The most obvious advantage is associated with the use of the “PRIOR OF CHAIN” record
specifier. (See Procedure Division, RETRIEVE Verb, Note 5.) Ability to move in both
directions is often extremely useful in problem solving. For example, a prior chain
field may be used in delinking a recordfrom a chain. Using both the prior and next chain
fields, the prior and next records are known and the chain can be easily patched to delink
a record. This is extremely helpful if the chain fields of a record are modified and
must, therefore, be delinked and relinked intoa new chain position. The modify command
requires that this relinking be completed immediately. If the prior record’s address is
not known, the entire chain will be searched until the prior record is discovered, so that
the record may be delinked. The deletion process has a delayed delinking scheme which
makes the designation of the chain prior field less important.
The use of chain prior fields has two disadvantages. First, the record size must be
increased to include the chain prior field. Second, the linking process is slower, because
the chain prior field of the next record must be modified when a new record is inserted.
Under two conditions, chainprior fields are automatically provided. If the “before current
record” ordering rule is selected, all record types in the chain will be assigned chain
prior fields for that chain. If the “last detail in chain” ordering rule is specified, the
master record (only) will be assigned a chain prior field.
GE-600 SERIES —
-40-
RANDOMIZE
Randomize
FUNCTION: To specify those fields of a calculated record which should be used to locate the
L
record in the data file.
FORMAT | RANDOMIZE ON data-name-2 | ;RANDOMIZE )|
| od
NOTES
1. This clause must be used when the Chain Definition entry is for a CALC chain.
2. Data-name-2 must be a field contained in the record being defined.
3. The randomizing routine of IDS uses as many fields as are specified in determining the
page in which the record is to be located.
GE-600 SERIES —
-41-
CHAIN-ORDER
Chain-Order
FUNCTION: To specify the criteria for sequencing the details of the chain.
FORMAT:
:CHAIN-ORDER IS
NOTES:
у
SORTED WITHIN TYPE 7
N
O
=
3
tr]
о
IN
CB
о В
Tw
> |]
ео
= |
ti (0
a |
1
This clause must be used in each Master Chain Definition entry.
When either of the SORTED options is specified, details will be added to the chain based
upon the content of the defined sort control fields of the detail records. When the (SORTED
WITHIN TYPE) option is used, records of the chain are maintained in sequence within
record type, independent of other types. When the (SORTED) option is used, the various
records of the chain are maintained in a single sequence regardless of the number of
record types in the chain. In this case, the sizes of the control fields of the various
records must be identical.
The last four options shown for this clause cause a detail to be inserted in the chain
relative to some other record in the chain. These options are:
FIRST - Insert first detail in chain relative to the master record.
LAST - Insert last detail in chain relative to the master record.
BEFORE - Insert record just before or prior to the current record of chain.
AFTER - Insert record just after the current record of chain.
GE-600 SERIES =
-49-
DUPLICATES
Duplicates
FUNCTION: To specify whether or not duplicate records may exist in the chain.
FORMAT:
i ( ARE FIRST YO
¡DUPLICATES ARE LAST
NOT ALLOWED
- _
NOTES:
1. This clause must be used whenever the chain has been defined as a sorted chain by the
CHAIN-ORDER clause.
2. When duplicates are allowed, the new detail may be positioned as the FIRST or LAST
of the string of duplicates.
3. When the NOT ALLOWED option is specified, an error condition will exist whenever
insertion of a duplicate record is attempted.
Phi 1 * ced
This clause noe ot be used when the chain is the CALC rhain In this rase duplicates
Ha
are never allowed by LDS,
GE-600 SERIES =
- 43-
SORT KEY
Sort Key
FUNCTION: To specify those data fields which
the chain named.
FORMAT Option 1.
ASCENDING
; KEY IS data-name-
DESCENDING
FORMAT Option 2.
will control the sequence of the detail record of
ASCENDING
3
DESCENDING
[ {ASCENDING RANGE KEY IS data-name-3 ]
NOTES:
1. Data-name-3 must be a field contained in
2. This clause must be used in one of its
as a SORTED or SORTED WITHIN TYPE
the detail record being defined.
two forms whenever the chain has been defined
chain.
3. When multiple sort control keys are required to define a chain sequence, the various
clauses required must be presented in sequence from major control field to minor,
thus establishing the sort level of each field.
4. The option 2 variation of this clause implies an additional function for the control field
named by data-name-3. In addition to controlling the sequence of the detail records
Payroll
System
Record
| Range Chain
Payroll
Range
Master
Payroll Chain
Pay
Records
within the chain, the value of the field also
delimits the upper maximum of the range of
values the field may assume. When a record
defined in this manner is retrieved using the
“RETRIEVE data-name-1” record specifier,
the value supplied in Working Storage for data-
name-3 is compared with the value of the field
in each detail of the chain. A match occurs
whenever the value of the field in Working
Storage is equal to or less than the value actually
stored in the record.
