rob@aoa.UUCP (Rob Everton) (10/07/87)
Seeking comments on the process of developing a small language
compiler system for the ibm pc family using Yacc and Lax
from a different machine, probably 3b1 unix pc's or something.
Has anyone tried this?
Does anyone have any comments for or against the two utilities
or the possibility of problems porting to a pc format processor?
Thanks in advance.
.
--
Rob Everton
...!{harvard,ima}!bbn!aoa!rob
...!{wjh12,mit-vax}!biomed!aoa!robwindley@iris.ucdavis.edu (Phil Windley) (10/08/87)
In article <468@aoa.UUCP> rob@aoa.UUCP (Rob Everton) writes: >Seeking comments on the process of developing a small language >compiler system for the ibm pc family using Yacc and Lax >from a different machine, probably 3b1 unix pc's or something. > >Has anyone tried this? > I recently *tried* (the operative word) to port an object-oriented language developed here at UCD to an IBM. I didn't have problems with YACC and LEX, just things like memory models, etc. Here's what I learned about YACC and LEX from it: a) If you have access to another machine you can run your .y and .l filesd trough YACC and LEX respectively, giving you .c programs. Those will compile using Microsoft C 4.0 just fine (I used ULTRIX 2.0 YACC and LEX, but I think that YACC and LEX are pretty standard, I could be wrong.) b) the parser will run fine. The lexer will need to be linked with the lex library. The lex library has four or five quite short routines like yyless(). I will leave it up to you to decide how you get the source and make a simlar library for your PC. c) Given the above two things, you don't need YACC and LEX for the PC if you have them available elsewhere. The parser we wrote (which is not as large as an ADA parser, but still not insignificant, works great. Since we were not careful with pointers/integers (and on a 32 bit machine, you don't really have to be) the rest runs fine on 68000's, VAX's, etc and crashes a 16 bit machine like the 80286. Phil Windley Robotics Research Lab University of California, Davis Phil Windley Robotics Research Lab University of California, Davis
dws@cseg.UUCP (David W. Summers) (10/08/87)
In article <468@aoa.UUCP>, rob@aoa.UUCP (Rob Everton) writes: > Seeking comments on the process of developing a small language > compiler system for the ibm pc family using Yacc and Lax > from a different machine, probably 3b1 unix pc's or something. > Has anyone tried this? > Does anyone have any comments for or against the two utilities > or the possibility of problems porting to a pc format processor? > Rob Everton > ...!{harvard,ima}!bbn!aoa!rob > ...!{wjh12,mit-vax}!biomed!aoa!rob I've written assemblers with Xenix Yacc and Lex and then compiled them with MicroSoft C. Works Great! Is this what you were asking about? The only trouble I've had is when I try to port GREAT BIG Lex or Yacc specifications. Then the C compiler runs out of table space/data space sometimes. Other than that I've had no problem porting stuff from these machines to the IBM PC: 1) IBM MainFrame (Amdahl/UTS Sys V) 2) AT&T 3B2/400 (Unix Sys V/2.0.4(?)) 3) Tandy 3000 HD (Xenix Sys V/286) Hope this helps! - David Summers ..!rutgers!pbox!romed!actsn!david ..!rutgers!pbox!romed!actsn!cseg!dws
farren@gethen.UUCP (Michael J. Farren) (10/09/87)
In article <340@ucdavis.ucdavis.edu> windley@iris.ucdavis.edu (Phil Windley) writes: > Since we were not careful with pointers/integers (and > on a 32 bit machine, you don't really have to be) the rest runs > fine on 68000's, VAX's, etc and crashes a 16 bit machine like the > 80286. Which just illustrates a point: you DO have to be careful with pointers and integers, dammit! You may not KNOW that your code will only ever run on 32-bit machines. Give everybody a break, especially those who will come after you and try and maintain and port your code, and DON'T TREAT POINTERS AND INTEGERS AS EQUIVALENT! (Emphatic enough?) As far as the original statement goes, I've had much success creating parsers with lex and moving the generated lex.yy.c code over to the PC, and recompiling with MSC. The functions of the lex library routines are described pretty adequately in the man page for lex, and aren't particularly hard to implement independently. -- ---------------- Michael J. Farren "... if the church put in half the time on covetousness unisoft!gethen!farren that it does on lust, this would be a better world ..." gethen!farren@lll-winken.arpa Garrison Keillor, "Lake Wobegon Days"
dave@westmark.UUCP (Dave Levenson) (10/12/87)
In article <468@aoa.UUCP>, rob@aoa.UUCP (Rob Everton) writes: > Seeking comments on the process of developing a small language > compiler system for the ibm pc family using Yacc and Lax > from a different machine, probably 3b1 unix pc's or something. > > Has anyone tried this? Yes, I have ported the cdecl utility (it translates between c declaration syntax and English) to MS-DOS from UNIX by doing exactly what you are asking about! I ran the lex and yacc portions of the make procedure under UNIX, and then compiled the resulting C code, along with the rest of cdecl, using the Microsoft C (rel 4.0) under MS-DOS 3.2. The result is that the utility runs under MS-DOS as it does under UNIX. I'm now hacking up the lex and yacc sources to understand the far and near extensions, so that it will be more useful to MS-C developers. The development system here is ideal for the task: an AT&T PC6300PLUS running Simul-Task. It allows me to run MS-DOS applications from within UNIX programs. What I use to make cdecl.exe is a single UNIX Makefile that runs lex and yacc as usual, and then runs cl as a dos command! The same makefile can be used to make cdecl (the unix executable) and cdecl.exe. It invokes cc to make the unix flavor, and cl to make the dos flavor. -- Dave Levenson Westmark, Inc. A node for news. Warren, NJ USA {rutgers | clyde | mtune | ihnp4}!westmark!dave