scott@etive.ed.ac.uk (Scott Larnach) (07/19/88)
This is a repost of an article which I posted earlier with a "uk" distribution. The silence was deafening... I wonder if what I'm asking is contrary to the spirit of FORTH...? Or is it just that noone's gotten around to doing it, or felt there was a need? Anyway: Is there a FORTH for Unix that would allow me to access Unix system calls, and create FORTH programs through the use of my favourite text editor. In other words, I'd like FORTH to behave as a "normal" language. I have seen FORTHs on the net for Unix, but they have had their own editors and file structures. Thanks in advance.
XRBEO@VPFVM.BITNET (Bruce O'Neel) (07/22/88)
A few comments from someone who's only had a little experience. Most of the time xxx for unix implys xxx in C. I've played around with forth written in C and the problem I've always had is getting C to have a "Jump" table for quick vectoring of the "CODE" words (which are actually written in C). A few (MANY?) C compilers can support pointers to functions which would seem to work if you had an array of pointers to functions, but just using a large switch statement would seem to be quite slow. Just a few random thoughts. Maybe someone with more experience could also comment? bruce
tobeye@eleazar.dartmouth.edu (Anthony V. Edwards) (07/22/88)
> A few (MANY?) C compilers can support pointers to functions which would > seem to work if you had an array of pointers to functions, but just using a > large switch statement would seem to be quite slow. Actually, function pointers are part of the original K&R C, so all compilers *should* have them (whether a programmer uses them or not is another matter). As for case statements being slow, good compilers should automatically set up a jump table when a large case statement is found. So a large case state- ment is not necessarily slow at all. Anthony
Ralf.Brown@B.GP.CS.CMU.EDU (07/22/88)
In article <8807211846.AA27919@jade.berkeley.edu> you write: }C). A few (MANY?) C compilers can support pointers to functions which would }seem to work if you had an array of pointers to functions, but just using a }large switch statement would seem to be quite slow. If it doesn't support pointers to functions, it isn't full C, but only a subset. Pointers to functions is one of the most useful features of C. -- {harvard,uunet,ucbvax}!b.gp.cs.cmu.edu!ralf -=-=- AT&T: (412)268-3053 (school) ARPA: RALF@B.GP.CS.CMU.EDU |"Tolerance means excusing the mistakes others make. FIDO: Ralf Brown at 129/31 | Tact means not noticing them." --Arthur Schnitzler BITnet: RALF%B.GP.CS.CMU.EDU@CMUCCVMA -=-=- DISCLAIMER? I claimed something?
rsl@bio-image.UUCP (Bob Lippert) (07/26/88)
In article <8807211846.AA27919@jade.berkeley.edu> Forth Interest Group International List <FIGI-L%SCFVM.bitnet@jade.berkeley.edu> writes: >A few comments from someone who's only had a little experience. > >Most of the time xxx for unix implys xxx in C. I've played around with forth >written in C and the problem I've always had is getting C to have a "Jump" >table for quick vectoring of the "CODE" words (which are actually written in >C). A few (MANY?) C compilers can support pointers to functions which would >seem to work if you had an array of pointers to functions, but just using a >large switch statement would seem to be quite slow. I have provided the following code as an example of how I have set things up for a Forth written in C implementation which I call FORTHinC. If anyone can come up with a faster or better way of doing this I would be interested. I'm currently re-structuring FORTHinC to make sure that it will run on UNIX (it was originally written for an IBM/PC) and will test it on UNIX and then release it to the public (in around 2 weeks from now). Cognetics , Inc. Robert S. Lippert 1275 N. Silo Ridge Dr. Ann Arbor, Mi. 48104 /*---------------------- FORTHinC starter edition ----------------------- This code is provided as an example of how FORTHinC is designed and can be used as