1: \ Copyright 1992 by the ANSI figForth Development Group
2: \
3: \ WARNING: This file is processed by m4. Make sure your identifiers
4: \ don't collide with m4's (e.g. by undefining them).
5: \
6: \
7: \
8: \ This file contains instructions in the following format:
9: \
10: \ forth name stack effect category [pronunciation]
11: \ [""glossary entry""]
12: \ C code
13: \ [:
14: \ Forth code]
15: \
16: \ The pronunciation is also used for forming C names.
17: \
18: \
19: \
20: \ These informations are automatically translated into C-code for the
21: \ interpreter and into some other files. I hope that your C compiler has
22: \ decent optimization, otherwise the automatically generated code will
23: \ be somewhat slow. The Forth version of the code is included for manual
24: \ compilers, so they will need to compile only the important words.
25: \
26: \ Note that stack pointer adjustment is performed according to stack
27: \ effect by automatically generated code and NEXT is automatically
28: \ appended to the C code. Also, you can use the names in the stack
29: \ effect in the C code. Stack access is automatic. One exception: if
30: \ your code does not fall through, the results are not stored into the
31: \ stack. Use different names on both sides of the '--', if you change a
32: \ value (some stores to the stack are optimized away).
33: \
34: \
35: \
36: \ The stack variables have the following types:
37: \
38: \ name matches type
39: \ f.* Bool
40: \ c.* Char
41: \ [nw].* Cell
42: \ u.* UCell
43: \ d.* DCell
44: \ ud.* UDCell
45: \ r.* Float
46: \ a_.* Cell *
47: \ c_.* Char *
48: \ f_.* Float *
49: \ df_.* DFloat *
50: \ sf_.* SFloat *
51: \ xt.* XT
52: \ wid.* WID
53: \ f83name.* F83Name *
54: \
55: \
56: \
57: \ In addition the following names can be used:
58: \ ip the instruction pointer
59: \ sp the data stack pointer
60: \ rp the parameter stack pointer
61: \ lp the locals stack pointer
62: \ NEXT executes NEXT
63: \ cfa
64: \ NEXT1 executes NEXT1
65: \ FLAG(x) makes a Forth flag from a C flag
66: \
67: \
68: \
69: \ Percentages in comments are from Koopmans book: average/maximum use
70: \ (taken from four, not very representative benchmarks)
71: \
72: \
73: \
74: \ To do:
75: \
76: \ throw execute, cfa and NEXT1 out?
77: \ macroize *ip, ip++, *ip++ (pipelining)?
78:
79: \ these m4 macros would collide with identifiers
80: undefine(`index')
81: undefine(`shift')
82:
83: noop -- fig
84: ;
85: :
86: ;
87:
88: lit -- w fig
89: w = (Cell)*ip++;
90:
91: execute xt -- core,fig
92: cfa = xt;
93: IF_TOS(TOS = sp[0]);
94: NEXT1;
95:
96: branch-lp+!# -- new branch_lp_plus_store_number
97: /* this will probably not be used */
98: branch_adjust_lp:
99: lp += (Cell)(ip[1]);
100: goto branch;
101:
102: branch -- fig
103: branch:
104: ip = (Xt *)(((Cell)ip)+(Cell)*ip);
105: :
106: r> dup @ + >r ;
107:
108: \ condbranch(forthname,restline,code)
109: \ this is non-syntactical: code must open a brace that is close by the macro
110: define(condbranch,
111: $1 $2
112: $3 ip = (Xt *)(((Cell)ip)+(Cell)*ip); NEXT;
113: }
114: else
115: ip++;
116:
117: $1-lp+!# $2_lp_plus_store_number
118: $3 goto branch_adjust_lp;
119: }
120: else
121: ip+=2;
122:
123: )
124:
125: condbranch(?branch,f -- f83 question_branch,
126: if (f==0) {
127: IF_TOS(TOS = sp[0]);
128: )
129:
130: condbranch((next),-- cmFORTH paren_next,
131: if ((*rp)--) {
132: )
133:
134: condbranch((loop),-- fig paren_loop,
135: Cell index = *rp+1;
136: Cell limit = rp[1];
137: if (index != limit) {
138: *rp = index;
139: )
140:
141: condbranch((+loop),n -- fig paren_plus_loop,
142: /* !! check this thoroughly */
143: Cell index = *rp;
144: /* sign bit manipulation and test: (x^y)<0 is equivalent to (x<0) != (y<0) */
145: /* dependent upon two's complement arithmetic */
146: Cell olddiff = index-rp[1];
147: #ifndef undefined
148: if ((olddiff^(olddiff+n))>=0 /* the limit is not crossed */
149: || (olddiff^n)>=0 /* it is a wrap-around effect */) {
150: #else
151: #ifndef MAXINT
152: #define MAXINT ((((Cell)1)<<(8*sizeof(Cell)-1))-1)
153: #endif
154: if(((olddiff^MAXINT) >= n) ^ ((olddiff+n) < 0)) {
155: #endif
156: #ifdef i386
157: *rp += n;
158: #else
159: *rp = index + n;
160: #endif
161: IF_TOS(TOS = sp[0]);
162: )
163:
164: condbranch((s+loop),n -- new paren_symmetric_plus_loop,
165: ""The run-time procedure compiled by S+LOOP. It loops until the index
166: crosses the boundary between limit and limit-sign(n). I.e. a symmetric
167: version of (+LOOP).""