This type of record is used principally for
segmenting a long sorted chain to improve
the access time to specific records of the
chain. An example of this data structure and
its application is shown here.
In this example, the Pay Records are each identified by a control field called PAYNO which
may contain a value of 1 to 10,000. These records are to be segmented by establishing 100 Pay-
roll Range Masters. Each such record controls a
GE-600 SERIES
-44-
range of 100 payroll numbers.
IDS
SORT-KEY (cont.)
The Range Masters contain a control field whose initial value is 100, 200, ... or 10,000. The
Range Master records are sequenced in the Range Chain in accordance with their control fields.
When a Pay Record is stored into the system, a master record of the Payroll Chain must first
be selected, The selected record would be the first Payroll Range Master in the Range Chain
that has a control field equal to or higher than that of the new Pay Record. A Pay Record with a
PAYNO of 2128 would select as its master the Payroll Range Master with the control field value
of 2200.
The option 2 clause above identifies the fact that an exact match is not required to select the
master of the Payroll Chain, in addition to serving its normal function as a sort control key
specification.
GE-600 SERIES =
- 45-
SELECT
Select
FUNCTION: To specify the criteria for selecting the specific master record from many master
records of a given type.
FORMAT:
UNIQUE
SELECT MASTER
CURRENT
NOTES:
1. One of the two forms of this clause must be used, unless the record is defined as a detail
of the CALC chain. This definition implies that the Page Header record is the unique
master.
When using the UNIQUE option, the master is selected by matching the data field values
in the master record with those supplied in Working Storage for the new detail. In the
purchase order example (see Chapter 4), the order chain exemplifies this form of master
record selection. As a new order record is entered into the system, the vendor code
(supplied with the data fields) of the order record causes the appropriate vendor record
to be retrieved. This vendor record is the master of the chain into which the new detail
(order record) will be inserted.
When CURRENT is specified, the master selected is the current master and the detail
is inserted in its chain. The item chain in the purchase order example demonstrates this
type of master selection. Since the item records represent all of the items of an order
and since they are stored as a batch at the same time as the order record is stored, the
current order record is the master to which the items should be associated.
GE-600 SERIES —
- 46-
MATCH-KEY
Match-Key
FUNCTION: To specify those data fields which must be initialized in Working-Storage to enable
unique identification of the detail record defined by this level 98
er
Y. Tr
Ly,
FORMAT: :MATCH-KEY IS data-name-4 (MATCH KEY ]
1. This clause must be used only when the SELECT UNIQUE clause is used.
2. The fields named by this clause depend upon the RETRIEVAL clauses specified for each
of the higher level master records defined for the hierarchical structure which includes
this detail record.
e When a master record is defined as a primary record, the data field which is
equivalent to the reference code of the master must be named as a MATCH-KEY for
this detail record.
e Whena master recordis defined asa secondary record, each of the data fields which
control the retrieval of that record must be named as a MATCH-KEY field for this
detail record.
e When a masler record is defined as a calculated record, the data ficlds definedas
RANDOMIZE control fields must be named as MATCH-KEY fields tor this detail
record.
3. The use of this clause effectively defines the fields named as though they were contained
in the current detail, although they are actually contained only in the master records of
chain hierarchy.
4. This clause is not required when the record is defined as a detail of the CALC chain.
GE-600 SERIES =
-47-
SYNONYM
Synonym
FUNCTION: To specify an alternate name for a field defined as a MATCH-KEY field.
FORMAT: [ ;SYNONYM data-name-5 EQUALS data-name-4 |]
NOTES:
1.
Data-name-4 must be a field that is defined as a MATCH-KEY field for that record.
Data-name-5 is a synonym or alternate name of that field.
The following example illustrates the use of this clause. The clause may be used when
a given record is defined as a detail in two chains each having the same record type as
its master. In this case, the detail must be linked into each of the chains based upon the
contents of its defined MATCH-KEY fields. Since a common record type is the master
for both chains, the match control fields are actually the same for each chain. A separate
field is set up to hold the alternate match field value for each match control field. The
SYNONYM clause equates the alternate field and its equivalent.
When this clause is used, a separate Working-Storage area is established for the field
identified by data-name-5. In the example above, the user must initialize this alternate
Working-Storage area with the MATCH-KEY field which controls the chain for which the
SYNONYM clause is used.
GE-600 SERIES =
- 48
LINKED-MASTER
Linked-Master
FUNCTION: To provide an extra chain field for each detail record of the chain which points
to the master record of the chain.
1. This optional clause can improve the operation of the system by providing a direct path
from each detail to the master of the chain and, thus, eliminating the need for processing
all of the intervening detail records serially.
GE-600 SERIES =
- 49-
SAMPLE CODING. The following example describes the records shownin Figure 16 in IDS source
language. This illustrates one method of preparing the Data Division for the purchase order
problem.
DATA DIVISION.
IDS SECTION.
