Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 1130, Mon Mar 11 04:49:41 2002 UTC	revision 1387, Mon Sep 22 17:10:40 2003 UTC
#	Line 13	Line 13
13	Description:	Description:
14
15	----------------------------------------------------------------------	----------------------------------------------------------------------
16		Name: Matthias Blume (blume (at) tti - c (dot) org)
17		Date: 2003/09/22 12:10:00 CDT
18		Tag: blume-20030918-110_43_2
19		Description:
20
21		Made a new interim version and bootfiles for developer's bootstrapping
22		convenience.
23
24		110.43.2 -- NEW BOOTFILES
25
26		----------------------------------------------------------------------
27		Name: Matthias Blume (blume (at) tti - c (dot) org)
28		Date: 2003/09/19 15:55:00 CDT
29		Tag: blume-20030919-cmdir
30		Description:
31
32		1. new-install.sh -> install.sh
33		2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
34		3. tweaked installer so that another name instead of .cm can be chosen
35		at install time (by setting the CM_DIR_ARC environment variable
36		during installation); once installation is complete, the name is
37		fixed
38
39		----------------------------------------------------------------------
40		Name: Matthias Blume (blume (at) tti - c (dot) org)
41		Date: 2003/09/18 16:00:00 CDT
42		Tag: blume-20030918-110_43_1
43		Description:
44
45		Made a new interim version and bootfiles for developer's bootstrapping
46		convenience.
47
48		110.43.1 -- NEW BOOTFILES
49
50		----------------------------------------------------------------------
51		Name: Matthias Blume (blume (at) tti - c (dot) org)
52		Date: 2003/09/18 15:20:00 CDT
53		Tag: blume-20030918-misc
54		Description:
55
56		1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
57		this seems to be controversial at the moment) and used those in
58		Posix.ProcEnv.times.
59
60		2. Added Time.{from,to}Nanoseconds to Time.
61
62		3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
63		For example, fromLargeInt never needs to look at more than 3 "big
64		digits" to get its 53 bits of precision.
65
66		----------------------------------------------------------------------
67		Name: Matthias Blume (blume (at) tti - c (dot) org)
68		Date: 2003/09/17 16:30:00 CDT
69		Tag: blume-20030917-real32-slices
70		Description:
71
72		Added an entry to the primitive environment
73		(compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
74		added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
75
76		Removed some of the "magic" constants in real64.sml and replaced them
77		with code that generates these values from their corresponding
78		integer counterparts.
79
80		Made all(?) the slice-related changes to the Basis and made everything
81		compile again...
82
83		----------------------------------------------------------------------
84		Name: Matthias Blume (blume (at) tti - c (dot) org)
85		Date: 2003/09/15 17:45:00 CDT
86		Tag: blume-20030915-rbase
87		Description:
88
89		Fixed bug in Real.fromLargeInt.
90
91		----------------------------------------------------------------------
92		Name: Matthias Blume (blume (at) tti - c (dot) org)
93		Date: 2003/09/13 18:11:00 CDT
94		Tag: blume-20030913-libinstall
95		Description:
96
97		Minor bugfix in config/libinstall (set anchor with path to
98		standalone tool after installing it, otherwise libraries that
99		need ml-lex or ml-yacc won't compile the first time the installer
100		runs).
101
102		----------------------------------------------------------------------
103		Name: Matthias Blume (blume (at) tti - c (dot) org)
104		Date: 2003/09/12 11:45:00 CDT
105		Tag: blume-20030912-various
106		Description:
107
108		- fixed bug in Real.toLargeInt
109		- fixed bug in Posix.ProcEnv.times
110		- changed inputLine functions to return an option
111		- minor installer improvements / bugfixes
112		- changed default @SMLalloc parameter for x86/celeron to 64k
113
114		----------------------------------------------------------------------
115		Name: Matthias Blume (blume (at) tti - c (dot) org)
116		Date: 2003/09/09 22:00:00 CDT
117		Tag: Release_110_43
118		Description:
119
120		New working release 110.43.  New bootfiles.
121
122		----------------------------------------------------------------------
123		Name: Matthias Blume (blume (at) tti - c (dot) org)
124		Date: 2003/09/09 19:20:00 CDT
125		Tag: blume-20030909-installer
126		Description:
127
128		Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
129		Modified config/install.bat to take advantage of it.  Also modified
130		config/install.sh (and called it config/new-install.sh) to take advantage
131		of it on Unix systems.  (The SML code is (supposed to be) platform-
132		independent.)