a starting point for any compiler or emulator written in C. This example will process a double nested looping structure and provides an indication of the processing speed of the inner loop relative to other languages. The following relative performance times are approximations. The actual times you get will depend upon your computer and C compiler: C Language = 1 FORTHinC = 9 BASIC = 19 -------------------------------------------------------------------------*/ #define DO 0 #define DOEND 1 #define NUMBER 2 #define FINISH 3 #define CALL 4 #define CRET 5 #define LITERAL 6 #define RNUMBER 7 #define MAXTOP 64 unsigned *pc, code[256]={NUMBER,30,DO,CALL,0,DOEND,FINISH, /*example called routine */ NUMBER,30000,DO,DOEND,CRET}; unsigned *calls[32] ={&code[7]}; unsigned doi; /* the do loop count */ /*----- define the stacks -----*/ unsigned *tstk[MAXTOP],ttop = -1; /* the call stack */ int istk[MAXTOP], itop = -1; /* the forth stack */ unsigned dostk[MAXTOP],dotop = -1; /* the do loop stack */ /*---------------------------------------------------------*/ void call() { unsigned *newpc; newpc = calls[ *pc++ ]; /* call routine */ tstk[++ttop] = pc; /* push return address */ pc = newpc; } /*---------------------------------------------------------*/ void cret() { pc = tstk[ttop--]; } /*---------------------------------------------------------*/ void walk (bump,dbump,end1,end2) /* walk forward thru code */ int bump,dbump,end1,end2; { int level,value; value = *pc; level=0; while( (value!=end1 && value!=end2) || level!=0 ) { if ( value==bump ) level += 1; if ( value==dbump ) level -= 1; switch (value) { /* skip over any literal areas that are found */ case CRET: pc=tstk[ttop--]; break; case LITERAL: pc++; pc += *pc; break; case NUMBER: pc += 2; break; case RNUMBER: pc += 3; break; default: pc++; break; } value = *pc; } } /*-------------------------------------------------------*/ int pop( ) { if ( itop<0 ) printf("Stack Underflow!"); else return( istk[ itop--]); } /*-------------------------------------------------------*/ void push(ent) int ent; { if ( itop >= MAXTOP ) { printf("\nStack Overflow!"); itop=MAXTOP; } istk[ ++itop] = ent; } /*---------------------------------------------------------*/ void do_op() { int iv1; dostk[++dotop]= doi; /* save previous doi */ iv1=pop(); if ( iv1<1 ) { doi=dostk[dotop--]; walk ( DO, DOEND, DOEND, -999); pc++; } else { doi=(unsigned)iv1; tstk[++ttop]=pc; /* also save start of do */ } } /*--------------------------------------------------------*/ void doend() { if (--doi>0) pc = tstk[ttop]; else { ttop--; /* throw away do start */ doi=dostk[dotop--]; /* and get previous doi */ } } /*-------------------------------------------------------*/ void number() { push ( (int)*pc++ ); } /*-------------------------------------------------------*/ void finish() {printf("\nExiting!FORTHinC"); exit(1); } /*-------------------------------------------------------*/ void (*dsp[200])()={ do_op,doend,number,finish,call,cret}; /************************* main *********************************/ main() { pc=code; /* start at first location in code */ innerloop: (*dsp[*pc++])(); /*dispatch to FORTHinC code*/ goto innerloop; }
cameron@symcom.math.uiuc.edu ( Cameron Smith) (03/09/91)
I would like to get a public domain Forth to use on my
NeXT computer. I have heard rumors of Forth's written
in C, but all the PD Forth's that I seem to be able to find
are in assembly language for IBM PC's. (Except for
a version called c-forth, dated 1985, which is an
implementation of fig-FORTH, except that the author
didn't implement <BUILDS...DOES> !)
If I could have my druthers I'd prefer a system that
used standard Unix files rather than blocks, but at
this point I won't refuse any suggestions.
Thanks in advance for any info. Please respond by
email and I'll summarize in about a week (if I do
get any responses).
--Cameron Smith
cameron@midd.cc.middlebury.edu (Please respond to
this address, *NOT* to any address that may
appear in the header of this msg! Thanks!)