168: /* !! check this thoroughly */
169: Cell index = *rp;
170: Cell diff = index-rp[1];
171: Cell newdiff = diff+n;
172: if (n<0) {
173: diff = -diff;
174: newdiff = -newdiff;
175: }
176: if (diff>=0 || newdiff<0) {
177: #ifdef i386
178: *rp += n;
179: #else
180: *rp = index + n;
181: #endif
182: IF_TOS(TOS = sp[0]);
183: )
184:
185: unloop -- core
186: rp += 2;
187: :
188: r> rdrop rdrop >r ;
189:
190: (for) ncount -- cmFORTH paren_for
191: /* or (for) = >r -- collides with unloop! */
192: *--rp = 0;
193: *--rp = ncount;
194: :
195: r> swap 0 >r >r >r ;
196:
197: (do) nlimit nstart -- fig paren_do
198: /* or do it in high-level? 0.09/0.23% */
199: *--rp = nlimit;
200: *--rp = nstart;
201: :
202: r> -rot swap >r >r >r ;
203:
204: (?do) nlimit nstart -- core-ext paren_question_do
205: *--rp = nlimit;
206: *--rp = nstart;
207: if (nstart == nlimit) {
208: IF_TOS(TOS = sp[0]);
209: goto branch;
210: }
211: else {
212: ip++;
213: }
214:
215: i -- n core,fig
216: n = *rp;
217:
218: j -- n core
219: n = rp[2];
220:
221: \ digit is high-level: 0/0%
222:
223: (emit) c -- fig paren_emit
224: putchar(c);
225: emitcounter++;
226:
227: (type) c_addr n -- fig paren_type
228: fwrite(c_addr,sizeof(Char),n,stdout);
229: emitcounter += n;
230:
231: (key) -- n fig paren_key
232: fflush(stdout);
233: /* !! noecho */
234: n = key();
235:
236: key? -- n fig key_q
237: fflush(stdout);
238: n = key_query;
239:
240: cr -- fig
241: puts("");
242: :
243: $0A emit ;
244:
245: move c_from c_to ucount -- core
246: memmove(c_to,c_from,ucount);
247: /* make an Ifdef for bsd and others? */
248: :
249: >r 2dup u< IF r> cmove> ELSE r> cmove THEN ;
250:
251: cmove c_from c_to u -- string
252: while (u-- > 0)
253: *c_to++ = *c_from++;
254: :
255: bounds ?DO dup c@ I c! 1+ LOOP drop ;
256:
257: cmove> c_from c_to u -- string c_move_up
258: while (u-- > 0)
259: c_to[u] = c_from[u];
260: :
261: dup 0= IF drop 2drop exit THEN
262: rot over + -rot bounds swap 1-
263: DO 1- dup c@ I c! -1 +LOOP drop ;
264:
265: fill c_addr u c -- core
266: memset(c_addr,c,u);
267: :
268: -rot bounds
269: ?DO dup I c! LOOP drop ;
270:
271: compare c_addr1 u1 c_addr2 u2 -- n string
272: ""Compare the strings lexicographically. If they are equal, n is 0; if
273: the first string is smaller, n is -1; if the first string is larger, n
274: is 1. Currently this is based on the machine's character
275: comparison. In the future, this may change to considering the current
276: locale and its collation order.""
277: n = memcmp(c_addr1, c_addr2, u1<u2 ? u1 : u2);
278: if (n==0)
279: n = u1-u2;
280: if (n<0)
281: n = -1;
282: else if (n>0)
283: n = 1;
284: :
285: rot 2dup - >r min swap -text dup
286: IF rdrop
287: ELSE drop r@ 0>
288: IF rdrop -1
289: ELSE r> 1 and
290: THEN
291: THEN ;
292:
293: -text c_addr1 u c_addr2 -- n new dash_text
294: n = memcmp(c_addr1, c_addr2, u);
295: if (n<0)
296: n = -1;
297: else if (n>0)
298: n = 1;
299: :
300: swap bounds
301: ?DO dup c@ I c@ = WHILE 1+ LOOP drop 0
302: ELSE c@ I c@ - unloop THEN -text-flag ;
303: : -text-flag ( n -- -1/0/1 )
304: dup 0< IF drop -1 ELSE 0> IF 1 ELSE 0 THEN THEN ;
305:
306: capscomp c_addr1 u c_addr2 -- n new
307: Char c1, c2;
308: for (;; u--, c_addr1++, c_addr2++) {
309: if (u == 0) {
310: n = 0;
311: break;
312: }
313: c1 = toupper(*c_addr1);
314: c2 = toupper(*c_addr2);
315: if (c1 != c2) {
316: if (c1 < c2)
317: n = -1;
318: else
319: n = 1;
320: break;
321: }
322: }
323: :
324: swap bounds
325: ?DO dup c@ toupper I c@ toupper = WHILE 1+ LOOP drop 0
326: ELSE c@ toupper I c@ toupper - unloop THEN -text-flag ;
327:
328: -trailing c_addr u1 -- c_addr u2 string dash_trailing
329: u2 = u1;
330: while (c_addr[u2-1] == ' ')
331: u2--;
332: :
333: BEGIN 1- 2dup + c@ bl = WHILE
334: dup 0= UNTIL ELSE 1+ THEN ;
335:
336: /string c_addr1 u1 n -- c_addr2 u2 string slash_string
337: c_addr2 = c_addr1+n;
338: u2 = u1-n;
339: :
340: tuck - >r + r> dup 0< IF - 0 THEN ;
341:
342: + n1 n2 -- n core,fig plus
343: n = n1+n2;
344:
345: - n1 n2 -- n core,fig minus
346: n = n1-n2;
347: :
348: negate + ;
349:
350: negate n1 -- n2 core,fig
351: /* use minus as alias */
352: n2 = -n1;
