Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 902, Wed Aug 15 21:17:05 2001 UTC	revision 1069, Fri Feb 15 21:00:05 2002 UTC
#	Line 8	Line 8
8	The form of an entry should be:	The form of an entry should be:
9
10	Name:	Name:
11	Date:	Date: yyyy/mm/dd
12	Tag: <post-commit CVS tag>	Tag: <post-commit CVS tag>
13	Description:	Description:
14
15	----------------------------------------------------------------------	----------------------------------------------------------------------
16	Name: Matthias Blume	Name: Matthias Blume
17		Date: 2002/02/15 16:00:00 EST
18		Tag: Release_110_39
19		Description:
20
21		Working release 110.39.  New bootfiles.
22
23		----------------------------------------------------------------------
24		Name: Matthias Blume
25		Date: 2002/02/15 14:17:00 EST
26		Tag: blume-20020215-showbindings
27		Description:
28
29		Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
30		the latter can be useful for exploring what bindings are available at
31		the interactive prompt.  (The first function returns only the list
32		of symbols that are really bound, the second prints those but also the
33		ones that CM's autoloading mechanism knows about.)
34
35		----------------------------------------------------------------------
36		Name: Matthias Blume
37		Date: 2002/02/15 12:08:00 EST
38		Tag: blume-20020215-iptrs
39		Description:
40
41		Two improvements to ml-nlffigen:
42
43		1. Write files only if they do not exist or if their current contents
44		do not coincide with what's being written.  (That is, avoid messing
45		with the time stamps unless absolutely necessary.)
46
47		2. Implement a "repository" mechanism for generated files related
48		to "incomplete pointer types".   See the README file for details.
49
50		----------------------------------------------------------------------
51		Name: Matthias Blume
52		Date: 2002/02/14 11:50:00 EST
53		Tag: blume-20020214-quote
54		Description:
55
56		Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
57		because of the new and improved tag generation scheme.  (Thanks to Allen
58		Leung for pointing it out.)
59
60		----------------------------------------------------------------------
61		Name: Lal George
62		Date: 2002/02/14 09:55:27 EST 2002
63		Tag: george-20020214-isabelle-bug
64		Description:
65
66		Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
67		of Isabelle on the x86.
68
69		From Allen:
70		-----------
71		I've found the problem:
72
73		in ra-core.sml, I use the counter "blocked" to keep track of the
74		true number of elements in the freeze queue.  When the counter goes
75		to zero, I skip examining the queue.  But I've messed up the
76		bookkeeping in combine():
77
78		else ();
79		case !ucol of
80		PSEUDO => (if !cntv > 0 then
81		(if !cntu > 0 then blocked := !blocked - 1 else ();
82		^^^^^^^^^^^^^^^^^^^^^^^
83		moveu := mergeMoveList(!movev, !moveu)
84		)
85		else ();
86
87		combine() is called to coalesce two nodes u and v.
88		I think I was thinking that if the move counts of u and v are both
89		greater than zero then after they are coalesced then one node is
90		removed from the freeze queue.  Apparently I was thinking that
91		both u and v are of low degree, but that's clearly not necessarily true.
92
93
94		02/12/2002:
95		Here's the patch.  HOL now compiles.
96
97		I don't know how this impact on performance (compile
98		time or runtime).  This bug caused the RA (especially on the x86)
99		to go thru the potential spill phase when there are still nodes on the
100		freeze queue.
101
102
103
104
105		----------------------------------------------------------------------
106		Name: Matthias Blume
107		Date: 2002/02/13 22:40:00 EST
108		Tag: blume-20020213-fptr-rtti
109		Description:
110
111		Fixed a bug in ml-nlffigen that was introduced with one of the previous
112		updates.
113
114		----------------------------------------------------------------------
115		Name: Matthias Blume
116		Date: 2002/02/13 16:41:00 EST
117		Tag: blume-20020213-cmlpq
118		Description:
119
120		Added new priority queue export symbols (which have just been added to
121		smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
122		not compile and the installer would choke.)
