27.8 Extending the compiler. 4tH 4tH Forth compiler

PDF

Dokument

Reklaam

CHAPTER 27. MODIFYING 4TH

27.8 Extending the compiler

521

So far we’ve only added words that are directly compiled into a token/argument pair. If that was the way all 4tH words worked, we would never have branches, variables or other things that make up a language. In fact, you would never construct this complex architecture, since there are easier ways to achieve the same functionality.

The secret lies in the last major function of comp_4th() we have discussed, which is called GetImmediate(). When you look at it for the first time, it looks quite like

GetConstant() and GetWord(). The associated table ImmedList[] has a lengthbyte and a name but instead of a token or a value the last field is a pointer to a function.

You’ll find these functions above GetImmediate() and they all start with Do..(). In fact, they are the icing on the cake. They make 4tH a compiler, since they allow non-linear compiling. That includes:

• Branching

• Comments

• Allocation of variables

• String handling

• Assertions

• Constants

If there is something you want to do that you cannot define in a single token/argument, this is the place where you have to be. But before you are starting to make new functions, note there are many functions that you can use.

E.g. when you want to compile a word, you don’t have to bother yourself with errorchecking or other 4tH-internals. Just make a call to CompileWord():

CompileWord (NIP, 0);

Extending the compiler is quite easy. We will illustrate that by using a very simple example.

We have some compiler-words that handle the radix at compile time. They are:

[BINARY]

[OCTAL]

[DECIMAL]

[HEX]

Now we want to add a new one called "[SEXTAL]", which sets the radix to six. The radix at compile-time is handled by a single variable called "Base". First we have to make a function, which sets "Base". We call it DoSextal(). Note that this function cannot receive or return any values, like all Do..() functions:

#ifndef ARCHAIC static void DoSextal (void)

#else

#endif static void DoSextal ()

{

Base = 6;

}

CHAPTER 27. MODIFYING 4TH

MarkLink()

MakeLink()

PairLink()

CompileMark()

RANCHING

Adds a link to the controlstack

Makes a back-link

Retrieves a link from the controlstack

Compiles a token and adds a link to the controlstack

YMBOLTABLE

AddSymbol()

MakeSymbol()

GetSymbol()

Adds a symbol to the symboltable

Adds the current word as a symbol to the symboltable

Retrieves an entry from the symboltable

GetNextWord()

ARSING

Gets next word in the source

DecodeSymbol() Decodes a symbol from source

DecodeLiteral() Gets a compiled literal expression

DecodeOperand() Gets a compiled expression

DecodeWord() Gets a name from source

DecodeName()

SkipSource()

Gets a previously delared name from source

Discards all source between two labels

CompileWord()

InlineWords()

MoveString()

CompileString()

OMPILING

Compiles a token and its argument

Compiles a sequence of tokens without arguments

TRINGS

Moves a string inside the String Segment

Compiles a token and its associated string

Table 27.3: comp_4th() basic API

522

Just place it anywhere before ImmedList[] and you’re safe. Now we have to make it work. Like the other Get..() functions, there is a table called ImmedList[] which drives this behaviour, so let’s get started:

1. Length byte = 8

2. Symboltable entries = 0

3. Additional tokens = -1

4. Delimiter = ""

5. Name = "[SEXTAL]"

6. Function = DoSextal

Now let’s update that table:

{ 7, 0, -1, "ALIGNED",

{ 7, 0, 0, "RECURSE",

{ 8, 0, -1, "[SEXTAL]",

{ 8, 1, -3, "CONSTANT",

{ 8, 1, -2, "VARIABLE",