353: :
354: invert 1+ ;
355:
356: 1+ n1 -- n2 core one_plus
357: n2 = n1+1;
358: :
359: 1 + ;
360:
361: 1- n1 -- n2 core one_minus
362: n2 = n1-1;
363: :
364: 1 - ;
365:
366: max n1 n2 -- n core
367: if (n1<n2)
368: n = n2;
369: else
370: n = n1;
371: :
372: 2dup < IF swap THEN drop ;
373:
374: min n1 n2 -- n core
375: if (n1<n2)
376: n = n1;
377: else
378: n = n2;
379: :
380: 2dup > IF swap THEN drop ;
381:
382: abs n1 -- n2 core
383: if (n1<0)
384: n2 = -n1;
385: else
386: n2 = n1;
387: :
388: dup 0< IF negate THEN ;
389:
390: * n1 n2 -- n core,fig star
391: n = n1*n2;
392: :
393: um* drop ;
394:
395: / n1 n2 -- n core,fig slash
396: n = n1/n2;
397: :
398: /mod nip ;
399:
400: mod n1 n2 -- n core
401: n = n1%n2;
402: :
403: /mod drop ;
404:
405: /mod n1 n2 -- n3 n4 core slash_mod
406: n4 = n1/n2;
407: n3 = n1%n2; /* !! is this correct? look into C standard! */
408: :
409: >r s>d r> fm/mod ;
410:
411: 2* n1 -- n2 core two_star
412: n2 = 2*n1;
413: :
414: dup + ;
415:
416: 2/ n1 -- n2 core two_slash
417: /* !! is this still correct? */
418: n2 = n1>>1;
419:
420: fm/mod d1 n1 -- n2 n3 core f_m_slash_mod
421: ""floored division: d1 = n3*n1+n2, n1>n2>=0 or 0>=n2>n1""
422: /* assumes that the processor uses either floored or symmetric division */
423: n3 = d1/n1;
424: n2 = d1%n1;
425: /* note that this 1%-3>0 is optimized by the compiler */
426: if (1%-3>0 && (d1<0) != (n1<0) && n2!=0) {
427: n3--;
428: n2+=n1;
429: }
430:
431: sm/rem d1 n1 -- n2 n3 core s_m_slash_rem
432: ""symmetric division: d1 = n3*n1+n2, sign(n2)=sign(d1) or 0""
433: /* assumes that the processor uses either floored or symmetric division */
434: n3 = d1/n1;
435: n2 = d1%n1;
436: /* note that this 1%-3<0 is optimized by the compiler */
437: if (1%-3<0 && (d1<0) != (n1<0) && n2!=0) {
438: n3++;
439: n2-=n1;
440: }
441: :
442: over >r dup >r abs -rot
443: dabs rot um/mod
444: r> 0< IF negate THEN
445: r> 0< IF swap negate swap THEN ;
446:
447: m* n1 n2 -- d core m_star
448: d = (DCell)n1 * (DCell)n2;
449: :
450: 2dup 0< and >r
451: 2dup swap 0< and >r
452: um* r> - r> - ;
453:
454: um* u1 u2 -- ud core u_m_star
455: /* use u* as alias */
456: ud = (UDCell)u1 * (UDCell)u2;
457:
458: um/mod ud u1 -- u2 u3 core u_m_slash_mod
459: u3 = ud/u1;
460: u2 = ud%u1;
461: :
462: dup IF 0 (um/mod) THEN nip ;
463: : (um/mod) ( ud ud--ud u)
464: 2dup >r >r dup 0<
465: IF 2drop 0
466: ELSE 2dup d+ (um/mod) 2* THEN
467: -rot r> r> 2over 2over du<
468: IF 2drop rot
469: ELSE dnegate d+ rot 1+ THEN ;
470:
471: m+ d1 n -- d2 double m_plus
472: d2 = d1+n;
473: :
474: s>d d+ ;
475:
476: d+ d1 d2 -- d double,fig d_plus
477: d = d1+d2;
478: :
479: >r swap >r over 2/ over 2/ + >r over 1 and over 1 and + 2/
480: r> + >r + r> 0< r> r> + swap - ;
481:
482: d- d1 d2 -- d double d_minus
483: d = d1-d2;
484: :
485: dnegate d+ ;
486:
487: dnegate d1 -- d2 double
488: /* use dminus as alias */
489: d2 = -d1;
490: :
491: invert swap negate tuck 0= - ;
492:
493: dmax d1 d2 -- d double
494: if (d1<d2)
495: d = d2;
496: else
497: d = d1;
498: :
499: 2over 2over d> IF 2swap THEN 2drop ;
500:
501: dmin d1 d2 -- d double
502: if (d1<d2)
503: d = d1;
504: else
505: d = d2;
506: :
507: 2over 2over d< IF 2swap THEN 2drop ;
508:
509: dabs d1 -- d2 double
510: if (d1<0)
511: d2 = -d1;
512: else
513: d2 = d1;
514: :
515: dup 0< IF dnegate THEN ;
516:
517: d2* d1 -- d2 double d_two_star
518: d2 = 2*d1;
519: :
520: 2dup d+ ;
521:
522: d2/ d1 -- d2 double d_two_slash
523: /* !! is this still correct? */
524: d2 = d1>>1;
525: :
526: dup 1 and >r 2/ swap 2/ [ 1 8 cells 1- lshift 1- ] Literal and
527: r> IF [ 1 8 cells 1- lshift ] Literal + THEN swap ;
528:
529: d>s d -- n double d_to_s
530: /* make this an alias for drop? */
531: n = d;
532: :
533: drop ;
534:
535: and w1 w2 -- w core,fig
536: w = w1&w2;
537:
538: or w1 w2 -- w core,fig
539: w = w1|w2;
540:
541: xor w1 w2 -- w core,fig
542: w = w1^w2;
543:
544: invert w1 -- w2 core
545: w2 = ~w1;
546: :
547: -1 xor ;
548:
549: rshift u1 n -- u2 core
550: u2 = u1>>n;
551:
552: lshift u1 n -- u2 core
553: u2 = u1<<n;
554:
555: \ comparisons(prefix, args, prefix, arg1, arg2, wordsets...)