MD INFORMATION=FILE: PAGE CONTAINS 1920 CHARACTERS
FILE CONTAINS 1020 PAGES.
01 VENDOR=REC TYPE IS 100 RETRIEVAL VIA CALC CHAIN.
02 VENDORNO SIZE IS 6 NUMERIC.
02 VENDOR-NAME SIZE IS 18 ALPHANUMERIC.
02 ADDRESS SIZE IS 24 ALPHANUMERIC.
02 CITY-STATE SIZE IS 20 ALPHANUMECIC.
98 CALC CHAIN DETAIL RANDOMIZE ON VENDORNO.
98 ORDER CHAIN MASTER: LINKED TO PRIOR
CHAIN-ORDER IS SORTED.
01 ORDER-REC TYPE IS 105: RETRIEVAL VIA CALC CHAIN.
02 ORDER-NO SIZE IS 5 NUMERIC.
02 ORDER-DATE SIZE IS 3 NUMERIC.
02 PURCHAGENTNO SIZE IS 2 NUMERIC SYNCHRONIZED RIGHT.
98 CALC CHAIN DETAIL RANDOMIZE ON ORDER-NO.
98 ORDER CHAIN DETAIL; SELECT UNIQUE MASTER
MATCH=KEY IS VENDORNO# ASCENDING KEY IS ORDER-NO
DUPLICATES NOT ALLOWED.
98 ITEM CHAIN MASTER CHAIN-ORDER IS SORTED.
01 ITEM=REC TYPE IS 110 RETRIEVAL VIA ITEM CHAIN.
02 ITEMNO SIZE IS 2 NUMERIC
02 MATLIDENT SIZE IS 18 AN.
02 ORDER-QTY PICTURE IS 9999V9.
98 ITEM CHAIN DETAIL SELECT CURRENT MASTER
ASCENDING KEY IS ITEMNO DUPLICATES NOT ALLOWED.
GE-600 SERIES =
-50-
Procedure Division
The IDS procedural statements store and retrieve data with respect to the mass storage device.
In addition, these statements have thc implicit responsibility of maintaining the structure of the
ERAN y м.” м Edo 6e 5 5e A A
data file that is created by the defined chain relationships. The IDS controller accomplishes these
functions.
The communication interfaces between the IDS controllerandthe balance of the COBOL Procedure
Division are the Working-Storage areas which are established for each record type defined by the
record description entries of the IDS Section. All COBOL references to data from the IDS file
are to these Working Storage areas.
The procedural statements of IDS may appear anywhere in the context of the COBOL Procedure
Division. An IDS sentence is always preceded by ENTER IDS. The sentence may contain any
number of IDS statements and must be terminated by a period, The IDS sentence may be further
delimited by the statement ENTER COBOL. However, this is not required, since the period
following the sentence is a sufficient delimiter, All other format rules for the COBOL Procedure
Division apply to the IDS language statements, A paragraph name or section name may be assigned
to an IDS sentence in a manner consistent with normal COBOL format,
The following pages describe how these various sentence formats may be used and the limitations
which apply to each.
GE-600 SERIES =
-51-
STORE
Store
FUNCTION: To place a record into the IDS data file and to establish any chain fields which may
be required.
FORMAT: STORE data-name-1 RECORD
NOTES:
1. Data-name-1 must be one defined for a record level (01) entry of the IDS Section of the
Data Division.
9 When this verb is used, the following is assumed: (1) The Working-Storage area for
this record has been initialized with an exact image of the record. (2) Any other control
fields required to provide unique identification of the master records of the defined
chains which include data-name-1 have been initialized in their respective Working-
Storage areas.
3. The record is placed into the file as defined by the PLACE or RETRIEVAL clauses of
the Record Description entry.
4. The reference code assigned to the record stored is left in the communication cell
DIRECT-REFERENCE after the storage process is complete.
5 The record stored is recorded as the CURRENT record of its type and the CURRENT
record in each chain in which it is a master or detail.
6. If the storage process creates a duplicate record in violation to any DUPLICATES NOT
ALLOWED clause, or if the unique or range master selected, cannot be retrieved, the
storage process is terminated with all linkages restored as before and an error condition
is noted.
7. When a primary record is stored, the reference code assigned to the record is moved to
the Working-Storage field which is equivalent to the reference code for the record.
GE-600 SERIES am
- 52-
RETRIEVE
Retrieve
FUNCTION: To cause a record to be made available in the memory buffers.
FORMAT:
data-name-1 a
RECORD
URRENT data-name-1
"a
r
| C
NEXT
RETRIEVE < PRIOR RECORD OF data-name-2 CHAIN
MASTER
EACH AT END GO TO procedure-name-1
DIRECT
NOTES:
1. The various options of the RETRIEVE verb are hereafter referred to as the record
соо @ бо нос
Spcclilers.
2. Data-name-1 must be the name of the record (01) level entry defined in the IDS Section
of the Data Division.
