ForthNet@willett.UUCP (ForthNet articles from GEnie) (01/05/90)
Date: 12-26-89 (21:06) Number: 1501 To: IAN GREEN Refer#: 1431 From: DOJUN YOSHIKAMI Read: NO Subj: POWERFUL SCRIPT PROCESSOR Status: PUBLIC MESSAGE Forth as a virtual machine........ not a bad idea, actually. I currently have a special project which consists of writing an interpreter which uses the forth interpreter to traverse the syntax tree. A couple things which may help: with a little tweaking, the syntax tree that comes out of a parser can be modified such that it resembles the headerless dictionary. Rather than being a syntax tree, I suppose we should call it a syntax DAG. At any rate, once that is built, you can use the forth interpretr to directly execute the syntax tree. Forth intrinsically lacks the nice memory heap utilities such as malloc and free (unles someone wrote them). ALLOT and FORGET just aren't the same. Thus I would forgo implementing the parser and scanner in forth. The code generator and typechecker, however, could be put together nicely. The inner interpretr is wonderful for traversing trees, the only trick you have to do is to make the contents of each node (or leaf) executable in some form. It would not be bad practice to try writing your own forth interpreter to get some idea of how the innards work (or even look at Kelly and Spies's "Forth a Text and REference" particulary in the last chapters). THe little project I have scratched to gether is on a UNIX box, so of course it is written in C. C seems to buck every now and then, especially when squeezing pointers to functions all over the place. I think what I am writing is some oddball flavour of subroutine-threaded code. Perhaps token threading would be more appropriate: use one token table when traversing for type checking, use the other for execution. (Even better, have a third token table for generating machine code). I would personally put forth as the back half of the analysis phase, and as the back end to the compiler. Since I can't seem to find a nice whizz-bang parser generator/scanner generator in Forth (such as yacc or lex) I'd leave that part to some other language (for now). At any rate, what part of the compiler were you specifically thinking of targetting Forth for? D. YOshikami ----- This message came from GEnie via willett through a semi-automated program. Report problems to: 'uunet!willett!dwp' or 'willett!dwp@gateway.sei.cmu.edu'
ForthNet@willett.UUCP (ForthNet articles from GEnie) (01/05/90)
Date: 12-27-89 (14:30) Number: 1507
To: DOJUN YOSHIKAMI Refer#: 1501
From: IAN GREEN Read: NO
Subj: POWERFUL SCRIPT PROCESSOR Status: PUBLIC MESSAGE
The ideas I am kicking call for a standard procedural language that
is compiled into a subset of Forth. To this end I am trying to gather up
details about how Forths are themselves implemented. The compiler itself
will be written in a procedural language (Modula-2) since I am most
familiar with the block structured style of programming techniques.
You mentioned some things about parser generators like YACC. Alas I
have looked at PC versions of these programs and can find no
documentation whatsoever. They all refer to the Unix documenation.
BTW today I bought a mouse so I will spend the next few days writing a
Module to support the rodent in my projects.
NET/Mail : British Columbia Forth Board - Burnaby BC - (604)434-5886
-----
This message came from GEnie via willett through a semi-automated program.
Report problems to: 'uunet!willett!dwp' or 'willett!dwp@gateway.sei.cmu.edu'