133
134		The installer can now install everything under Win32
135		as well as under *nix as long as it compiles.
136
137		Other changes:
138
139		- made CML compile again under Win32
140		- made eXene compile under Win32 (by providing a fake structure UnixSock
141		and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
142		- fixed a bug in nowhere: it assumed that type OS.Process.status is the
143		same as type int;  under Win32 it isn't
144		- fixed some slice-related problems in the win32-specific parts of CML
145		- added a functor argument "sameVol" to os-path-fn.sml in the Basis
146		(under Win32, the volume name is case-insensitive, and the
147		OS.Path code compares volume names for equality)
148
149		----------------------------------------------------------------------
150		Name: Matthias Blume (blume (at) tti - c (dot) org)
151		Date: 2003/09/08 11:55:00 CDT
152		Tag: blume-20030908-fullpath
153		Description:
154
155		Made Win32 version of OS.FileSys.fullPath return current directory
156		when given an empty string.  This is what the spec says, and incidentally,
157		CM depends on it.  (CM otherwise goes into an infinite loop in certain
158		cases when presented with the name of a non-existing .cm file.)
159
160		----------------------------------------------------------------------
161		Name: Matthias Blume (blume (at) tti - c (dot) org)
162		Date: 2003/09/04 16:30:00 CDT
163		Tag: blume-20030905-slices-etc
164		Description:
165
166		1. Changed interface to vectors and arrays in Basis to match
167		(draft) Basis spec.
168		2. Added signatures and implementations of slices according to
169		Basis spec.
170		3. Edited source code throughout the system to make it compile again
171		under 1. and 2.  (In some cases code had to be added to have it
172		match the new signatures.)
173		4. MLRISC should be backward-compatible: the copies of the originals
174		of files that needed to change under 3. were retained, the .cm files
175		check the compiler version number and use old versions when
176		appropriate.
177		5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
178		dirstream -> string option (in accordance with Basis spec).
179		6. When generating code that counts lines, ml-lex used function
180		CharVector.foldli, taking advantage of its old interface.
181		This has been replaced with the corresponding code from
182		CharVectorSlice.  (html-lex must be re-lexed!)
183		7. BitArray in smlnj-lib/Util has been extended/modified to match the
184		new MONO_ARRAY signature.  (Do we need BitArraySlice?)
185		8. Removed temporary additions (fromInternal, toInternal) from the
186		(now obsolete) IntInf in smlnj-lib/Util.
187		9. Cleaned up structure Byte.
188		10. Added localOffset, scan, and fromString to Date (according to spec).
189		Cleaned/corrected implementation of Date.
190		(Still need to check for correctness; implement better canonicalizeDate.)
191		11. Added "scan" to signature IEEE_REAL.
192		12. Some improvements to IntInf [in particular: efficiency-hack for
193		mod and rem when second operand is 2 (for parity checks).]
194		13. Changed representation of type Time.time, using a single IntInf.int
195		value counting microseconds.  This considerably simplified the
196		implementation of structure Time.  We now support negative time
197		values; scan and fromString handle signs.
198		14. Functor PrimIO now takes two additional arguments (VectorSlice and
199		ArraySlice).
200
201		----------------------------------------------------------------------
202		Name: Matthias Blume (blume (at) tti - c (dot) org)
203		Date: 2003/08/28 17:00:00 CDT
204		Tag: blume-20030828-intinf
205		Description:
206
207		This is a major update which comes with a version number bump
208		(110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
209		and an implementation of IntInf in the Basis.
210
211		There are a fairly large number of related changes and updates throughout
212		the system:
213
214		Basis:
215		- Implemented IntInf.
216		- Made LargeInt a projection of IntInf (by filtering through INTEGER).
217		- Added some missing Real64 operations, most notably Real.toLargeInt.
218		- Added FixedInt as a synonym for Int32.
219
220		compiler:
221		* Added support for a built-in intinf type.
222		- literals
223		- pattern matching
224		- conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
225		- overloading on literals and operations
226
227		This required adding a primitive type intinf, some additional
228		primops, and implementations for several non-trivial intinf
229		operations in Core.  (The intinf type is completely abstract
230		to the compiler; all operations get delegated back to the Core.)
231
232		* Intinf equality is handled by polyequal.  However, the compiler
233		does not print its usual warning in this case (since polyequal
234		is the right thing to do there).
235
236		* Improved the organization of structure InlineT.