556: define(comparisons,
557: $1= $2 -- f $6 $3equals
558: f = FLAG($4==$5);
559:
560: $1<> $2 -- f $7 $3different
561: /* use != as alias ? */
562: f = FLAG($4!=$5);
563:
564: $1< $2 -- f $8 $3less
565: f = FLAG($4<$5);
566:
567: $1> $2 -- f $9 $3greater
568: f = FLAG($4>$5);
569:
570: $1<= $2 -- f new $3less_or_equal
571: f = FLAG($4<=$5);
572:
573: $1>= $2 -- f new $3greater_or_equal
574: f = FLAG($4>=$5);
575:
576: )
577:
578: comparisons(0, n, zero_, n, 0, core, core-ext, core, core-ext)
579: comparisons(, n1 n2, , n1, n2, core, core-ext, core, core)
580: comparisons(u, u1 u2, u_, u1, u2, new, new, core, core-ext)
581: comparisons(d, d1 d2, d_, d1, d2, double, new, double, new)
582: comparisons(d0, d, d_zero_, d, 0, double, new, double, new)
583: comparisons(du, ud1 ud2, d_u_, ud1, ud2, new, new, double-ext, new)
584:
585: within u1 u2 u3 -- f core-ext
586: f = FLAG(u1-u2 < u3-u2);
587: :
588: over - >r - r> u< ;
589:
590: sp@ -- a_addr fig spat
591: a_addr = sp+1;
592:
593: sp! a_addr -- fig spstore
594: sp = a_addr;
595: /* works with and without TOS caching */
596:
597: rp@ -- a_addr fig rpat
598: a_addr = rp;
599:
600: rp! a_addr -- fig rpstore
601: rp = a_addr;
602:
603: fp@ -- f_addr new fp_fetch
604: f_addr = fp;
605:
606: fp! f_addr -- new fp_store
607: fp = f_addr;
608:
609: ;s -- fig semis
610: ip = (Xt *)(*rp++);
611:
612: >r w -- core,fig to_r
613: *--rp = w;
614:
615: r> -- w core,fig r_from
616: w = *rp++;
617:
618: r@ -- w core,fig r_fetch
619: /* use r as alias */
620: /* make r@ an alias for i */
621: w = *rp;
622:
623: rdrop -- fig
624: rp++;
625:
626: i' -- w fig i_tick
627: w=rp[1];
628:
629: 2>r w1 w2 -- core-ext two_to_r
630: *--rp = w1;
631: *--rp = w2;
632:
633: 2r> -- w1 w2 core-ext two_r_from
634: w2 = *rp++;
635: w1 = *rp++;
636:
637: 2r@ -- w1 w2 core-ext two_r_fetch
638: w2 = rp[0];
639: w1 = rp[1];
640:
641: 2rdrop -- new two_r_drop
642: rp+=2;
643:
644: over w1 w2 -- w1 w2 w1 core,fig
645:
646: drop w -- core,fig
647:
648: swap w1 w2 -- w2 w1 core,fig
649:
650: dup w -- w w core,fig
651:
652: rot w1 w2 w3 -- w2 w3 w1 core rote
653:
654: -rot w1 w2 w3 -- w3 w1 w2 fig not_rote
655: :
656: rot rot ;
657:
658: nip w1 w2 -- w2 core-ext
659: :
660: swap drop ;
661:
662: tuck w1 w2 -- w2 w1 w2 core-ext
663: :
664: swap over ;
665:
666: ?dup w -- w core question_dupe
667: if (w!=0) {
668: IF_TOS(*sp-- = w;)
669: #ifndef USE_TOS
670: *--sp = w;
671: #endif
672: }
673: :
674: dup IF dup THEN ;
675:
676: pick u -- w core-ext
677: w = sp[u+1];
678: :
679: 1+ cells sp@ + @ ;
680:
681: 2drop w1 w2 -- core two_drop
682: :
683: drop drop ;
684:
685: 2dup w1 w2 -- w1 w2 w1 w2 core two_dupe
686: :
687: over over ;
688:
689: 2over w1 w2 w3 w4 -- w1 w2 w3 w4 w1 w2 core two_over
690: :
691: 3 pick 3 pick ;
692:
693: 2swap w1 w2 w3 w4 -- w3 w4 w1 w2 core two_swap
694: :
695: >r -rot r> -rot ;
696:
697: 2rot w1 w2 w3 w4 w5 w6 -- w3 w4 w5 w6 w1 w2 double two_rote
698: :
699: >r >r 2swap r> r> 2swap ;
700:
701: \ toggle is high-level: 0.11/0.42%
702:
703: @ a_addr -- w fig fetch
704: w = *a_addr;
705:
706: ! w a_addr -- core,fig store
707: *a_addr = w;
708:
709: +! n a_addr -- core,fig plus_store
710: *a_addr += n;
711:
712: c@ c_addr -- c fig cfetch
713: c = *c_addr;
714:
715: c! c c_addr -- fig cstore
716: *c_addr = c;
717:
718: 2! w1 w2 a_addr -- core two_store
719: a_addr[0] = w2;
720: a_addr[1] = w1;
721: :
722: tuck ! cell+ ! ;
723:
724: 2@ a_addr -- w1 w2 core two_fetch
725: w2 = a_addr[0];
726: w1 = a_addr[1];
727: :
728: dup cell+ @ swap @ ;
729:
730: d! d a_addr -- double d_store
731: /* !! alignment problems on some machines */
732: *(DCell *)a_addr = d;
733:
734: d@ a_addr -- d double d_fetch
735: d = *(DCell *)a_addr;
736:
737: cell+ a_addr1 -- a_addr2 core cell_plus
738: a_addr2 = a_addr1+1;
739: :
740: [ cell ] Literal + ;
741:
742: cells n1 -- n2 core
743: n2 = n1 * sizeof(Cell);
744: :
745: [ cell ]
746: [ 2/ dup ] [IF] 2* [THEN]
747: [ 2/ dup ] [IF] 2* [THEN]
748: [ 2/ dup ] [IF] 2* [THEN]
749: [ 2/ dup ] [IF] 2* [THEN]
750: [ drop ] ;
751:
752: char+ c_addr1 -- c_addr2 core care_plus
753: c_addr2 = c_addr1 + 1;
754: :
755: 1+ ;
756:
757: (chars) n1 -- n2 gforth paren_cares
758: n2 = n1 * sizeof(Char);
759: :
760: ;
761:
762: count c_addr1 -- c_addr2 u core
763: u = *c_addr1;
764: c_addr2 = c_addr1+1;
765: :
766: dup 1+ swap c@ ;
767:
768: (bye) n -- toolkit-ext paren_bye
769: return (Label *)n;
770:
771: system c_addr u -- n own
772: n=system(cstr(c_addr,u,1));
773:
774: getenv c_addr1 u1 -- c_addr2 u2 new
775: c_addr2 = getenv(cstr(c_addr1,u1,1));
776: u2=strlen(c_addr2);
777:
778: popen c_addr u n -- wfileid own
779: static char* mode[2]={"r","w"};
780: wfileid=(Cell)popen(cstr(c_addr,u,1),mode[n]);
781:
782: pclose wfileid -- wior own
783: wior=pclose((FILE *)wfileid);
784:
785: time&date -- nsec nmin nhour nday nmonth nyear facility-ext time_and_date
786: struct timeval time1;
787: struct timezone zone1;
788: struct tm *ltime;
789: gettimeofday(&time1,&zone1);
790: ltime=localtime(&time1.tv_sec);
791: nyear =ltime->tm_year+1900;
792: nmonth=ltime->tm_mon+1;
793: nday =ltime->tm_mday;
794: nhour =ltime->tm_hour;
795: nmin =ltime->tm_min;
796: nsec =ltime->tm_sec;
797:
798: ms n -- facility-ext
799: struct timeval timeout;
800: timeout.tv_sec=n/1000;
801: timeout.tv_usec=1000*(n%1000);
802: (void)select(0,0,0,0,&timeout);
803:
804: allocate u -- a_addr wior memory
805: a_addr = (Cell *)malloc(u);
806: wior = a_addr==NULL; /* !! Define a return code */
807:
808: free a_addr -- wior memory
809: free(a_addr);
810: wior = 0;
811:
812: resize a_addr1 u -- a_addr2 wior memory
813: a_addr2 = realloc(a_addr1, u);
814: wior = a_addr2==NULL; /* !! Define a return code */
815:
816: (f83find) c_addr u f83name1 -- f83name2 new paren_f83find
817: for (; f83name1 != NULL; f83name1 = f83name1->next)
818: if (F83NAME_COUNT(f83name1)==u &&
819: strncasecmp(c_addr, f83name1->name, u)== 0 /* or inline? */)
820: break;
821: f83name2=f83name1;
822: :
823: BEGIN dup WHILE
824: >r dup r@ cell+ c@ $1F and =
825: IF 2dup r@ cell+ char+ capscomp 0=
826: IF 2drop r> EXIT THEN THEN
827: r> @
828: REPEAT nip nip ;
829:
830: (hashfind) c_addr u a_addr -- f83name2 new paren_hashfind
831: F83Name *f83name1;
832: f83name2=NULL;
833: while(a_addr != NULL)
834: {
835: f83name1=(F83Name *)(a_addr[1]);
836: a_addr=(Cell *)(a_addr[0]);
837: if (F83NAME_COUNT(f83name1)==u &&
838: strncasecmp(c_addr, f83name1->name, u)== 0 /* or inline? */)
839: {
840: f83name2=f83name1;
841: break;
842: }
843: }
844: :
845: BEGIN dup WHILE
846: 2@ >r >r dup r@ cell+ c@ $1F and =
847: IF 2dup r@ cell+ char+ capscomp 0=
848: IF 2drop r> rdrop EXIT THEN THEN
849: rdrop r>
850: REPEAT nip nip ;
851:
852: (hashkey) c_addr u1 -- u2 new paren_hashkey
853: u2=0;
854: while(u1--)
855: u2+=(Cell)toupper(*c_addr++);
856: :
857: 0 -rot bounds ?DO I c@ toupper + LOOP ;
858:
859: (hashkey1) c_addr u ubits -- ukey new paren_hashkey1
860: ""ukey is the hash key for the string c_addr u fitting in ubits bits""
861: /* this hash function rotates the key at every step by rot bits within
862: ubits bits and xors it with the character. This function does ok in
863: the chi-sqare-test. Rot should be <=7 (preferably <=5) for
864: ASCII strings (larger if ubits is large), and should share no
865: divisors with ubits.