3. Data-name-2 must be the name of a chainas defined by a level 98 entry of the IDS Section
of the Data Division.
4, Regardless of the record specifier used, this verb causes the record referenced to be
retrieved and made available in the memory buffer. This action may or may not require
that a page be transmitted from the mass storage device, since the record may already
be in memory. No otheraction, suchas moving the record to Working Storage, is implied
by this verb.
5. Of the seven record specifiers which may be used with the RETRIEVE verb, two may be
classified as absolute. This means that only one record will satisfy the retrieval
specification, whenever one of these two specifiers is executed.
RETRIEVE data-name-1 RECORD. The recordretrieveddependsonthe RETRIEVAL
clause defined as a part of the level 01 entry in the IDS Section and upon the value
contained in the control fields of Working-Storage whichuniquely identify the record.
RETRIEVE DIRECT. The record to be retrieved is identified by the reference code
stored in a communication cell named DIRECT REFERENCE. The user is re-
sponsible for initializing the communication cell prior to the execution of this com -
mand.
GE-000 SERIES =
-53-
RETRIEVE
(cont. )
oO,
The other five record specifiers may be classified as relative, since the actual
record retrieved is a function of what has transpired previously.
7
NEXT
RETRIEVE PRIOR RECORD OF data-name-2 CHAIN.
MASTER
In this case, record retrieval depends upon the CURRENT record within the chain
specified. If NEXTor PRIOR isused, the appropriate record is retrieved regardless
of the record type. When MASTER is specified, all intervening records are ignored
and the master record of the chain named is retrieved. These record specifiers can
be used only if some record has already been processed which is a member of the
specified type of chain.
RETRIEVE CURRENT data-name-1 RECORD. This record specifier instructs the
system to retrieve the last record of the type specified that was processed by any
STORE or RETRIEVE verb. If the last record processed had been deleted, it, of
course, would not be available and an error condition would exist.
RETRIEVE EACH. This record specifier provides for a serial search of the mass
storage device beginning with the reference code contained in a communication cell
called FIRST-REFERENCE. This command retrieves the first record found, unless
the reference code of the record retrieved is greater than the contents of a com-
munication cell called LAST-REFERENCE. At this point the system exits according
to the AT END statement. The user mustinitialize the communication cells with the
appropriate reference codes. The user may retrieve all of the records in the file
through the repeated use of this record specifier. As a record is retrieved, the
reference code value contained in the FIRST-REFERENCE cell is incremented by
one, so that subsequent use of this verb is initialized.
The reference code of the record retrieved is placed in the communication cell named
DIRECT-REFERENCE after the retrieval process is complete.
The record retrieved is recorded as the CURRENT record of its type and the CURRENT
record in each chain in which it is a master or detail.
If a record cannot be retrieved according to the specifications of the retrieval command,
an error condition is noted.
GE-600 SERIES —
-54-
IDS IMPERATIVE STATEMENTS. The imperative statements includedin this section are provided
as a part of the IDS language to extend the function of the basic STORE and RETRIEVE verbs. Four
of these statements apply only to the RETRIEVE verb: the GO and PERFORM statements may be
~~ че =. = аа
used with eithei vel,
When these statements are used, they must occur in the order in which they are to be executed,
a +ai A 4 +4 +h + 1a
They may be contained within the sentence beginning with the basic verb and ending with a period,
or they may be used as separate sentences to accomplish their intended result, In the latter
case, the statements must be preceded by ENTER IDS,
The specific formats of these statements and detailed discussions of the restrictions and limitations
associated with each appear on the following pages.
GE-600 SERIES —
-55-
MOVE
Move
FUNCTION: To cause the record last processed to be moved from the buffer to Working Storage
or to cause selected fields of the record to be moved,
FORMAT: ;MOVE | data-name-1 Г. data-name-2 |
NOTES:
1. The implied source of the MOVE is the record last processed and available within the
buffer.
2. This statement must be used before any reference can be made to the data in the record.
3. When the statement includes the list of fields identified by data-name-1, data-name-2,
etc., only those fields are moved to Working Storage. Otherwise, all fields are moved.
4. When the record is moved to Working Storage, each data field as defined by the 02
entries of the Record Descriptions, is unpacked into an integral number of words as if
the fields had been defined as SYNCHRONIZED, These fields are SYNCHRONIZED LEFT
unless SYNCHRONIZED RIGHT was specified in the 02 entry.
GE-600 SERIES =
-56-
HEAD
Head
FUNCTION: To enable the automatic retrieval and move to Working-Storage of the master
ELCCOLU UL UE CALI SPCCLILEU.
E
>
О
Ld
1. Data-name-1 must be the name of the chain as defined by a level 98 entry of the Data
Division. Further, some record within the data-name-1 chain must have been previously
referenced by an IDS statement.
2. This statement can best be illustrated with the example of a data structure shown below
in IDS shorthand.