123
124		----------------------------------------------------------------------
125		Name: Matthias Blume
126		Date: 2002/02/13 16:15:00 EST
127		Tag: blume-20020213-various
128		Description:
129
130		1. More tweaks to ml-nlffigen:
131
132		- better internal datastructures (resulting in slight speedup)
133		- "-match" option requires exact match
134		- "localized" gensym counters (untagged structs/unions nested within
135		other structs/unions or within typedefs get a fresh counter; their
136		tag will be prefixed by a concatenation of their parents' tags)
137		- bug fixes (related to calculation of transitive closure of types
138		to be included in the output)
139
140		2. Minor Basis updates:
141
142		- added implementations for List.collate and Option.app
143
144		----------------------------------------------------------------------
145		Name: Matthias Blume
146		Date: 2002/02/11 15:55:00 EST
147		Tag: blume-20020211-gensym
148		Description:
149
150		Added a "-gensym" option to command line of ml-nlffigen.  This can be
151		used to specify a "stem" -- a string that is inserted in all "gensym'd"
152		names (ML structure names that correspond to unnamed C structs, unions,
153		and enums), so that separate runs of ml-nlffigen do not clash.
154
155		----------------------------------------------------------------------
156		Name: Matthias Blume
157		Date: 2002/02/11 12:05:00 EST
158		Tag: blume-20020211-gensml
159		Description:
160
161		A quick fix for a problem with GenSML (in the pgraph-util library):
162		Make generation of toplevel "local" optional.  (Strictly speaking,
163		signature definitions within "local" are not legal SML.)
164
165		Other than that: updates to INSTALL and cm/TODO.
166
167		----------------------------------------------------------------------
168		Name: Matthias Blume
169		Date: 2002/02/08 15:00:00 EST
170		Tag: blume-20020208-uniquepid
171		Description:
172
173		0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
174
175		1. The installer (config/install.sh) has gotten smarter:
176
177		- Configuration options are a bit easier to specify now
178		(in config/targets).
179		- Bug in recognizing .tar.bz2 files fixed.
180		- Installer automatically resolves dependencies between
181		configuration options (e.g., if you ask for eXene, you will
182		also get cml -- regardless whether you asked for it or not).
183		- Installer can run in "quieter mode" by setting the environment
184		variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
185		"completely silent", though.
186		- Build HashCons library as part of smlnj-lib.
187
188		2. A new scheme for assigning persistent identifiers to compilation
189		units (and, by extension, to types etc.) has been put into place.
190		This fixes a long-standing bug where types and even dynamic values
191		can get internally confused, thereby compromising type safety
192		(abstraction) and dynamic correctness.  See
193
194		http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
195
196		for an example of how things could go wrong until now.
197
198		The downside of the new scheme is that pids are not quite as
199		persistent as they used to be: CM will generate a fresh pid
200		for every compilation unit that it thinks it sees for the first
201		time.  That means that if you compile starting from a clean, fresh
202		source tree at two different times, you end up with different
203		binaries.
204
205		Cutoff recompilation, however, has not been compromised because
206		CM keeps pid information in special caches between runs.
207
208		----------------------------------------------------------------------
209		Name: Lal George
210		Date: 2002/02/07 15:34:13 EST 2002
211		Tag: <none>
212		Description:
213
214		Compilers that generate assembly code may produce  global labels
215		whose value is resolved at link time. The various peephole optimization
216		modules did not take this in account.
217
218		TODO. The Labels.addrOf function should really return an option
219		type so that clients are forced to deal with this issue, rather
220		than an exception being raised.
221
222		----------------------------------------------------------------------
223		Name: Lal George
224		Date: 2002/02/06 13:55:02 EST
225		Tag: george-20020206-ra-breakup
226		Description:
227
228		1. A bug fix from Allen.
229
230		A typo causes extra fstp %st(0)'s to be generated at compensation
231		edges, which might cause stack underflow traps at runtime.  This
232		occurs in fft where there are extraneous fstps right before the 'into'
233		trap instruction (in this case they are harmless since none of the
234		integers overflow.)
235
236		2. Pulled out various utility modules that were embedded in the modules
237		of the register allocator. I need these modules for other purposes, but
238		they are not complete enough to put into a library (just yet).
239		----------------------------------------------------------------------
240		Name: Matthias Blume
241		Date: 2002/01/31 16:05:00 EST
242		Tag: blume-20020131-sparc-ccalls
243		Description:
244
245		1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
246		of extra stack space by mistake.  Fixed.
247
248		2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
249
250		----------------------------------------------------------------------
251		Name: Allen Leung
252		Date: 2002/01/30
253		Tag: leunga-20020130-nowhere-bug-fix
254		Description:
255
256		MLRISC bug fixes:
257		1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
258		2. MachineInt.fromString was negating its value.