866: */
867: unsigned rot = ((char []){5,0,1,2,3,4,5,5,5,5,3,5,5,5,5,7,5,5,5,5,7,5,5,5,5,6,5,5,5,5,7,5,5})[ubits];
868: Char *cp = c_addr;
869: for (ukey=0; cp<c_addr+u; cp++)
870: ukey = ((((ukey<<rot) | (ukey>>(ubits-rot)))
871: ^ toupper(*cp))
872: & ((1<<ubits)-1));
873: :
874: dup rot-values + c@ over 1 swap lshift 1- >r
875: tuck - 2swap r> 0 2swap bounds
876: ?DO dup 4 pick lshift swap 3 pick rshift or
877: I c@ toupper xor
878: over and LOOP
879: nip nip nip ;
880: Create rot-values
881: 5 c, 0 c, 1 c, 2 c, 3 c, 4 c, 5 c, 5 c, 5 c, 5 c,
882: 3 c, 5 c, 5 c, 5 c, 5 c, 7 c, 5 c, 5 c, 5 c, 5 c,
883: 7 c, 5 c, 5 c, 5 c, 5 c, 6 c, 5 c, 5 c, 5 c, 5 c,
884: 7 c, 5 c, 5 c,
885:
886: (parse-white) c_addr1 u1 -- c_addr2 u2 new paren_parse_white
887: /* use !isgraph instead of isspace? */
888: Char *endp = c_addr1+u1;
889: while (c_addr1<endp && isspace(*c_addr1))
890: c_addr1++;
891: if (c_addr1<endp) {
892: for (c_addr2 = c_addr1; c_addr1<endp && !isspace(*c_addr1); c_addr1++)
893: ;
894: u2 = c_addr1-c_addr2;
895: }
896: else {
897: c_addr2 = c_addr1;
898: u2 = 0;
899: }
900: :
901: BEGIN dup WHILE over c@ bl <= WHILE 1 /string
902: REPEAT THEN 2dup
903: BEGIN dup WHILE over c@ bl > WHILE 1 /string
904: REPEAT THEN nip - ;
905:
906: close-file wfileid -- wior file close_file
907: wior = FILEIO(fclose((FILE *)wfileid)==EOF);
908:
909: open-file c_addr u ntype -- w2 wior file open_file
910: w2 = (Cell)fopen(cstr(c_addr, u, 1), fileattr[ntype]);
911: wior = FILEEXIST(w2 == NULL);
912:
913: create-file c_addr u ntype -- w2 wior file create_file
914: Cell fd;
915: fd = open(cstr(c_addr, u, 1), O_CREAT|O_RDWR|O_TRUNC, 0666);
916: if (fd > -1) {
917: w2 = (Cell)fdopen(fd, fileattr[ntype]);
918: assert(w2 != NULL);
919: wior = 0;
920: } else {
921: assert(fd == -1);
922: wior = FILEIO(fd);
923: w2 = 0;
924: }
925:
926: delete-file c_addr u -- wior file delete_file
927: wior = FILEEXIST(unlink(cstr(c_addr, u, 1)));
928:
929: rename-file c_addr1 u1 c_addr2 u2 -- wior file-ext rename_file
930: char *s1=cstr(c_addr2, u2, 1);
931: wior = FILEEXIST(rename(cstr(c_addr1, u1, 0), s1));
932:
933: file-position wfileid -- ud wior file file_position
934: /* !! use tell and lseek? */
935: ud = ftell((FILE *)wfileid);
936: wior = 0; /* !! or wior = FLAG(ud<0) */
937:
938: reposition-file ud wfileid -- wior file reposition_file
939: wior = FILEIO(fseek((FILE *)wfileid, (long)ud, SEEK_SET));
940:
941: file-size wfileid -- ud wior file file_size
942: struct stat buf;
943: wior = FILEEXIST(fstat(fileno((FILE *)wfileid), &buf));
944: ud = buf.st_size;
945:
946: resize-file ud wfileid -- wior file resize_file
947: wior = FILEIO(ftruncate(fileno((FILE *)wfileid), (Cell)ud));
948:
949: read-file c_addr u1 wfileid -- u2 wior file read_file
950: /* !! fread does not guarantee enough */
951: u2 = fread(c_addr, sizeof(Char), u1, (FILE *)wfileid);
952: wior = FILEIO(u2<u1 && ferror((FILE *)wfileid));
953: /* !! who performs clearerr((FILE *)wfileid); ? */
954:
955: read-line c_addr u1 wfileid -- u2 flag wior file read_line
956: /*
957: Cell c;
958: flag=-1;
959: for(u2=0; u2<u1; u2++)
960: {
961: *c_addr++ = (Char)(c = getc((FILE *)wfileid));
962: if(c=='\n') break;
963: if(c==EOF)
964: {
965: flag=FLAG(u2!=0);
966: break;
967: }
968: }
969: wior=FILEIO(ferror((FILE *)wfileid));
970: */
971: if ((flag=FLAG(!feof((FILE *)wfileid) &&
972: fgets(c_addr,u1+1,(FILE *)wfileid) != NULL))) {
973: wior=FILEIO(ferror((FILE *)wfileid));
974: u2 = strlen(c_addr);
975: u2-=((u2>0) && (c_addr[u2-1]==NEWLINE));
976: }
977: else {
978: wior=0;
979: u2=0;
980: }
981:
982: write-file c_addr u1 wfileid -- wior file write_file
983: /* !! fwrite does not guarantee enough */
984: {
985: Cell u2 = fwrite(c_addr, sizeof(Char), u1, (FILE *)wfileid);
986: wior = FILEIO(u2<u1 && ferror((FILE *)wfileid));
987: }
988:
989: flush-file wfileid -- wior file-ext flush_file
990: wior = FILEIO(fflush((FILE *) wfileid));
991:
992: comparisons(f, r1 r2, f_, r1, r2, new, new, float, new)
993: comparisons(f0, r, f_zero_, r, 0., float, new, float, new)
994:
995: d>f d -- r float d_to_f
996: r = d;
997:
998: f>d r -- d float f_to_d
999: /* !! basis 15 is not very specific */
1000: d = r;
1001:
1002: f! r f_addr -- float f_store
1003: *f_addr = r;
1004:
1005: f@ f_addr -- r float f_fetch
1006: r = *f_addr;
1007:
1008: df@ df_addr -- r float-ext d_f_fetch
1009: #ifdef IEEE_FP
1010: r = *df_addr;
1011: #else
1012: !! df@
1013: #endif
1014:
1015: df! r df_addr -- float-ext d_f_store
1016: #ifdef IEEE_FP
1017: *df_addr = r;
1018: #else
1019: !! df!
1020: #endif
1021:
1022: sf@ sf_addr -- r float-ext s_f_fetch
1023: #ifdef IEEE_FP
1024: r = *sf_addr;
1025: #else
1026: !! sf@
1027: #endif
1028:
1029: sf! r sf_addr -- float-ext s_f_store
1030: #ifdef IEEE_FP
1031: *sf_addr = r;
1032: #else
1033: !! sf!