REC-AD
N J CHAIN-A2
rent rm rire МН a ir NI TE .
|
\ CHAIN-A1
REC-XR REC-AB
ae
m
CHAIN-X CHAIN-A
In this case, assume that REC-AA was the record initially retrieved by the RETRIEVE
verb. At this point, three chains include REC-AA and, therefore, three possible master
records may be referenced by the HEAD statement. Notice, however, that once HEAD
has been used to reference CHAIN-A, the next higher level CHAIN-A1 can be referenced,
3. This statement includes an implied move of the record retrieved to Working-Storage.
4. After execution of this statement, the master records retrieved are the CURRENT
records of their respective types. They become the CURRENT records in each chain
in which they are defined as details. However, they are not the CURRENT records in
chains in which they are defined as master records. In those chains, the detail record
which leads to the master is the CURRENT record.
9. Note that the function of the statement is very similar to that of the RETRIEVE MASTER
RECORD statement, except for the manner in which CURRENT of chain is maintained.
(Note 4.)
GE-600 SERIES tos
„57-
MODIFY
Modify
FUNCTION: To modify the contents of a field of the record last processed and to maintain any
chains which may be controlled by the modified field.
FORMAT: :MODIFY data-name-1 [ , data-name-2 ....
NOTES:
1.
This statement causes the contents of Working Storage to replace the contents of the
fields specified in the record in the buffer.
Data-name-1 may be any field contained in the record specified by the RETRIEVE verb.
In addition, data-name-1 may be a field that has been defined as a MATCH-KEY field
by the detail Chain Definition (level 98) entry associated with the record retrieved.
When the field to be modified is a MATCH-KEY field, the association of the detail recora
with its master is changed. The record is delinked from its old master, its new master
is retrieved, and the record is linked to its new master according to the CHAIN-ORDER
clause for the chain. Note that, in this case, the field specified does not change, but,
instead, the record becomes a part of a different chain.
A field in the record retrieved thatis defined as a sequence key field may also be changed.
In this case, the field is changed, the record is delinked from its master, and the record
is relinked to the master in accordance with the new value of its sequence key field.
Note that when the field to be modified is a control field such as those mentioned in
Notes 3 and 4, a conflict in the use of Working-Storage can occur. When the original
retrieval action was accomplished by the “RETRIEVE data-name” record specifier,
the appropriate Working-Storage fields for the various control fields had to be initialized
with the values currently contained in the data structure. Since the modify function
assumes the change values are in Working-Storage, a conflict is apparent. To resolve
this problem the user should perform the following functions:
1) RETRIEVE data-name RECORD
2) Initialize Working -Storage with change values
3) MODIFY file name
GE-600 SERIES —
-58-
DELETE
Delete
FUNCTION: To delete the record retrieved with all of its de ependent detail records and to
ds mana Tier aaa Anna Freon bir отл | ni omnanifia Ad eta e A tario MANI Ary ws
Optional: y PELLUALLL CCA Laiñ та ПСБ When specified d TC La il оС Куре OCcTur LULA ing
the deletion process
e “и Er В or wd Ws BA e A LLE
FORMAT
:DELETE ON data-name-1 DETAIL
| MOVE |JTO WORKING -STORAGE |
| HEAD data-name-2 CHAIN [ HEAD...) |
L PERFORM procedure-name ]
[ со то procedure-name |
Г a —
| | OTHERWISE \ |
| € ? ON data-uailié-o DL TAIL... |
NOTES: a
1. The implicit object of the DELETE statement is the record last retrieved by the
RETRIEVE verb.
2. The deletion process deletes a record only when there are no dependent details in its
chains. When details are present, the system first attempts to delete the dependent
detail records. Since the hierarchical data structure of IDS may involve many levels
of detail records, this statement assumes powerful capabilities and should be used with
due consideration .
3. The execution of a DELETE statement makes the record retrieved unavailable for any
further processing, and an attempt to reference such a record results in an error
condition.
4. The conditional statement “ON data-name-1 DETAIL” is used only when it is necessary
to interrupt the deletion process when a dependent detail of the type named by data-name-1l
is encountered. When the statement is used, various imperative statements immediately
following are executed prior tothe actual deletion of the detail record. After the execution
of these statements, the deletion process is continued unless one of the statements was a
GO TO statement. In this case, control is not returned to the deletion process. When
the record encountered is not the type named by data-name-1, it is compared with the
type named by data-name-3. The reserved words OTHERWISE or ELSE separate the
tests for different record types that may be encountered. A record encountered which does
not match any of the specified record types is deleted in the normal manner.
9. As a record is deleted it is not implicitly moved to Working -Storage.
GE-600 SERIES | —
-59-
GO
Go
FUNCTION: To depart from the normal in-line sequence of procedures.
FORMAT: :GO TO procedure-name -1
NOTES:
1. Procedure-name-1 may be any COBOL or IDS procedural paragraph in the Procedure
Division.
2. When this statement is encountered within the IDS sentence, all subsequent statement are
bypassed and control is transferred to the procedure named.