259
260		----------------------------------------------------------------------
261		Name: Matthias Blume
262		Date: 2002/01/29
263		Tag: blume-20020129-INSTALL
264		Description:
265
266		- Added somewhat detailed installation instructions (file INSTALL).
267		- Fixed curl-detection bug in config/install.sh.
268		- It is now possible to select the URL getter using the URLGETTER
269		environment variable:
270
271		not set / "unknown"      --> automatic detection (script tries wget,
272		curl, and lynx)
273		"wget" / "curl" / "lynx" --> use the specified program (script "knows"
274		how to properly invoke them)
275		other                    --> use $URLGETTER directly, it must take
276		precisely two command-line arguments
277		(source URL and destination file name)
278
279		----------------------------------------------------------------------
280		Name: Matthias Blume
281		Date: 2002/01/28
282		Tag: blume-20020128-sparc-ccalls
283		Description:
284
285		- Fixed problem with calculation of "used" registers in sparc-c-calls.
286		- Make use of the allocParam argument in sparc-c-calls.
287
288		----------------------------------------------------------------------
289		Name: Matthias Blume
290		Date: 2002/01/28
291		Tag: blume-20020128-allocParam
292		Description:
293
294		John Reppy:  Changes c-calls API to accept client-callback for
295		allocating extra stack space.
296		me: Corresponding changes to mlriscGen (using a dummy argument that
297		does not change the current behavior).
298
299		----------------------------------------------------------------------
300		Name: Matthias Blume
301		Date: 2002/01/28 12:00:00
302		Tag: Release_110_38
303		Description:
304
305		This time for real!!!
306
307		----------------------------------------------------------------------
308		Name: Matthias Blume
309		Date: 2002/01/28 10:56:00 EST
310		Tag: blume-20020128-retraction
311		Description:
312
313		0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
314		with blume-Release_110_38-retracted.)
315
316		1. Fixed a problem with incorrect rounding modes in real64.sml.
317		(Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
318
319		2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
320		and enums fixed.  The naming of corresponding ML identifiers should
321		now be consistent again.
322
323		----------------------------------------------------------------------
324		Name: Allen Leung
325		Date: 2002/01/27
326		Tag: leunga-20020127-nowhere
327		Description:
328
329		Added a target called nowhere in the configuration scripts.
330		Enabling this will build the MLRISC 'nowhere' tool (for translating
331		programs with where-clauses into legal SML code) during installation.
332
333		----------------------------------------------------------------------
334		Name: Matthias Blume
335		Date: 2002/01/25 21:27:00 EST
336		Tag: blume-Release_110_38-retracted
337		Description:
338
339		Call it a (working) release!  Version is 110.38. Bootfiles are ready.
340
341		README will be added later.
342
343		!!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
344		(Release_110_38) removed.  Reason: Last-minute bug fixes.
345
346		----------------------------------------------------------------------
347		Name: Matthias Blume
348		Date: 2002/01/25
349		Tag: blume-20020125-ffi
350		Description:
351
352		A large number of tweaks and improvements to ml-nlffi-lib and
353		ml-nlffigen:
354
355		- ML represenation types have been streamlined
356		- getter and setter functions work with concrete values, not abstract
357		ones where possible
358		- ml-nlffigen command line more flexible (see README file there)
359		- some bugs have been fixed (hopefully)
360
361		----------------------------------------------------------------------
362		Name: Lal George
363		Date: 2002/01/24
364		Tag: george-20020124-risc-ra-interface
365		Description:
366
367		There is a dramatic simplification in the interface to the
368		register allocator for RISC architectures as a result of making
369		parallel copy instructions explicit.
370
371		----------------------------------------------------------------------
372		Name: Matthias Blume
373		Date: 2002/01/22
374		Tag: blume-20020122-x86-ccalls
375		Description:
376
377		Bug fix for c-calls on x86 (having to do with how char- and
378		short-arguments are being handled).
379
380		----------------------------------------------------------------------
381		Name: Matthias Blume
382		Date: 2002/01/21
383		Tag: blume-20020121-ff
384		Description:
385
386		Another day of fiddling with the FFI...
387
388		1. Bug fix/workaround:  CKIT does not complain about negative array
389		dimensions, so ml-nlffigen has to guard itself against this possibility.
390		(Otherwise a negative dimension would send it into an infinite loop.)
