Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 1140, Wed Mar 13 18:59:03 2002 UTC	revision 1253, Fri Jul 5 19:54:58 2002 UTC
#	Line 11	Line 11
11	Date: yyyy/mm/dd	Date: yyyy/mm/dd
12	Tag: <post-commit CVS tag>	Tag: <post-commit CVS tag>
13	Description:	Description:
14
15		----------------------------------------------------------------------
16		Name: Matthias Blume
17		Date: 2002/07/05 16:00 EDT
18		Tag: Release_110_41
19		Description:
20
21		New working release.  New bootfiles.
22
23		----------------------------------------------------------------------
24		Name: Matthias Blume
25		Date: 2002/07/05 10:25:00 EDT
26		Tag: blume-20020705-btimp
27		Description:
28
29		Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
30		other clients can set up backtracing support.
31
32		----------------------------------------------------------------------
33		Name: Matthias Blume
34		Date: 2002/06/25 17:23:00 EDT
35		Tag: blume-20020625-fpmax
36		Description:
37
38		Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
39		case.  (The sense of the isNaN test was reversed -- which made min and
40		max always return their first argument.)
41
42		----------------------------------------------------------------------
43		Name: Matthias Blume
44		Date: 2002/06/11
45		Tag: blume-20020611-unixpath
46		Description:
47
48		Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
49
50		----------------------------------------------------------------------
51		Name: Matthias Blume
52		Date: 2002/06/10 16:35:00 EDT
53		Tag: blume-20020610-ieeereal
54		Description:
55
56		I back-ported my implementation of IEEEReal.fromString from the
57		idlbasis-devel branch so that we can test it.
58
59		Another small change is that ppDec tries to give more information
60		than just "<sig>" in the case of functors.  However, this code is
61		broken in some mysterious way if the functor's body's signature
62		has not been declared by ascription but gets inferred from the
63		implementation.  This needs fixing...
64
65		----------------------------------------------------------------------
66		Name: Matthias Blume
67		Date: 2002/05/31
68		Tag: blume-20020531-btrace-mode
69		Description:
70
71		Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
72		the wayside when I switched over to using Controls everywhere.)
73
74		----------------------------------------------------------------------
75		Name: Lal George
76		Date: 2002/05/23 12:21:40 EDT
77		Tag: george-20020523-visual-labels
78		Description:
79
80		Labels are now displayed in the graphical output to make
81		the fall-through and target blocks obvious.
82
83		----------------------------------------------------------------------
84		Name: Matthias Blume
85		Date: 2002/05/22 11:03:00 EDT
86		Tag: blume-20020522-shrink
87		Description:
88
89		John tweaked yesterday's fix for 1131 to handle an out-of-memory
90		situation that comes up when allocating huge arrays.
91
92		----------------------------------------------------------------------
93		Name: Matthias Blume
94		Date: 2002/05/21 16:00:00 EDT
95		Tag: Release_110_40
96		Description:
97
98		New working release (110.40).  New bootfiles.
99
100		[Also: John Reppy fixed GC bug 1131.]
101
102		----------------------------------------------------------------------
103		Name: Matthias Blume
104		Date: 2002/05/21 12:35:00 EDT
105		Tag: blume-20020521-cmdoc
106		Description:
107
108		CM documentation update.
109
110		----------------------------------------------------------------------
111		Name: Matthias Blume
112		Date: 2002/05/21 10:55:00 EDT
113		Tag: blume-20020521-misc
114		Description:
115
116		- John tweaked runtime to be silent on heap export (except when
117		GC messages are on).
118		- I added a few more things (cross-compiling versions of CMB) to
119		config/preloads (as suggestions).
120
121		----------------------------------------------------------------------
122		Name: Matthias Blume
123		Date: 2002/05/20 22:25:00 EDT
124		Tag: blume-20020520-controls
125		Description:
126
127		- Added ControlUtil structure to control-lib.cm.
128		- Use it throughout.
129		- Used Controls facility to define MLRISC controls (as opposed to
130		registering MLRISC control ref cells with Controls after the fact)
131		- Fixed messed-up controls priorities.
132
133		* Removed again all the stuff from config/preloads that one wouldn't
134		be able to preload at the time the initial heap image is built.
135		(Many libraries, e.g., CML, do not exist yet at this time.  The
136		only libraries that can be preloaded via config/preloads are those
137		that come bundled with the bootfiles.)