GE-600 SERIES 2
-60-
PERFORM
Perform
FUNCTION: To depart from the normal in-line sequence of procedures in order to execute a
specific procedure and then return to the normal sequence.
FORMAT: ¡PERFORM procedure-name-1
NOTES:
1. Procedure-name-1 may be any COBOL procedural paragraph in the Procedure Division.
2. For other details concerning the PERFORM statement see the GE-625/635 COBOL
Reference Manual, CPB-1007. Only the simple PERFORM (option 1) is recognized
within an IDS sentence.
3. The procedure to be executed may not contain any IDS statement.
GE-600 SERIES —
-61-
IDS CONDITIONAL
STATEMENTS
IDS CONDITIONAL STATEMENTS. The conditional statements nf TDS are a logical extension
of the basic STORE and RETRIEVE verbs. Generally, they involve the key word IF, followed by the
condition to be tested, followed by the imperative statements to be performed.
IDS conditional statements are of two general forms’ either form may appear in the string of
statements following a basic verb.
The specific formats ot these statements and a discussion of their restrictions and limitations
are outlined below.
FORMAT Form 1.
IF data-name-1 RECORD [ siatement- C:statement-2...] }
{ OTHERWISE
(statement: [ :statement-4... 1)
ELSE
NOTES:
1. This form may only follow a RETRIEVE DIRECT, RETRIEVE EACH, or RETRIEVE
e RECORD OF data-name CHAIN.
PRIOR
2. Statement-1, statement-2, etc., may be any imperative statements. Statement-3,
statement-4, etc., may be any imperative statements, or a conditional of this form
(form 1).
3. This form of conditional is an implicit comparison of the record type just retrieved with
the record type named by data-name-1. If the record type of the record last processed
equals the record type named, the condition is true and statement-1, statement-2, etc.,
will be executed in order. Then control is transferred to the next sentence. If the con-
dition is false statement-3, statement-4, etc., will be executed. If the clause beginning with
the key words OTHERWISE or ELSE is omitted and the condition is false, control will be
transferred to the next sentence.
FORMAT Form 2.
OTHERWISE
¡IF ERROR statement-1 statement-2 L ‚statement-3... ]
ELSE
NOTES:
1. This form may only follow a STORE or RETRIEVE verb or a MODIFY, DELETE, HEAD
or MOVE imperative statement.
2. Statement-1 may only be a GO TOora PERFORM imperative. Statement-2, statement-3,
etc., may be any imperative statement appropriate to the basic verb, or a conditional of
form 1, if appropriate.
3. This form signals the occurrence of any logical or physical error as a result of some IDS
command. The specific errors which may occurare a function of the command executed.
A detailed coding scheme for identifying errors is defined in Appendix C. The user pro-
gram may determine the type of error by referring toa communication cell named
ERROR-REFERENCE.
GE-600 SERIES —
-62-
CLOSE IDS
Close IDS
FUNCTION: To force the writing to the file unit of any pages in memory which have been modified.
FORMAT: CLOSE
NOTES:
1. This statement must be executed before any COBOL STOP RUN statement. No automatic
closing takes place.
GE-600 SERIES
-63-
OPEN IDS
Open
IDS
FUNCTION: To initialize the processing of the IDS data file.
FORMAT:
RETRIEVAL
OPEN FOR
UPDATE
[WITH AUTHORITY -KEY integer-1 |]
NOTES:
1.
2.
This statement must be executed before any other IDS verb is executed.
When the IDS file is opened for RETRIEVAL, the STORE, DELETE, and MODIFY state-
ments of IDS are not operative. An attempt to use these statements under these conditions
RETRIEVAL
UPDATE
results in an error condition during program execution. When the FOR
clause is not used, UPDATE is assumed by IDS.
The AUTHORITY-KEY clause enables access to various record types which may be pro-
tected by a defined AUTHORITY code. (See Data Division, Record Description). The
value of integer-1 may not exceed 4095, )'
10
When this clause is used, each reference to a record of the IDS file involves a match of
the AUTHORITY value defined for the record with the AUTHORITY-KEY supplied. When
a valid match occurs, the IDS verb is allowed to function normally. Otherwise, the
function of the verb is aborted and an error condition is returned to the user’s program.
The exact details of the matching process may be modified with each installation, to suit
individual requirements.
GE-600 SERIES ==
-64-
GECOS SYSTEM CONTROL
The IDS Translator is a “system program” which may be called by GECOS. The $ IDS control
card is used to call the IDS Translator from system storage. This control card is identical to the
COBOL control card except for the contents of columns 8-12 which contain “IDS” instead of
“COBOL.” The various processing options which may be specified are identical with those speci-
fied for COBOL. (See GE-625/635 Comprehensive Operating Supervisor, CPB-1002A for details
of the COBOL control card.)