391
392		2. Some of the abstract types (light objects, light pointers, most "base"
393		types) in structure C are now eqtypes.
394
395		3. Added constructors and test functions for NULL function pointers.
396
397		----------------------------------------------------------------------
398		Name: Matthias Blume
399		Date: 2002/01/18
400		Tag: blume-20020118-ready-for-new-release
401		Description:
402
403		Made config/srcarchiveurl point to a new place.  (Will provide boot
404		files shortly.)
405
406		Maybe we christen this to be 110.38?
407
408		----------------------------------------------------------------------
409		Name: Matthias Blume
410		Date: 2002/01/18
411		Tag: blume-20020118-more-ffifiddle
412		Description:
413
414		Today's FFI fiddling:
415
416		- Provided a structure CGetSet with "convenient" versions of C.Get.* and
417		C.Set.* that use concrete (MLRep.*) arguments and results instead
418		of abstract ones.
419
420		- Provided word-style bit operations etc. for "int" representation
421		types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
422		and Long.
423
424		----------------------------------------------------------------------
425		Name: Matthias Blume
426		Date: 2002/01/18
427		Tag: blume-20020118-use-x86-fp
428		Description:
429
430		Now that x86-fast-fp seems to be working, I turned it back on again
431		by default.  (Seems to work fine now, even with the FFI.)
432
433		Other than that, I added some documentation about the FFI to
434		src/ml-nlffigen/README and updated the FFI test examples in
435		src/ml-nlffi-lib/Tests/*.
436
437		----------------------------------------------------------------------
438		Name: Allen Leung
439		Date: 2002/01/17
440		Tag: leunga-20020117-x86-fast-fp-call
441		Description:
442
443		1. Fixed a problem with handling return fp values when x86's fast fp
444		mode is turned on.
445
446		2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
447		of %f32.
448
449		3. Added a constructor INT32lit to the ast of MLRISC tools.
450
451		----------------------------------------------------------------------
452		Name: Matthias Blume
453		Date: 2002/01/16
454		Tag: blume-20020116-ffifiddle
455		Description:
456
457		More fiddling with the FFI interface:
458
459		- Make constness 'c instead of rw wherever possible.  This eliminates
460		the need for certain explicit coercions.  (However, due to ML's
461		value polymorphism, there will still be many cases where explicit
462		coercions are necessary.  Phantom types are not the whole answer
463		to modeling a subtyping relationship in ML.)
464
465		- ro/rw coersions for pointers added.  (Avoids the detour through */&.)
466
467		- "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
468		clumsy workaround for varargs problem.)
469
470		----------------------------------------------------------------------
471		Name: Lal George
472		Date: 2002/01/15
473		Tag: <none>
474		Description:
475
476		1. Since COPY instructions are no longer native to the architecture,
477		a generic functor can be used to implement the expandCopies function.
478
479		2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
480		TEXT segment.
481
482		----------------------------------------------------------------------
483		Name: Matthias Blume
484		Date: 2002/01/15
485		Tag: blume-20020115-ffiupdates
486		Description:
487
488		1. Fix for bug resulting in single-precision float values being returned
489		incorrectly from FFI calls.
490
491		2. Small modifications to C FFI API:
492
493		- memory-allocation routines return straight objects (no options)
494		and raise an exception in out-of-memory situations
495		- unsafe extensions to cast between function pointers and pointers
496		from/to ints
497		- added structure C_Debug as an alternative to structure C where
498		pointer-dereferencing (|*| and |*!) always check for null-pointers
499		- added open_lib' to DynLinkage;  open_lib' works like open_lib
500		but also takes a (possibly empty) list of existing library handles
501		that the current library depends on
502
503		----------------------------------------------------------------------
504		Name: Matthias Blume
505		Date: 2002/01/10
506		Tag: blume-20020110-newffigen
507		Description:
508
509		1. Updates to portable graph code.
510
511		2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
512		more scalable now so that even huge interfaces such as the one
513		for GTK compile in finite time and space. :-)
514		See src/ml-nlffigen/README for details on what's new.
515
516		----------------------------------------------------------------------
517		Name: Lal George
518		Date: 2001/01/09 14:31:35 EST 2002
519		Tag: george-20011206-rm-native-copy
520		Description:
521
522		Removed the native COPY and FCOPY instructions
523		from all the architectures and replaced it with the
524		explicit COPY instruction from the previous commit.