138
139		----------------------------------------------------------------------
140		Name: Matthias Blume
141		Date: 2002/05/20 10:59:00 EDT
142		Tag: blume-20020520-preloads
143		Description:
144
145		Added a lot of commented-out suggestions for things to be included
146		in config/preloads.
147
148		----------------------------------------------------------------------
149		Name: Allen Leung
150		Date: 2002/05/18 14:20:00 EDT
151		Tag: leunga-20020518-mdl
152		Description:
153
154		o Made the mdl tool stuff compile and run again.
155		o I've disabled all the stuff that depends on RTL specifications; they
156		are all badly broken anyway.
157
158		----------------------------------------------------------------------
159		Name: Matthias Blume
160		Date: 2002/05/17 16:49:00 EDT
161		Tag: blume-20020517-controls
162		Description:
163
164		0. John Reppy made several modifications to the SML/NJ library.
165		In particular, there is a shiny new controls-lib.cm.
166
167		1. Pushed new controls interface through compiler so that everything
168		compiles again.
169
170		2. Added FormatComb and FORMAT_COMB to the CML version of the
171		SML/NJ library (so that CML compiles again).
172
173		3. Modified init scripts because XXX_DEFAULT environment variables
174		are no longer with us.  (Boot-time initialization is now done
175		using the same environment variables that are also used for
176		startup-time initialization of controls.)
177
178		----------------------------------------------------------------------
179		Name: Lal George
180		Date: 2002/05/15 09:20:10 EDT
181		Tag: george-20020515-pseudo-op-decls
182		Description:
183
184		All pseudo-ops emitted before the first segment declaration
185		such as TEXT, DATA, and BSS directives are assumed to be global
186		declarations and are emitted first in the assembly file. This is
187		useful in a number of situations where one has pseudo-ops that are not
188		specific to any segment, and also works around the constraint that one
189		cannot have client pseudo-ops in the TEXT segment.
190
191		Because no segment is associated with these declarations it is
192		an error to allocate any space or objects before the first segment
193		directive and an exception will be raised. However, we cannot make
194		this check for client pseudo-ops.
195
196		These top level declarations are a field in the CFG graph_info.
197		In theory you can continue to add to this field after the CFG has been
198		built -- provided you know what you are doing;-)
199
200		----------------------------------------------------------------------
201		Name: Matthias Blume
202		Date: 2002/05/13 16:40:00 EDT
203		Tag: blume-20020513-pp-etc
204		Description:
205
206		A few minor bugfixes:
207
208		- Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
209		(Bogus printouts for redefined bindings still occur.  Compiler
210		bug should no longer occur now.  We need to redo the prettyprinter
211		from scratch.)
212
213		- CM pathname printer now also adds escape sequences for ( and )
214
215		- commend and docu fixes for ml-nlffi
216
217		----------------------------------------------------------------------
218		Name: Matthias Blume
219		Date: 2002/05/10 16:40:00 EDT
220		Tag: blume-20020510-erg-textio
221		Description:
222
223		Applied the following bugfix provided by Emden Gansner:
224
225		Output is corrupted when outputSubstr is used rather than output.
226
227		The problem occurs when a substring
228
229		ss = (s, dataStart, dataLen)
230
231		where dataStart > 0, fills a stream buffer with avail bytes left.
232		avail bytes of s, starting at index dataStart, are copied into the
233		buffer, the buffer is flushed, and then the remaining dataLen-avail
234		bytes of ss are copied into the beginning of the buffer. Instead of
235		starting this copy at index dataStart+avail in s, the current code
236		starts the copy at index avail.
237
238		Fix:
239		In text-io-fn.sml, change line 695 from
240		val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
241		to
242		val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
243
244		----------------------------------------------------------------------
245		Name: Matthias Blume
246		Date: 2002/04/12 13:55:00 EDT
247		Tag: blume-20020412-assyntax
248		Description:
249
250		1. Grabbed newer assyntax.h from the XFree86 project.
251		2. Fiddled with how to compile X86.prim.asm without warnings.
252		3. (Very) Minor cleanup in CM.
253
254		----------------------------------------------------------------------
255		Name: Matthias Blume
256		Date: 2002/04/01 (no joke!) 17:07:00 EST
257		Tag: blume-20020401-x86div
258		Description:
259
260		Added full support for div/mod/rem/quot on the x86, using the machine
261		instruction's two results (without clumsily recomputing the remainder)
262		directly where appropriate.