At the time of allocation for the IDS Translator, sufficient resources (memory and peripheral de-
vices) are allocated to provide for COBOL. When the translator has completed its function, it
passes control to COBOL using the GECOS entry point GECALL. If the translator encounters a
serious error, COBOL is not called and the job is terminated after anpropriate explanation has
been placed on the “execution report” associated with the job. Figure 20 is a flow diagram of the
compilation process of an IDS prog rau.
OBJECT PROGRAM EXECUTION
The object program created from the IDS source language consists of a modular set of subroutines
which interpretively execute the store and retrieve commands. GELOAD loads these subroutines
as a result of the calls generated by the compilation process.
Because of the interpretive mode of execution, the complete data description of the IDS data file,
also generated by the translator, must be available to these routines.
The Input/Output Controller module of IDS controls the mass storage device. It transfers pages
in and out of the core in response to commands to retrieve a specific record, or to store a specific
record. In order to minimize the seek and transfer time, an inventory of pages is maintained in
memory. Increasing the number of pages stored in core increases the possibility that the one
needed next will already be there. To further improve the possibility of finding the page desired
in core, the Input/Output Controller keeps track of the sequence of page utilization and holds the
most recently active data pages in core. Pages which are infrequently accessed are retired from
core as others are called in. The Input/Output Controller notes which pages have been modified
and writes only the modified pages back tothe mass memory device. Sufficient core memory must
have been allocated to the object program to ensure a minimum of two pages retained in memory,
but usually a more practical number is four or more.
GE-600 SERIES E
-65-
Exception
Report
IDS -COBOL
Source
Language
IDS
Translator
Language
N
N
N
N
N
N
COBOL
Library Listing
COBOL or
Compiler > Exception
Report
Input
GMAP Assembly
— Listing
Object
Code
GELOAD BR “Map
Map
Subroutine
Library
(Includes IDS
Subroutines)
Figure 20. IDS Compilation Process
GE-600 SERIES —
-66-
ASSIGNMENT OF IDS BUFFERS
The buffer area for the Input/Output Controller module of IDS is established when the IDS file
a )
Ww
opened. The user defines the size of this area (that is, the number of buffers established
employing one of the two following procedures.
1. A labeled common area (.QAREA) may be used. Its size is specified by a GELOAD
control card as shown below:
1 Q 18
À VU AU
$ USE .QMAX/1/,.QAREA /n/,.QMIN/1/
The control card must be inserted in the object deck so that GELOAD will encounter it
prior to its retrieval of the IDS routine from the library. (See CPB-1002.)
The value supplied for “n” must be equal to the number of words to be usedfor buffers.
When the file is opened, the size of .QAREA will be determined and divided into “m”
buffers, where “m” is the integral portion of the area size divided by page size. The
resulting value of “m” must be at least 2 and is limited to 50 buffers.
2. If the above control card procedure is not used, the OPEN routine of IDS will attempt
to use the area of memory not assigned to other program segments. This “available
area” is defined by the contents of word 31( of the program asinitialized by GELOAD
during the loading process. As in the procedure above, the “available area” is then
divided into some number of buffers depending on the size of the area. Again a minimum
of 2 and a maximum of 50 buffers is established.
When the IDS buffers are established in the latter manner, the user must insure that the area to
be used is not concurrently used by any other segment of the program. For example, a program
which calls upon IDS and SORT/MERGE would cause a conflict in the use of memory, since both
routines would attempt to use the available area. To guard against such a possibility, the OPEN
routine of IDS determines that the available area is actually available and sets a bit in the GECOS
“switch word” indicating the assignment of the area to IDS. If some other routine has pre-
viously made use of the area, the program is aborted. The use of the labeled common area, as
outlined for procedure 1 above, would, of course, eliminate any potential problem of this type.
FILE UNIT INITIALIZATION
Prior to the operation of any IDS program, the disc storage unit must have been initialized with a
Page Header record as the first record of each page in the IDS data environment. See Appendix A
for the detail format of the Page Header record.
In addition to the Page Header record, a single Environment record must be stored as the first
record of page 1 of the data environment. The Environment record is used each time the IDS
file is opened to accomplish necessary initialization of the Input/Output Controller module of IDS.
A generalized utility program will be available to accomplish this disc storage unit initialization.
GE-600 SERIES ms
-67-
APPENDIX A
IDS RECORD FORMATS
PAGE HEADER RECORD FORMAT
CHARACTERS 1 2 3 4 5 6 7 8 9 10 11 12
REFERENCE CODE RECORD CALC CHAIN NEXT SPACE
PAGE NO. TYPE LINK AVAILABLE “
AVAILABLE LINE NO. FLAGS —
ENVIRONMENT RECORD
4 5 6 7 8 9 10 11 12
RECORD PAGE |
(22)
FILE NAME >
TYPICAL DATA RECORD FORMAT
1 2 3 4 5 6 7 8 9 10 11 12
REF, IC)
CODE RECORD RECORD CHAIN-"A" CHAIN-A
LINE | TYPE SIZE NEXT LINK PRIOR
NO. | A
CHAIN-B
LINK NEXT LINK DATA FIELDS
(2) DELETE SWITCH
NOTE: Reference Code is a composite of the Page Numbers from the Page Header record and the
unique Line Numbers from the data record.