1034: #endif
1035:
1036: f+ r1 r2 -- r3 float f_plus
1037: r3 = r1+r2;
1038:
1039: f- r1 r2 -- r3 float f_minus
1040: r3 = r1-r2;
1041:
1042: f* r1 r2 -- r3 float f_star
1043: r3 = r1*r2;
1044:
1045: f/ r1 r2 -- r3 float f_slash
1046: r3 = r1/r2;
1047:
1048: f** r1 r2 -- r3 float-ext f_star_star
1049: ""@i{r3} is @i{r1} raised to the @i{r2}th power""
1050: r3 = pow(r1,r2);
1051:
1052: fnegate r1 -- r2 float
1053: r2 = - r1;
1054:
1055: fdrop r -- float
1056:
1057: fdup r -- r r float
1058:
1059: fswap r1 r2 -- r2 r1 float
1060:
1061: fover r1 r2 -- r1 r2 r1 float
1062:
1063: frot r1 r2 r3 -- r2 r3 r1 float
1064:
1065: float+ f_addr1 -- f_addr2 float float_plus
1066: f_addr2 = f_addr1+1;
1067:
1068: floats n1 -- n2 float
1069: n2 = n1*sizeof(Float);
1070:
1071: floor r1 -- r2 float
1072: ""round towards the next smaller integral value, i.e., round toward negative infinity""
1073: /* !! unclear wording */
1074: r2 = floor(r1);
1075:
1076: fround r1 -- r2 float
1077: ""round to the nearest integral value""
1078: /* !! unclear wording */
1079: #ifdef HAVE_RINT
1080: r2 = rint(r1);
1081: #else
1082: r2 = floor(r1+0.5);
1083: /* !! This is not quite true to the rounding rules given in the standard */
1084: #endif
1085:
1086: fmax r1 r2 -- r3 float
1087: if (r1<r2)
1088: r3 = r2;
1089: else
1090: r3 = r1;
1091:
1092: fmin r1 r2 -- r3 float
1093: if (r1<r2)
1094: r3 = r1;
1095: else
1096: r3 = r2;
1097:
1098: represent r c_addr u -- n f1 f2 float
1099: char *sig;
1100: Cell flag;
1101: Cell decpt;
1102: sig=ecvt(r, u, &decpt, &flag);
1103: n=(r==0 ? 1 : decpt);
1104: f1=FLAG(flag!=0);
1105: f2=FLAG(isdigit(sig[0])!=0);
1106: memmove(c_addr,sig,u);
1107:
1108: >float c_addr u -- flag float to_float
1109: /* real signature: c_addr u -- r t / f */
1110: Float r;
1111: char *number=cstr(c_addr, u, 1);
1112: char *endconv;
1113: while(isspace(number[--u]) && u>0);
1114: switch(number[u])
1115: {
1116: case 'd':
1117: case 'D':
1118: case 'e':
1119: case 'E': break;
1120: default : u++; break;
1121: }
1122: number[u]='\0';
1123: r=strtod(number,&endconv);
1124: if((flag=FLAG(!(Cell)*endconv)))
1125: {
1126: IF_FTOS(fp[0] = FTOS);
1127: fp += -1;
1128: FTOS = r;
1129: }
1130: else if(*endconv=='d' || *endconv=='D')
1131: {
1132: *endconv='E';
1133: r=strtod(number,&endconv);
1134: if((flag=FLAG(!(Cell)*endconv)))
1135: {
1136: IF_FTOS(fp[0] = FTOS);
1137: fp += -1;
1138: FTOS = r;
1139: }
1140: }
1141:
1142: fabs r1 -- r2 float-ext
1143: r2 = fabs(r1);
1144:
1145: facos r1 -- r2 float-ext
1146: r2 = acos(r1);
1147:
1148: fasin r1 -- r2 float-ext
1149: r2 = asin(r1);
1150:
1151: fatan r1 -- r2 float-ext
1152: r2 = atan(r1);
1153:
1154: fatan2 r1 r2 -- r3 float-ext
1155: ""@i{r1/r2}=tan@i{r3}. The standard does not require, but probably
1156: intends this to be the inverse of @code{fsincos}. In gforth it is.""
1157: r3 = atan2(r1,r2);
1158:
1159: fcos r1 -- r2 float-ext
1160: r2 = cos(r1);
1161:
1162: fexp r1 -- r2 float-ext
1163: r2 = exp(r1);
1164:
1165: fexpm1 r1 -- r2 float-ext
1166: ""@i{r2}=@i{e}**@i{r1}@minus{}1""
1167: #ifdef HAVE_EXPM1
1168: extern double expm1(double);
1169: r2 = expm1(r1);
1170: #else
1171: r2 = exp(r1)-1.;
1172: #endif
1173:
1174: fln r1 -- r2 float-ext
1175: r2 = log(r1);
1176:
1177: flnp1 r1 -- r2 float-ext
1178: ""@i{r2}=ln(@i{r1}+1)""
1179: #ifdef HAVE_LOG1P
1180: extern double log1p(double);
1181: r2 = log1p(r1);
1182: #else
1183: r2 = log(r1+1.);
1184: #endif
1185:
1186: flog r1 -- r2 float-ext
1187: ""the decimal logarithm""
1188: r2 = log10(r1);
1189:
1190: falog r1 -- r2 float-ext
1191: ""@i{r2}=10**@i{r1}""
1192: #ifdef HAVE_POW10
1193: extern double pow10(double);
1194: r2 = pow10(r1);
1195: #else
1196: #ifndef M_LN10
1197: #define M_LN10 2.30258509299404568402
1198: #endif
1199: r2 = exp(r1*M_LN10);
1200: #endif
1201:
1202: fsin r1 -- r2 float-ext
1203: r2 = sin(r1);
1204:
1205: fsincos r1 -- r2 r3 float-ext
1206: ""@i{r2}=sin(@i{r1}), @i{r3}=cos(@i{r1})""
1207: r2 = sin(r1);
1208: r3 = cos(r1);
1209:
1210: fsqrt r1 -- r2 float-ext
1211: r2 = sqrt(r1);
1212:
1213: ftan r1 -- r2 float-ext
1214: r2 = tan(r1);
1215: :
1216: fsincos f/ ;
1217:
1218: fsinh r1 -- r2 float-ext
1219: r2 = sinh(r1);
1220: :