237
238		* A word about conversion primops:
239		If conversions involving intinf do not cancel out during
240		CPS contract, then the compiler must insert calls to Core functions.
241		Since all core access must be resolved already during the FLINT
242		translate phase, it would be too late a the time of CPS contract
243		to add new Core calls.  For this reason, conversion primops
244		for intinf carry two arguments:  1. the numeric argument that
245		they are supposed to convert, and 2. the Core function that
246		can help with this conversion if necessary.  If CPS contract
247		eliminates a primop, then the associated Core function becomes
248		dead and goes away.  Intinf conversion primops that do not get
249		eliminated by CPS contract get rewritten into calls of their
250		core functions by a separate, new phase.
251
252		interactive system:
253		- Control.Print.intinfDepth controls max length of intinf constants
254		being printed. (Analogous to Control.Print.stringDepth.)
255		- Cleanup in printutil and pputil: got rid of unused stuff and
256		duplicates; replaced some of the code with code that makes better
257		use of library functionality.
258
259		CM:
260		Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
261		are no longer silent.
262
263		CKIT:
264		Fixed mismatched uses of Int32 and LargeInt.  I always decided
265		in favor of LargeInt -- which is now the same as IntInf.
266		CKIT-knowledgable people should check whether this is what's
267		intended and otherwise change things back to using Int32 or
268		FixedInt.
269
270		Throughout the code:
271		Started using IntInf.int literals and built-in operations
272		(e.g., comparison with 0) where this seems appropriate.
273
274
275		----------------------------------------------------------------------
276		Name: Dave MacQueen (dbm@cs.uchicago.edu)
277		Date: 2003/08/13 11:36:00 CDT
278		Tag: dbm-20030813-mcz-merge1
279		Description:
280
281		Merging changes from the mcz-branch development branch into trunk.
282		These changes involve replacement of the emulated old prettyprinter
283		interface with direct use of the SML/NJ Lib PP library, and fixing
284		of a couple of bugs (895, 1186) relating to error messages.  A new
285		prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
286		has been added.
287
288		----------------------------------------------------------------------
289		Name: Matthias Blume (blume (at) tti - c (dot) org)
290		Date: 2003/08/11 15:45:00 CDT
291		Tag: blume-20030811-windows
292		Description:
293
294		Version number bumped to 110.42.9.  NEW BOOTFILES!!!
295
296		http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
297
298		This patch restores SML/NJ's ability to run under win32.  There are a
299		number of changes, including fixes for several bugs that had gone
300		unnoticed until now:
301
302		- uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
303		trying to run the win32 version from within cygwin.)
304
305		- There are a number of simple .bat scripts that substitute for their
306		corresponding Unix shell-scripts.  (See below.)
307
308		- The internals of ml-build have been modified slightly.  The main
309		difference is that instead of calling ".link-sml" (or link-sml.bat)
310		using OS.Process.system, the ML process delegates this task back
311		to the script.  Otherwise problems arise in mixed environments such
312		as Cygwin where scripts look and work like Unix scripts, but
313		where OS.Process.system cannot run them.
314
315		- In CM, the srcpath pickler used native pathname syntax -- which
316		is incorrect in the case of cross-compilation.  The new pickle format
317		is independent of platform-specific naming conventions.
318
319		- Path configuration files (such as lib/pathconfig) can now choose
320		between native and standard syntax.  Placing a line of the form
321
322		standard!
323
324		into the file causes all subsequent paths to be interpreted using
325		CM standard pathname syntax (= Unix conventions); a line
326
327		native!
328
329		switches back to native style.  This was needed so that
330		path config files can be written portably, see src/system/pathconfig.
331
332		- Runtime system:
333
334		- win32-filesys.c:  get_file_time and set_file_time now
335		access modification time, not creation time.
336
337		- I/O code made aware of new array representation.
338
339		- Bug fixes in X86.prim.masm.
340
341		- src/system/makeml made aware of win32. (For use under cygwin
342		and other Unix-environments for windows.)
343
344		- In Basis, fixed off-by-one error in win32-io.sml (function vecF)
345		which caused BinIO.inputAll to fail consistently.
346
347		.bat scripts:
348
349		Windows .bat scripts assume that SMLNJ_HOME is defined.
350
351		- sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
352		applications (sml, ml-yacc, ml-lex).
353		- ml-build.bat: analogous to ml-build.
354		- config\install.bat:  Analogous to config/install.sh.  This requires
355		that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
356		(nmake etc. must be on the path, and vcvars32 must have been run.)
357		Moreover, sources for ml-lex and ml-yacc need to exist under src,
358		and the bootfile hierarchy must have been unpacked under
359		sml.boot.x86-win32.