525
526		It is now possible to simplify many of the optimizations
527		modules that manipulate copies. This has not been
528		done in this change.
529
530		----------------------------------------------------------------------
531		Name: Lal George
532		Date: 2001/12/06 16:50:13 EST 2001
533		Tag: george-20011206-mlrisc-instruction
534		Description:
535
536		Changed the representation of instructions from being fully abstract
537		to being partially concrete. That is to say:
538
539		from
540		type instruction
541
542		to
543		type instr                              (* machine instruction *)
544
545		datatype instruction =
546		LIVE of {regs: C.cellset, spilled: C.cellset}
547		| KILL of {regs: C.cellset, spilled: C.cellset}
548		| COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
549		| ANNOTATION of {i: instruction, a: Annotations.annotation}
550		| INSTR of instr
551
552		This makes the handling of certain special instructions that appear on
553		all architectures easier and uniform.
554
555		LIVE and KILL say that a list of registers are live or killed at the
556		program point where they appear. No spill code is generated when an
557		element of the 'regs' field is spilled, but the register is moved to
558		the 'spilled' (which is present, more for debugging than anything else).
559
560		LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
561		We used to generate:
562
563		DEFFREG f1
564		f1 := f2 + f3
565		trapb
566
567		but now generate:
568
569		f1 := f2 + f3
570		trapb
571		LIVE {regs=[f1,f2,f3], spilled=[]}
572
573		Furthermore, the DEFFREG (hack) required that all floating point instruction
574		use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
575		defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
576		in a cleaner alpha implementation. (Hopefully, intel will not get rid of
577		this architecture).
578
579		COPYXXX is intended to replace the parallel COPY and FCOPY  available on
580		all the architectures. This will result in further simplification of the
581		register allocator that must be aware of them for coalescing purposes, and
582		will also simplify certain aspects of the machine description that provides
583		callbacks related to parallel copies.
584
585		ANNOTATION should be obvious, and now INSTR represents the honest to God
586		machine instruction set!
587
588		The <arch>/instructions/<arch>Instr.sml files define certain utility
589		functions for making porting easier -- essentially converting upper case
590		to lower case. All machine instructions (of type instr) are in upper case,
591		and the lower case form generates an MLRISC instruction. For example on
592		the alpha we have:
593
594		datatype instr =
595		LDA of {r:cell, b:cell, d:operand}
596		| ...
597
598		val lda : {r:cell, b:cell, d:operand} -> instruction
599		...
600
601		where lda is just (INSTR o LDA), etc.
602
603		----------------------------------------------------------------------
604		Name: Matthias Blume
605		Date: 2001/11/22 21:40:00 EST
606		Tag: Release_110_37
607		Description:
608
609		Release 110.37.  This time for real.
610
611		----------------------------------------------------------------------
612		Name: Matthias Blume
613		Date: 2001/11/21 16:35:00 EST
614		Tag: blume-20011121-foot-in-mouth
615		Description:
616
617		Removed the "Release_110_37" tag because of a serious bug.
618		This will be re-tagged once the bug is fixed.
619
620		----------------------------------------------------------------------
621		Name: Matthias Blume
622		Date: 2001/11/21 16:14:00 EST
623		Tag: blume-20011121-forgottenfile
624		Description:
625
626		Forgot to add a file.  (Just a .tex-file -- part of
627		the CM manual source.)
628
629		----------------------------------------------------------------------
630		Name: Matthias Blume
631		Date: 2001/11/21 16:10:00 EST
632		Tag: blume-20011121-invalid_110_37
633		Description:
634
635		Note: I removed the original tag "Release_110_37" from this commit
636		because we found a serious bug in all non-x86 backends.
637		- Matthias
638
639		1. Modifications to the SML/NJ code generator and to the runtime system
640		so that code object name strings are directly inserted into code
641		objects at code generation time.  The only business the runtime system
642		has with this is now to read the name strings on occasions.
643		(The encoding of the name string has also changed somewhat.)
644
645		2. CM now implements a simple "set calculus" for specifying export lists.
646		In particular, it is now possible to refer to the export lists of
647		other libraries/groups/sources and form unions as well as differences.
648		See the latest CM manual for details.
649
650		3. An separate notion of "proxy" libraries has again be eliminated from
651		CM's model.  (Proxy libraries are now simply a special case of using
652		the export list calculus.)