263
264		Some more extensive power-of-two support was added to the x86 instruction
265		selector (avoiding expensive divs, mods, and muls where they can be
266		replaced with cheaper shifts and masks).  However, this sort of thing
267		ought to be done earlier, e.g., within the CPS optimizer so that
268		all architectures benefit from it.
269
270		The compiler compiles to a fixed point, but changes might be somewhat
271		fragile nevertheless.  Please, report any strange things that you might
272		see wrt. div/mod/quot/rem...
273
274		----------------------------------------------------------------------
275		Name: Matthias Blume
276		Date: 2002/03/29 17:22:00
277		Tag: blume-20020329-div
278		Description:
279
280		Fixed my broken div/mod logic.  Unfortunately, this means that the
281		inline code for div/mod now has one more comparison than before.
282		Fast paths (quotient > 0 or remainder = 0) are not affected, though.
283		The problem was with quotient = 0, because that alone does not tell
284		us which way the rounding went.  One then has to look at whether
285		remainder and divisor have the same sign...  :(
286
287		Anyway, I replaced the bootfiles with fresh ones...
288
289		----------------------------------------------------------------------
290		Name: Matthias Blume
291		Date: 2002/03/29 14:10:00 EST
292		Tag: blume-20020329-inlprims
293		Description:
294
295		NEW BOOTFILES!!!    Version number bumped to 110.39.3.
296
297		Primops have changed. This means that the bin/boot-file formats have
298		changed as well.
299
300		To make sure that there is no confusion, I made a new version.
301
302
303		CHANGES:
304
305		* removed REMT from mltree (remainder should never overflow).
306
307		* added primops to deal with divisions of all flavors to the frontend
308
309		* handled these primops all the way through so they map to their respective
310		MLRISC support
311
312		* used these primops in the implementation of Int, Int32, Word, Word32
313
314		* removed INLDIV, INLMOD, and INLREM as they are no longer necessary
315
316		* parameterized INLMIN, INLMAX, and INLABS by a numkind
317
318		* translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
319		floating point
320
321		* used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
322		and Real (but Real.abs maps to a separate floating-point-only primop)
323
324
325		TODO items:
326
327		* Hacked Alpha32 instruction selection, disabling the selection of REMx
328		instructions because the machine instruction encoder cannot handle
329		them.  (Hppa, PPC, and Sparc instruction selection did not handle
330		REM in the first place, and REM is supported by the x86 machine coder.)
331
332		* Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
333		selection phase.  (The two can be streamlined because the hardware
334		delivers both quotient and remainder at the same time anyway.)
335
336		* Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
337		(Currently the behavior is inconsistent both across architectures and
338		wrt. the draft Basis spec.)
339
340		* Word8 should eventually be handled natively, too.
341
342		* There seems to be one serious bug in mltree-gen.sml.  It appears, though,
343		as if there currently is no execution path that could trigger it in
344		SML/NJ.  (The assumptions underlying functions arith and promotable do not
345		hold for things like multiplication and division.)
346
347		----------------------------------------------------------------------
348		Name: Matthias Blume
349		Date: 2002/03/27 16:27:00 EST
350		Tag: blume-20020327-mlrisc-divisions
351		Description:
352
353		Added support for all four division operations (ML's div, mod, quot,
354		and rem) to MLRISC.  In the course of doing so, I also rationalized
355		the naming (no more annoying switch-around of DIV and QUOT), by
356		parameterizing the operation by div_rounding_mode (which can be either
357		DIV_TO_ZERO or DIV_TO_NEGINF).
358
359		The generic MLTreeGen functor takes care of compiling all four
360		operations down to only round-to-zero div.
361
362		Missing pieces:
363
364		* Doing something smarter than relying on MLTreeGen on architectures
365		like, e.g., the x86 where hardware division delivers both quotient and
366		remainder at the same time.  With this, the implementation of the
367		round-to-neginf operations could be further streamlined.
368
369		* Remove inlining support for div/mod/rem from the frontend and replace it
370		with primops that get carried through to the backend.  Do this for all
371		int and word types.