360		The script is very primitive and does a poor job at error checking.
361		It only installs the base system, ml-lex, and ml-yacc.  No other
362		libraries are being installed (i.e., you get only those that
363		are part of the compiler.)
364		- link-sml.bat:  analogous to .link-sml, but not currently used
365
366		Unrelated bug fixes:
367
368		- ml-nlffigen now exports structures ST_* corresponding to incomplete
369		types.
370		- Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
371		otherwise.)
372
373		----------------------------------------------------------------------
374		Name: Dave MacQueen (macqueen@cs.uchicago.edu)
375		Date: 2003/06/17
376		Tag: macqueen-20030617-bug895
377		Description:
378
379		Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
380		Tag will be used for new development branch (mcz-branch) for
381		use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
382
383		----------------------------------------------------------------------
384		Name: Matthias Blume (blume (at) tti - c (dot) org)
385		Date: 2003/05/27 16:55:00 CDT
386		Tag: blume-20030527-polyeq
387		Description:
388
389		Tried to eliminated most cases of polymorphic equality.
390
391		----------------------------------------------------------------------
392		Name: Matthias Blume (blume (at) tti - c (dot) org)
393		Date: 2003/05/21 17:45:00 CDT
394		Tag: blume-20030517-complete
395		Description:
396
397		Two changes:
398
399		1. Added a flag for controlling whether non-exhaustive bindings will
400		be treated as errors (default is false).
401		2. Cleaned up the *entire* source tree so that CMB.make goes through
402		without a single non-exhaustive match- or bind warning.
403
404		----------------------------------------------------------------------
405		Name: Matthias Blume (blume (at) tti - c (dot) org)
406		Date: 2003/05/17 10:20:00 CDT
407		Tag: blume-20030517-absyn
408		Description:
409
410		1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
411
412		This mainly affects the quality of error messages.  However, some
413		of the code is now more straightforward than before. (Treatment of
414		the above four constructs in translate.sml is much simpler than
415		the "macro-expansion" that was going on before.  Plus, the mach-
416		compiler no longer gets invoked just to be able to compile an
417		if-expression.)
418
419		2. The ErrorMsg.Error exception is now caught and absorbed by the
420		interactive loop.
421
422		----------------------------------------------------------------------
423		Name: Allen Leung
424		Date: 2003/05/16 13:05:00 CDT
425		Tag: leunga-20030516-cygwin-runtime
426		Description:
427
428		Ported the runtime system to cygwin, which uses the unix
429		x86-unix bin files.  Missing/buggy features:
430
431		o getnetbyname, getnetbyaddr: these functions seem to be missing in
432		the Cygwin library.
433		o Ctrl-C handling may be flaky.
434		o Windows system calls and Windows I/O are not supported.
435
436		A new set of binfiles is located at:
437
438		http://www.dorsai.org/~leunga/boot.x86-unix.tgz
439
440		This is only needed for bootstrapping the cygwin version of smlnj.
441		Other x86 versions can use the existing binfiles.
442
443		----------------------------------------------------------------------
444		Name: Matthias Blume
445		Date: 2003/04/08 15:42:00 CDT
446		Tag: blume-20030408-listpair
447		Description:
448
449		1. Added a target 'mlrisc' to installer.
450
451		2. Added missing elements to structure ListPair.
452
453		----------------------------------------------------------------------
454		Name: Allen Leung
455		Date: 2003/01/07 10:40:00 EST
456		Tag: leunga-20030107-int-rem
457		Description:
458
459		Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
460		The arguments to the SUBL instruction were swapped.
461
462		----------------------------------------------------------------------
463		Name: Matthias Blume
464		Date: 2002/12/12 16:25:00 EST
465		Tag: blume-20021212-risc-ra
466		Description:
467
468		Fixed a serious bug in the rewrite code for FP spilling/reloading that
469		sent the RA into an infinite loop when floating point registers get
470		spilled.  (Because of this bug, e.g., nucleic stopped compiling between
471		110.37 and 110.38.)
472		There was another set of potential problems related to the handling of
473		MLRISC annotations (but those did not yet cause real problems, apparently).
474
475		----------------------------------------------------------------------
476		Name: Matthias Blume
477		Date: 2002/12/06 22:40:00 EST
478		Tag: blume-20021206-cm-fileid
479		Description:
480
481		Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
482		This fixes the problem of CM getting confused by files that suddenly
483		change their identity (e.g., by getting unlinked and recreated by some
484		text editor such as vi).  There might be a better/cheaper/cleaner way
485		of doing this, but for now this will have to do.