653
654		4. Some of the existing libraries now take advantage of the new set
655		calculus.
656		(Notice that not all libraries have been converted because some
657		of the existing .cm-files are supposed to be backward compatible
658		with 110.0.x.)
659
660		5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
661		"General.exnMessage"!  The former relies on a certain hook to be
662		initialized, and that often does not happen in the stand-alone case.)
663
664		----------------------------------------------------------------------
665		Name: Lal George
666		Date: 2001/11/21  13:56:18 EST
667		Tag: george-2001121-pseudo-ops
668		Description:
669
670		Implemented a complete redesign of MLRISC pseudo-ops. Now there
671		ought to never be any question of incompatabilities with
672		pseudo-op syntax expected by host assemblers.
673
674		For now, only modules supporting GAS syntax are implemented
675		but more should follow, such as MASM, and vendor assembler
676		syntax, e.g. IBM as, Sun as, etc.
677
678		----------------------------------------------------------------------
679		Name: Matthias Blume
680		Date: 2001/11/14 11:52:00 EST
681		Tag: blume-20011114-srcname
682		Description:
683
684		1. Routed the name of the current source file to mlriscgen where it
685		should be directly emitted into the code object.  (This last part
686		is yet to be done.)
687
688		2. Some cleanup of the pgraph code to make it match the proposal that
689		I put out the other day.  (The proposal notwithstanding, things are
690		still in flux here.)
691
692		----------------------------------------------------------------------
693		Name: Lal George
694		Date: 2001/11/14 09:44:04 EST
695		Tag:
696		Description:
697
698		Fix for a backpatching bug reported by Allen.
699
700		Because the boundary between short and long span-dependent
701		instructions is +/- 128, there are an astounding number of
702		span-dependent instructions whose size is over estimated.
703
704		Allen came up with the idea of letting the size of span
705		dependent instructions be non-monotonic, for a maxIter
706		number of times, after which the size must be monotonically
707		increasing.
708
709		This table shows the number of span-dependent instructions
710		whose size was over-estimated as a function of maxIter, for the
711		file Parse/parse/ml.grm.sml:
712
713		maxIter            # of instructions:
714		10                      687
715		20                      438
716		30                      198
717		40                        0
718
719		In compiling the compiler, there is no significant difference in
720		compilation speed between maxIter=10 and maxIter=40. Actually,
721		my measurements showed that maxIter=40 was a tad faster than
722		maxIter=10! Also 96% of the  files in the compiler reach a fix
723		point within 13 iterations, so fixing maxIter at 40, while high,
724		is okay.
725
726		----------------------------------------------------------------------
727		Name: Matthias Blume
728		Date: 2001/10/31 15:25:00 EST
729		Tag: blume-20011031-pgraph
730		Description:
731
732		CKIT:
733		* Changed the "Function" constructor of type Ast.ctype to carry optional
734		argument identifiers.
735		* Changed the return type of TypeUtil.getFunction accordingly.
736		* Type equality ignores the argument names.
737		* TypeUtil.composite tries to preserve argument names but gives up quickly
738		if there is a mismatch.
739
740		installation script:
741		* attempts to use "curl" if available (unless "wget" is available as well)
742
743		CM:
744		* has an experimental implementation of "portable graphs" which I will
745		soon propose as an implementation-independent library format
746		* there are also new libraries $/pgraph.cm and $/pgraph-util.cm
747
748		NLFFI-LIB:
749		* some cleanup (all cosmetic)
750
751		NLFFIGEN:
752		* temporarily disabled the mechanism that suppresses ML output for
753		C definitions whose identifiers start with an underscore character
754		* generate val bindings for enum constants
755		* user can request that only one style (light or heavy) is being used;
756		default is to use both (command-line arguments: -heavy and -light)
757		* fixed bug in handling of function types involving incomplete pointers
758		* generate ML entry points that take record arguments (i.e., using
759		named arguments) for C functions that have a prototype with named
760		arguments
761		(see changes to CKIT)
762
763		----------------------------------------------------------------------
764		Name: Allen Leung
765		Date: 2001/10/27 20:34:00 EDT
766		Tag: leunga-20011027-x86-fast-fp-call
767		Description:
768
769		Fixed the bug described in blume-20010920-slowfp.
770
771		The fix involves
772		1. generating FCOPYs in FSTP in ia32-svid
773		2. marking a CALL with the appropriate annotation
774
775		----------------------------------------------------------------------
776		Name: Matthias Blume
777		Date: 2001/10/16 11:32:00 EDT
778		Tag: blume-20011016-netbsd
779		Description:
780
781		Underscore patch from Chris Richards (fixing problem with compiling
782		runtime system under recent NetBSD).