372
373		----------------------------------------------------------------------
374		Name: Matthias Blume
375		Date: 2002/03/25 17:25:00 EST
376		Tag: blume-20020325-divmod
377		Description:
378
379		I improved (hopefully without breaking them) the implementation of Int.div,
380		Int.mod, and Int.rem.   For this, the code in translate.sml now takes
381		advantage of the following observations:
382
383		Let  q = x quot y      r = x rem y
384		d = x div  y      m = x mod y
385
386		where "quot" is the round-to-zero version of integer division that
387		hardware usually provides.  Then we have:
388
389		r = x - q * y        where neither the * nor the - will overflow
390		d = if q >= 0 orelse x = q * y then q else q - 1
391		where neither the * nor the - will overflow
392		m = if q >= 0 orelse r = 0 then r else r + y
393		where the + will not overflow
394
395		This results in substantial simplification of the generated code.
396		The following table shows the number of CFG nodes and edges generated
397		for
398		fun f (x, y) = x OPER y
399		(* with OPER \in div, mod, quot, rem *)
400
401
402		OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
403		--------------------------------------------------------
404		div |         24 |         39 |         12 |         16
405		mod |         41 |         71 |         12 |         16
406		quot |          8 |         10 |          8 |         10
407		rem |         10 |         14 |          8 |         10
408
409
410		----------------------------------------------------------------------
411		Name: Matthias Blume
412		Date: 2002/03/25 22:06:00 EST
413		Tag: blume-20020325-cprotobug
414		Description:
415
416		Fixed a bug in cproto (c prototype decoder).
417
418		----------------------------------------------------------------------
419		Name: Matthias Blume
420		Date: 2002/03/25 16:00:00 EST
421		Tag: blume-20020325-raw-primops
422		Description:
423
424		I did some cleanup to Allen's new primop code and
425		replaced yesterday's bootfiles with new ones.
426		(But they are stored in the same place.)
427
428		----------------------------------------------------------------------
429		Name: Matthias Blume
430		Date: 2002/03/24 22:40:00 EST
431		Tag: blume-20020324-bootfiles
432		Description:
433
434		Made the bootfiles that Allen asked for.
435
436		----------------------------------------------------------------------
437		Name: Allen Leung
438		Date: 2002/03/23 15:50:00 EST
439		Tag: leunga-20020323-flint-cps-rcc-primops
440		Description:
441
442		1. Changes to FLINT primops:
443
444		(* make a call to a C-function;
445		* The primop carries C function prototype information and specifies
446		* which of its (ML-) arguments are floating point. C prototype
447		* information is for use by the backend, ML information is for
448		* use by the CPS converter. *)
449		| RAW_CCALL of { c_proto: CTypes.c_proto,
450		ml_args: ccall_type list,
451		ml_res_opt: ccall_type option,
452		reentrant : bool
453		} option
454		(* Allocate uninitialized storage on the heap.
455		* The record is meant to hold short-lived C objects, i.e., they
456		* are not ML pointers.  With the tag, the representation is
457		* the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
458		*)
459		| RAW_RECORD of {tag:bool,sz:int}
460		and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
461
462		2.  These CPS primops are now overloaded:
463
464		rawload of {kind:numkind}
465		rawstore of {kind:numkind}
466
467		The one argument form is:
468
469		rawload {kind} address
470
471		The two argument form is:
472
473		rawload {kind} [ml object, byte-offset]
474
475		3. RAW_CCALL/RCC now takes two extra arguments:
476
477		a. The first is whether the C call is reentrant, i.e., whether
478		ML state should be saved and restored.
479		b. The second argument is a string argument specifying the name of
480		library and the C function.
481
482		These things are currently not handled in the code generator, yet.
483
484		4. In CProto,
485
486		An encoding type of "bool" means "ml object" and is mapped into
487		C prototype of PTR.  Note that "bool" is different than "string",
488		even though "string" is also mapped into PTR, because "bool"
489		is assigned an CPS type of BOGt, while "string" is assigned INT32t.
490
491		5. Pickler/unpicker
492
493		Changed to handle RAW_RECORD and newest RAW_CCALL
494
495		6. MLRiscGen,
496
497		1. Changed to handle the new rawload/rawstore/rawrecord operators.
498		2. Code for handling C Calls has been moved to a new module CPSCCalls,
499		in the file CodeGen/cpscompile/cps-c-calls.sml
500
501		7. Added the conditional move operator
502
503		condmove of branch
504
505		to cps.  Generation of this is still buggy so it is currently
506		disabled.