GE-600 SERIES
-69-
APPENDIX B
IDS RESERVED WORDS
IDS uses all the reserved words specified for COBOL. In addition, it employs the reserved words
listed below. The user must avoid using words on both these lists for data-names.
ALLOWED
AUTHORITY
CALC
CHAIN
CHAIN-ORDER
CURRENT
DELETE
DIRECT
DUPLICATES
EACH
ERROR-REFERENCE
FIELD
FIRST-REFERENCE
HEAD
IDS
INTERVAL
LAST-REFERENCE
GE-600 SERIES
LINKED
MASTER
MATCH-KEY
MODIFY
NEAR
PAGE -RANGE
PRIOR
UNIQUE
RANDOMIZE
RECORD-TYPE
RETRIEVAL
RETRIEVE
SORTED
STORE
SYNONYM
VIA
WITHIN
IDS
-71-
APPENDIX C
IDS ERROR CONDITIONS
Two types of error conditions may occur during the execution of an IDS program. The first of
these, involves those situations whichare data dependent and must be anticipated by the procedural
logic of the program,
These errors are returned to the user program and may be tested by
reference to a cell called ERROR-REFERENCE. The following codes are in this category.
RO1
R02
RO3
RO9
D01
501
The retrieval of a record depends upon the selection of the current
master of a given type. That record has not been retrieved or has
been deleted.
Record cannot be retrieved because the record or one of its masters
has been deleted.
The primary record retrieved is not the same record type as that
specified in the Record Definition.
A + r я A р A + + + .
Attempt tc retrieve a record which does not exist within the data
structure identified by supplied control fields.
Attempt to retrieve the current record of a given type when that record
type has not been previously retrieved or has been deleted.
Attempt to retrieve DIRECT when DIRECT-REFERENCE is zero.
Attempt to retrieve a record which has been deleted and its Delete
switch is set.
Reference code of the record to be retrieved is not found within the
specified page.
Reference code of the record to be retrieved is outside the total range
of pages of the file.
Attempt to store an unallowed duplicate record.
Attempt to store a record when there is no space available within the
range of pages specified for the record.
The second type of error involves those situations which are the result of improper use of the
IDS functions, invalid definition of the data file, or hardware malfunctions which cannot be recovered
by the software.
These conditions, which are listed on the following page, result in an abort of
the program with the following message printed on the execution report:
***x ABORTED BY IDS QUIT ROUTINE REASON CODE XX ***
GE-000 SERIES =
-73-
In addition to the above message, a memory dump of the program is produced. Any time an abort
occurs for any reason within an IDS program, the IDS file is first CLOSED with the appropriate
IDS pages restored to the mass storage device.
Reason Code Description
01 Attempt to store or retrieve a record with the IDS file not opened.
02 Capacity of the disc file allocated viaGECOS is not large enough for the
number of pages specified by the environment record.
03 Insufficient memory allocated to provide at least 2 buffers.
04 Authority -key does not match authority lock.
05 Invalid device-type allocation.
10 Definition Error-No detail definition for secondary or calculated
record.
11 IDS subroutine table overflow. QASC, QDLTE and QSTOR must be
reassembled.
12 Definition Error-Unique field for primary record has not been defined.
13 Definition Error-No record definition has been established for this
record type.
14 Usage Error-Attempt to head chain for which no current record exists.
15 Attempt to Modify, Delete or Store a record with processing mode not
equal to “Update.”
16 Field to be modified or moved not defined for the current record.
17 Attempt to delete a record not previously retrieved.
18 Attempt to retrieve next in a chain for which no current record exists.
19 A control definition has been encountered which has an invalid control
code.
20 A sort control or a randomize control field is specified with a record
increment of zero.
21 Record definition of a record to be stored indicates that the recordis
less than 6 characters long.
22 Record definition of a record to be storedindicates that the record size
is greater than page size 22.
23 Attempt to store a page header record.
24 Attempt to store a primary record with no unique field defined.
GE-600 SERIES m
-74-
Reason Code
25
DS
cn
27
en
DI
1
en
en
56
57
58
Description
Attempt to store a record which is a detail in one or more chains
when no storage chain is specified.
Record type of record retrieved is not defined for the chain specified.
Invalid conditional action specified for a conditional delete statement.
Linking of a record cannot be completed. Either the record cannot be
retrieved from the disc or, for some other reason, the chain-pointer has
been lost or is invalid.
Page read from mass storage device is not the page requested.
Attempt to delink a record when the chain table “NEXT?” is zero.
(Subroutine error)
Attempt to link a record into a chain with the chain table “NEXT” equal
to zero. (Subroutine error)
GE-600 SERIES —
-75-
">