1221: fexpm1 fdup fdup 1. d>f f+ f/ f+ f2/ ;
1222:
1223: fcosh r1 -- r2 float-ext
1224: r2 = cosh(r1);
1225: :
1226: fexp fdup 1/f f+ f2/ ;
1227:
1228: ftanh r1 -- r2 float-ext
1229: r2 = tanh(r1);
1230: :
1231: f2* fexpm1 fdup 2. d>f f+ f/ ;
1232:
1233: fasinh r1 -- r2 float-ext
1234: r2 = asinh(r1);
1235: :
1236: fdup fdup f* 1. d>f f+ fsqrt f/ fatanh ;
1237:
1238: facosh r1 -- r2 float-ext
1239: r2 = acosh(r1);
1240: :
1241: fdup fdup f* 1. d>f f- fsqrt f+ fln ;
1242:
1243: fatanh r1 -- r2 float-ext
1244: r2 = atanh(r1);
1245: :
1246: fdup f0< >r fabs 1. d>f fover f- f/ f2* flnp1 f2/
1247: r> IF fnegate THEN ;
1248:
1249: \ The following words access machine/OS/installation-dependent ANSI
1250: \ figForth internals
1251: \ !! how about environmental queries DIRECT-THREADED,
1252: \ INDIRECT-THREADED, TOS-CACHED, FTOS-CACHED, CODEFIELD-DOES */
1253:
1254: >body xt -- a_addr core to_body
1255: a_addr = PFA(xt);
1256:
1257: >code-address xt -- c_addr new to_code_address
1258: ""c_addr is the code address of the word xt""
1259: /* !! This behaves installation-dependently for DOES-words */
1260: c_addr = CODE_ADDRESS(xt);
1261:
1262: >does-code xt -- a_addr new to_does_code
1263: ""If xt ist the execution token of a defining-word-defined word,
1264: a_addr is the start of the Forth code after the DOES>; Otherwise the
1265: behaviour is undefined""
1266: /* !! there is currently no way to determine whether a word is
1267: defining-word-defined */
1268: a_addr = (Cell *)DOES_CODE(xt);
1269:
1270: code-address! n xt -- new code_address_store
1271: ""Creates a code field with code address c_addr at xt""
1272: MAKE_CF(xt, symbols[CF(n)]);
1273: CACHE_FLUSH(xt,PFA(0));
1274:
1275: does-code! a_addr xt -- new does_code_store
1276: ""creates a code field at xt for a defining-word-defined word; a_addr
1277: is the start of the Forth code after DOES>""
1278: MAKE_DOES_CF(xt, a_addr);
1279: CACHE_FLUSH(xt,PFA(0));
1280:
1281: does-handler! a_addr -- new does_jump_store
1282: ""creates a DOES>-handler at address a_addr. a_addr usually points
1283: just behind a DOES>.""
1284: MAKE_DOES_HANDLER(a_addr);
1285: CACHE_FLUSH(a_addr,DOES_HANDLER_SIZE);
1286:
1287: /does-handler -- n new slash_does_handler
1288: ""the size of a does-handler (includes possible padding)""
1289: /* !! a constant or environmental query might be better */
1290: n = DOES_HANDLER_SIZE;
1291:
1292: toupper c1 -- c2 new
1293: c2 = toupper(c1);
1294:
1295: \ local variable implementation primitives
1296: @local# -- w new fetch_local_number
1297: w = *(Cell *)(lp+(Cell)(*ip++));
1298:
1299: @local0 -- w new fetch_local_zero
1300: w = *(Cell *)(lp+0*sizeof(Cell));
1301:
1302: @local1 -- w new fetch_local_four
1303: w = *(Cell *)(lp+1*sizeof(Cell));
1304:
1305: @local2 -- w new fetch_local_eight
1306: w = *(Cell *)(lp+2*sizeof(Cell));
1307:
1308: @local3 -- w new fetch_local_twelve
1309: w = *(Cell *)(lp+3*sizeof(Cell));
1310:
1311: f@local# -- r new f_fetch_local_number
1312: r = *(Float *)(lp+(Cell)(*ip++));
1313:
1314: f@local0 -- r new f_fetch_local_zero
1315: r = *(Float *)(lp+0*sizeof(Float));
1316:
1317: f@local1 -- r new f_fetch_local_eight
1318: r = *(Float *)(lp+1*sizeof(Float));
1319:
1320: laddr# -- c_addr new laddr_number
1321: /* this can also be used to implement lp@ */
1322: c_addr = (Char *)(lp+(Cell)(*ip++));
1323:
1324: lp+!# -- new lp_plus_store_number
1325: ""used with negative immediate values it allocates memory on the
1326: local stack, a positive immediate argument drops memory from the local
1327: stack""
1328: lp += (Cell)(*ip++);
1329:
1330: lp- -- new minus_four_lp_plus_store
1331: lp += -sizeof(Cell);
1332:
1333: lp+ -- new eight_lp_plus_store
1334: lp += sizeof(Float);
1335:
1336: lp+2 -- new sixteen_lp_plus_store
1337: lp += 2*sizeof(Float);
1338:
1339: lp! c_addr -- new lp_store
1340: lp = (Address)c_addr;
1341:
1342: >l w -- new to_l
1343: lp -= sizeof(Cell);
1344: *(Cell *)lp = w;
1345:
1346: f>l r -- new f_to_l
1347: lp -= sizeof(Float);
1348: *(Float *)lp = r;
1349:
1350: up! a_addr -- new up_store
1351: up0=up=(char *)a_addr;
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to primitives CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [gforth] / gforth