486
487		----------------------------------------------------------------------
488		Name: Matthias Blume
489		Date: 2002/10/28 09:50:00 EST
490		Tag: blume-20021028-typecheck
491		Description:
492
493		Exported structure Typecheck from $smlnj/viscomp/core.cm.
494
495		----------------------------------------------------------------------
496		Name: Matthias Blume
497		Date: 2002/10/17 09:10:00 EDT
498		Tag: Release_110_42
499		Description:
500
501		In good old tradition, there has been a slight hiccup so that we have
502		to patch 110.42 after the fact.   The old release tag has been replaced
503		(see below).
504
505		The change solves a problem with two competing approaches the
506		configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
507		each other's way.
508
509		This change only affects the runtime system code and the installer script.
510		(No new bootfiles.)
511
512		----------------------------------------------------------------------
513		Name: Matthias Blume
514		Date: 2002/10/16 12:00:00 EDT
515		Tag: Release_110_42_removed
516		Description:
517
518		New working release.  New bootfiles.
519
520		----------------------------------------------------------------------
521		Name: Matthias Blume
522		Date: 2002/10/10 13:10:00 EDT
523		Tag: blume-20021010-ppc-divs
524		Description:
525
526		The mltree operator DIVS must be implemented with an overflow check on
527		the PPC because the hardware indicates divide-by-zero using "overflow" as
528		well.
529
530		----------------------------------------------------------------------
531		Name: Matthias Blume
532		Date: 2002/07/23 11:45:00
533		Tag: blume-20020723-smlnj-home
534		Description:
535
536		Sml now senses the SMLNJ_HOME environment variable.  If this is set,
537		then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
538		CM_PATHCONFIG is also set), the path configuration file is assumed
539		to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
540		the entire tree to some other place and everything will "just work".
541
542		(Companion commands such as ml-build and ml-makedepend also sense this
543		variable.)
544
545		----------------------------------------------------------------------
546		Name: Matthias Blume
547		Date: 2002/07/12 21:19:00 EDT
548		Tag: blume-20020712-liveness
549		Description:
550
551		Exported two useful "step" functions from liveness module (MLRISC).
552
553		----------------------------------------------------------------------
554		Name: Matthias Blume
555		Date: 2002/07/05 16:00 EDT
556		Tag: Release_110_41
557		Description:
558
559		New working release.  New bootfiles.
560
561		----------------------------------------------------------------------
562		Name: Matthias Blume
563		Date: 2002/07/05 10:25:00 EDT
564		Tag: blume-20020705-btimp
565		Description:
566
567		Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
568		other clients can set up backtracing support.
569
570		----------------------------------------------------------------------
571		Name: Matthias Blume
572		Date: 2002/06/25 17:23:00 EDT
573		Tag: blume-20020625-fpmax
574		Description:
575
576		Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
577		case.  (The sense of the isNaN test was reversed -- which made min and
578		max always return their first argument.)
579
580		----------------------------------------------------------------------
581		Name: Matthias Blume
582		Date: 2002/06/11
583		Tag: blume-20020611-unixpath
584		Description:
585
586		Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
587
588		----------------------------------------------------------------------
589		Name: Matthias Blume
590		Date: 2002/06/10 16:35:00 EDT
591		Tag: blume-20020610-ieeereal
592		Description:
593
594		I back-ported my implementation of IEEEReal.fromString from the
595		idlbasis-devel branch so that we can test it.
596
597		Another small change is that ppDec tries to give more information
598		than just "<sig>" in the case of functors.  However, this code is
599		broken in some mysterious way if the functor's body's signature
600		has not been declared by ascription but gets inferred from the
601		implementation.  This needs fixing...
602
603		----------------------------------------------------------------------
604		Name: Matthias Blume
605		Date: 2002/05/31
606		Tag: blume-20020531-btrace-mode
607		Description:
608
609		Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
610		the wayside when I switched over to using Controls everywhere.)
611
612		----------------------------------------------------------------------
613		Name: Lal George
614		Date: 2002/05/23 12:21:40 EDT
615		Tag: george-20020523-visual-labels
616		Description:
617
618		Labels are now displayed in the graphical output to make
619		the fall-through and target blocks obvious.
620
621		----------------------------------------------------------------------
622		Name: Matthias Blume
623		Date: 2002/05/22 11:03:00 EDT
624		Tag: blume-20020522-shrink
625		Description:
626
627		John tweaked yesterday's fix for 1131 to handle an out-of-memory
628		situation that comes up when allocating huge arrays.