783
784		----------------------------------------------------------------------
785		Name: Allen Leung
786		Date: 2001/10/12 17:18:32 EDT 2001
787		Tag: leung-20011012-x86-printflowgraph
788		Description:
789
790		X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
791
792		----------------------------------------------------------------------
793		Name: Lal George
794		Date: 2001/10/11 23:51:34 EDT
795		Tag: george-20011011-too-many-instrs
796		Description:
797
798		The representation of a program point never expected to see more
799		than 65536 instructions in a basic block!
800
801		----------------------------------------------------------------------
802		Name: Lal George
803		Date: 2001/10/09 09:41:37 EDT
804		Tag: george-20011008-mlrisc-labels
805		Description:
806
807		Changed the machine description files to support printing of
808		local and global labels in assembly code, based on host assembler
809		conventions.
810
811		----------------------------------------------------------------------
812		Name: Matthias Blume
813		Date: 2001/09/25 15:25:00 EDT
814		Tag: blume-20010925-exninfo
815		Description:
816
817		I provided a non-hook implementation of exnName (at the toplevel) and
818		made the "dummy" implementation of exnMessage (at the toplevel) more
819		useful: if nothing gets "hooked in", then at least you are going to
820		see the exception name and a message indicating why you don't see more.
821
822		[For the time being, programs that need exnMessage and want to use
823		ml-build should either use General.exnMessage (strongly recommended) or
824		refer to structure General at some other point so that CM sees a
825		static dependency.]
826
827		[Similar remarks go for "print" and "use":  If you want to use their
828		functionality in stand-alone programs generated by ml-build, then use
829		TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
830
831		----------------------------------------------------------------------
832		Name: Matthias Blume
833		Date: 2001/09/20 17:28:00 EDT
834		Tag: blume-20010920-slowfp
835		Description:
836
837		Allen says that x86-fast-fp is not safe yet, so I turned it off again...
838
839		----------------------------------------------------------------------
840		Name: Matthias Blume
841		Date: 2001/09/20 17:20:00 EDT
842		Tag: blume-20010920-canonicalpaths
843		Description:
844
845		0. Updated the BOOT file (something that I forgot to do earlier).
846
847		1. Small internal change to CM so that it avoids "/../" in filenames
848		as much as possible (but only where it is safe).
849
850		2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
851		that arguments that contain delimiters are passed through correctly.
852		This change also means that all "special" arguments of the form
853		@SMLxxx... must come first.
854
855		3. Changed install script to put relative anchor names for tool commands
856		into pathconfig.
857
858		----------------------------------------------------------------------
859		Name: Matthias Blume
860		Date: 2001/09/18 15:35:00 EDT
861		Tag: blume-20010918-readme11036
862		Description:
863
864		Added README files.
865
866		----------------------------------------------------------------------
867		Name: Matthias Blume
868		Date: 2001/09/18 11:45:00 EDT
869		Tag: Release_110_36 (retag)
870		Description:
871
872		Fixed mistake in config/preloads. Retagged as 110.36.
873
874		----------------------------------------------------------------------
875		Name: Matthias Blume
876		Date: 2001/09/18 09:40:00 EDT
877		Tag: Release_110_36_orig (tag changed)
878		Description:
879
880		New version (110.36).  New bootfiles.
881
882		----------------------------------------------------------------------
883		Name: Matthias Blume
884		Date: 2001/09/14 16:15:00 EDT
885		Tag: blume-20010914-x86fastfp
886		Description:
887
888		John committed some changes that Allen made, in particular a (hopefully)
889		correctly working version of the x86-fp module.
890
891		I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
892		flag to "true".  Everything seems to compile to a fixpoint ok, and
893		"mandelbrot" speeds up by about 15%.
894
895		----------------------------------------------------------------------
896		Name: Matthias Blume
897		Date: 2001/09/13 11:20:00 EDT
898		Tag: blume-20010913-minimal
899		Description:
900
901		1. Stefan Monnier's patch to fix a miscompilation problem that
902		was brought to light by John Reppy's work on Moby.