507
508		----------------------------------------------------------------------
509		Name: Lal George
510		Date: 2002/03/22 14:18:25 EST
511		Tag: george-20020322-cps-branch-prob
512		Description:
513
514		Implemented the Ball-Larus branch prediction-heuristics, and
515		incorporated graphical viewers for control flow graphs.
516
517		Ball-Larus Heuristics:
518		---------------------
519		See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
520
521		By design it uses the Dempster-Shafer theory for combining
522		probabilities.  For example, in the function:
523
524		fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
525
526		the ball-larus heuristics predicts that the n=0 is unlikely
527		(OH-heuristic), and the 'then' branch is unlikely because of the
528		RH-heuristic -- giving the 'then' branch an even lower combined
529		probability using the Dempster-Shafer theory.
530
531		Finally, John Reppy's loop analysis in MLRISC, further lowers the
532		probability of the 'then' branch because of the loop in the else
533		branch.
534
535
536		Graphical Viewing:
537		------------------
538		I merely plugged in Allen's graphical viewers into the compiler. The
539		additional code is not much. At the top level, saying:
540
541		Control.MLRISC.getFlag "cfg-graphical-view" := true;
542
543		will display the graphical view of the control flow graph just before
544		back-patching.  daVinci must be in your path for this to work. If
545		daVinci is not available, then the default viewer can be changed
546		using:
547
548		Control.MLRISC.getString "viewer"
549
550		which can be set to "dot" or "vcg" for the corresponding viewers. Of
551		course, these viewers must be in your path.
552
553		The above will display the compilation unit at the level of clusters,
554		many of which are small, boring, and un-interesting. Also setting:
555
556		Control.MLRISC.getInt "cfg-graphical-view_size"
557
558		will display clusters that are larger than the value set by the above.
559
560
561		----------------------------------------------------------------------
562		Name: Matthias Blume
563		Date: 2002/03/21 22:20:00 EST
564		Tag: blume-20020321-kmp-bugfix
565		Description:
566
567		Changed the interface to the KMP routine in PreString and fixed
568		a minor bug in one place where it was used.
569
570		----------------------------------------------------------------------
571		Name: Allen Leung
572		Date: 2002/03/21 20:30:00 EST
573		Tag: leunga-20020321-cfg
574		Description:
575
576		Fixed a potential problem in cfg edge splitting.
577
578		----------------------------------------------------------------------
579		Name: Allen Leung
580		Date: 2002/03/21 17:15:00 EST
581		Tag: leunga-20020321-x86-fp-cfg
582		Description:
583
584		1. Recoded the buggy parts of x86-fp.
585
586		a. All the block reordering code has been removed.
587		We now depend on the block placement phases to do this work.
588
589		b. Critical edge splitting code has been simplified and moved into the
590		CFG modules, as where they belong.
591
592		Both of these were quite buggy and complex.  The code is now much, much
593		simpler.
594
595		2. X86 backend.
596
597		a. Added instructions for 64-bit support.  Instruction selection for
598		64-bit has not been committed, however, since that
599		requires changes to MLTREE which haven't been approved by
600		Lal and John.
601
602		b. Added support for FUCOMI and FUCOMIP when generating code for
603		PentiumPro and above.  We only generate these instructions in
604		the fast-fp mode.
605
606		c. Added cases for JP and JNP in X86FreqProps.
607
608		3. CFG
609
610		CFG now has a bunch of methods for edge splitting and merging.
611
612		4. Machine description.
613
614		John's simplification of MLTREE_BASIS.fcond broke a few machine
615		description things:
616
617		rtl-build.{sig,sml} and hppa.mdl fixed.
618
619		NOTE: the machine description stuff in the repository is still broken.
620		Again, I can't put my fixes in because that involves
621		changes to MLTREE.
622
623		----------------------------------------------------------------------
624		Name: Matthias Blume
625		Date: 2002/03/20 15:55:00 EST
626		Tag: blume-20020320-kmp
627		Description:
628
629		Implemented Knuth-Morris-Pratt string matching in PreString and used
630		it for String.isSubstring, Substring.isSubstring, and
631		Substring.position.
632
633		(Might need some stress-testing.  Simple examples worked fine.)
634
635		----------------------------------------------------------------------
636		Name: Matthias Blume
637		Date: 2002/03/19 16:37:00 EST
638		Tag: blume-20020319-witnesses
639		Description:
640
641		Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
642
643		This implements a generic mechanism for changing constness qualifiers
644		anywhere within big C types without resorting to outright "casts".