903
904		2. Implemented a minimal "structure Compiler" that contains just
905		"version" and "architecture".  The minimal version will be
906		available when the full version is not.  This is for backward-
907		compatibility with code that wants to test Compiler.version.
908
909		----------------------------------------------------------------------
910		Name: Matthias Blume
911		Date: 2001/08/28 14:03:00 EDT
912		Tag: blume-20010828-ml-lex
913		Description:
914
915		Fix for bug 1581, received from Neophytos Michael.
916
917		----------------------------------------------------------------------
918		Name: Matthias Blume
919		Date: 2001/08/27 11:20:00 EDT
920		Tag: blume-20010827-readme11035
921		Description:
922
923		Fleshed out the README file for 110.35.
924
925		----------------------------------------------------------------------
926		Name: Matthias Blume
927		Date: 2001/08/24 17:10:00 EDT
928		Tag: Release_110_35
929		Description:
930
931		New version number (110.35).  New bootfiles.
932
933		----------------------------------------------------------------------
934		Name: Lal George
935		Date: 2001/08/24 13:47:18 EDT 2001
936		Tag: george-20010824-MLRISC-graphs
937		Description:
938
939		removed clusters from MLRISC completely and replaced with graphs.
940
941		----------------------------------------------------------------------
942		Name: Matthias Blume
943		Date: 2001/08/23 17:50:00 EDT
944		Tag: blume-20010823-toplevel
945		Description:
946
947		- some reorganization of the code that implements various kinds of
948		environments in the compiler (static, dynamic, symbolic, combined)
949		- re-implemented the EnvRef module so that evalStream works properly
950		(if the stream contains references to "use", "CM.make", etc.)
951		- cleaned up evalloop.sml and interact.sml (but they need more cleaning)
952
953		----------------------------------------------------------------------
954		Name: Matthias Blume
955		Date: 2001/08/20 15:50 EDT
956		Tag: blume20010820-slipup
957		Description:
958
959		I forgot to commit a few files.  Here they are...
960
961		----------------------------------------------------------------------
962		Name: Matthias Blume
963		Date: 2001/08/20 15:35:00 EDT
964		Tag: blume-20010820-debugprof
965		Description:
966
967		!!!! NEW BOOTFILES !!!!
968
969		This is another round of reorganizing the compiler sources.  This
970		time the main goal was to factor out all the "instrumentation"
971		passes (for profiling and backtracing) into their own library.
972		The difficulty was to do it in such a way that it does not depend
973		on elaborate.cm but only on elabdata.cm.
974
975		Therefore there have been further changes to both elaborate.cm and
976		elabdata.cm -- more "generic" things have been moved from the former
977		to the latter.  As a result, I was forced to split the assignment
978		of numbers indicating "primtyc"s into two portions: SML-generic and
979		SML/NJ-specific.  Since it would have been awkward to maintain,
980		I bit the bullet and actually _changed_ the mapping between these
981		numbers and primtycs.  The bottom line of this is that you need
982		a new set of bin- and bootfiles.
983
984		I have built new bootfiles for all architectures, so doing a fresh
985		checkout and config/install.sh should be all you need.
986
987		The newly created library's name is
988
989		$smlnj/viscomp/debugprof.cm
990
991		and its sources live under
992
993		src/compiler/DebugProf
994
995		----------------------------------------------------------------------
996		Name: Matthias Blume
997	Date: 2001/08/15 17:15:00 EDT	Date: 2001/08/15 17:15:00 EDT
998	Tag: blume-20010815-compreorg	Tag: blume-20010815-compreorg
999	Description:	Description:
#	Line 2853	Line 3833
3833	elaborator). There were a lot of changes during my "linkpath" trials	elaborator). There were a lot of changes during my "linkpath" trials
3834	that could have been reverted to their original state but weren't.	that could have been reverted to their original state but weren't.
3835	Please, don't be too harsh on me for messing with this code a bit more	Please, don't be too harsh on me for messing with this code a bit more
3836	than what was strictly necessary...  (I _did_ resist the tempation	than what was strictly necessary...  (I _did_ resist the temptation
3837	of doing any "global reformatting" to avoid an untimely death at	of doing any "global reformatting" to avoid an untimely death at
3838	Dave's hands. :)	Dave's hands. :)
3839

---

Colored Diff Long Colored Diff Unidiff Context Diff Side by Side

Legend: Removed from v.902   changed lines   Added in v.1069

---

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