645		(So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
646		modify the constness at the outermost level.)
647		The implementation of "convert" is based on the idea of "witness"
648		values -- values that are not used by the operation but whose types
649		"testify" to their applicability.  On the implementation side, "convert"
650		is simply a projection (returning its second curried argument).  With
651		cross-module inlining, it should not result in any machine code being
652		generated.
653
654		----------------------------------------------------------------------
655		Name: Matthias Blume
656		Date: 2002/03/15 16:40:00 EST
657		Tag: blume-20020315-basis
658		Description:
659
660		Provided (preliminary?) implementations for
661
662		{String,Substring}.{concatWith,isSuffix,isSubstring}
663
664		and
665
666		Substring.full
667
668		Those are in the Basis spec but they were missing in SML/NJ.
669
670		----------------------------------------------------------------------
671		Name: Matthias Blume
672		Date: 2002/03/14 21:30:00 EST
673		Tag: blume-20020314-controls
674		Description:
675
676		Controls:
677		---------
678
679		1. Factored out the recently-added Controls : CONTROLS stuff and put
680		it into its own library $/controls-lib.cm.  The source tree for
681		this is under src/smlnj-lib/Controls.
682
683		2. Changed the names of types and functions in this interface, so they
684		make a bit more "sense":
685
686		module -> registry
687		'a registry -> 'a group
688
689		3. The interface now deals in ref cells only.  The getter/setter interface
690		is (mostly) gone.
691
692		4. Added a function that lets one register an already-existing ref cell.
693
694		5. Made the corresponding modifications to the rest of the code so that
695		everything compiles again.
696
697		6. Changed the implementation of Controls.MLRISC back to something closer
698		to the original.  In particular, this module (and therefore MLRISC)
699		does not depend on Controls.  There now is some link-time code in
700		int-sys.sml that registers the MLRISC controls with the Controls
701		module.
702
703		CM:
704		---
705
706		* One can now specify the lambda-split aggressiveness in init.cmi.
707
708		----------------------------------------------------------------------
709		Name: Allen Leung
710		Date: 2002/03/13 17:30:00 EST
711		Tag: leunga-20020313-x86-fp-unary
712		Description:
713
714		Bug fix for:
715
716		> leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
717		> Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
718		> - fun f(x,(y,z)) = Real.~ y;
719		> [autoloading]
720		> [autoloading done]
721		>       fchsl   (%eax), 184(%esp)
722		> Error: MLRisc bug: X86MCEmitter.emitInstr
723		>
724		> uncaught exception Error
725		>   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
726
727		The problem was that the code generator did not generate any fp registers
728		in this case, and the ra didn't know that it needed to run the X86FP phase to
729		translate the pseudo fp instruction.   This only happened with unary fp
730		operators in certain situations.
731
732	----------------------------------------------------------------------	----------------------------------------------------------------------
733	Name: Matthias Blume	Name: Matthias Blume
734	Date: 2002/03/13 14:00:00 EST	Date: 2002/03/13 14:00:00 EST
#	Line 50	Line 768
768	provides a bit more "stability":  Once CM has seen a compilation	provides a bit more "stability":  Once CM has seen a compilation
769	unit, it keeps its identity constant (as long as you do not delete	unit, it keeps its identity constant (as long as you do not delete
770	those crucial CM/GUID/* files).  This means that when you change	those crucial CM/GUID/* files).  This means that when you change
771	and interface, compiler, then go back to the old interface, and	an interface, compile, then go back to the old interface, and
772	compile again, you arrive at the original pid.	compile again, you arrive at the original pid.
773
774	There now also is a mechanism that instructs CM to use the plain	There now also is a mechanism that instructs CM to use the plain
#	Line 125	Line 843
843	----------------------------------------------------------------------	----------------------------------------------------------------------
844	Name: Allen Leung	Name: Allen Leung
845	Date: 2002/03/11 10:30:00 EST	Date: 2002/03/11 10:30:00 EST
846	Tag: leunga-20020310-runtime-string0	Tag: leunga-20020311-runtime-string0
847	Description:	Description:
848
849	The representation of the empty string now points to a	The representation of the empty string now points to a

---

Colored Diff Long Colored Diff Unidiff Context Diff Side by Side

Legend: Removed from v.1140   changed lines   Added in v.1253

---

root@smlnj-gforge.cs.uchicago.edu	ViewVC Help
Powered by ViewVC 1.0.0