629
630		----------------------------------------------------------------------
631		Name: Matthias Blume
632		Date: 2002/05/21 16:00:00 EDT
633		Tag: Release_110_40
634		Description:
635
636		New working release (110.40).  New bootfiles.
637
638		[Also: John Reppy fixed GC bug 1131.]
639
640		----------------------------------------------------------------------
641		Name: Matthias Blume
642		Date: 2002/05/21 12:35:00 EDT
643		Tag: blume-20020521-cmdoc
644		Description:
645
646		CM documentation update.
647
648		----------------------------------------------------------------------
649		Name: Matthias Blume
650		Date: 2002/05/21 10:55:00 EDT
651		Tag: blume-20020521-misc
652		Description:
653
654		- John tweaked runtime to be silent on heap export (except when
655		GC messages are on).
656		- I added a few more things (cross-compiling versions of CMB) to
657		config/preloads (as suggestions).
658
659		----------------------------------------------------------------------
660		Name: Matthias Blume
661		Date: 2002/05/20 22:25:00 EDT
662		Tag: blume-20020520-controls
663		Description:
664
665		- Added ControlUtil structure to control-lib.cm.
666		- Use it throughout.
667		- Used Controls facility to define MLRISC controls (as opposed to
668		registering MLRISC control ref cells with Controls after the fact)
669		- Fixed messed-up controls priorities.
670
671		* Removed again all the stuff from config/preloads that one wouldn't
672		be able to preload at the time the initial heap image is built.
673		(Many libraries, e.g., CML, do not exist yet at this time.  The
674		only libraries that can be preloaded via config/preloads are those
675		that come bundled with the bootfiles.)
676
677		----------------------------------------------------------------------
678		Name: Matthias Blume
679		Date: 2002/05/20 10:59:00 EDT
680		Tag: blume-20020520-preloads
681		Description:
682
683		Added a lot of commented-out suggestions for things to be included
684		in config/preloads.
685
686		----------------------------------------------------------------------
687		Name: Allen Leung
688		Date: 2002/05/18 14:20:00 EDT
689		Tag: leunga-20020518-mdl
690		Description:
691
692		o Made the mdl tool stuff compile and run again.
693		o I've disabled all the stuff that depends on RTL specifications; they
694		are all badly broken anyway.
695
696		----------------------------------------------------------------------
697		Name: Matthias Blume
698		Date: 2002/05/17 16:49:00 EDT
699		Tag: blume-20020517-controls
700		Description:
701
702		0. John Reppy made several modifications to the SML/NJ library.
703		In particular, there is a shiny new controls-lib.cm.
704
705		1. Pushed new controls interface through compiler so that everything
706		compiles again.
707
708		2. Added FormatComb and FORMAT_COMB to the CML version of the
709		SML/NJ library (so that CML compiles again).
710
711		3. Modified init scripts because XXX_DEFAULT environment variables
712		are no longer with us.  (Boot-time initialization is now done
713		using the same environment variables that are also used for
714		startup-time initialization of controls.)
715
716		----------------------------------------------------------------------
717		Name: Lal George
718		Date: 2002/05/15 09:20:10 EDT
719		Tag: george-20020515-pseudo-op-decls
720		Description:
721
722		All pseudo-ops emitted before the first segment declaration
723		such as TEXT, DATA, and BSS directives are assumed to be global
724		declarations and are emitted first in the assembly file. This is
725		useful in a number of situations where one has pseudo-ops that are not
726		specific to any segment, and also works around the constraint that one
727		cannot have client pseudo-ops in the TEXT segment.
728
729		Because no segment is associated with these declarations it is
730		an error to allocate any space or objects before the first segment
731		directive and an exception will be raised. However, we cannot make
732		this check for client pseudo-ops.
733
734		These top level declarations are a field in the CFG graph_info.
735		In theory you can continue to add to this field after the CFG has been
736		built -- provided you know what you are doing;-)
737
738		----------------------------------------------------------------------
739		Name: Matthias Blume
740		Date: 2002/05/13 16:40:00 EDT
741		Tag: blume-20020513-pp-etc
742		Description:
743
744		A few minor bugfixes:
745
746		- Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
747		(Bogus printouts for redefined bindings still occur.  Compiler
748		bug should no longer occur now.  We need to redo the prettyprinter
749		from scratch.)
750
751		- CM pathname printer now also adds escape sequences for ( and )
752
753		- commend and docu fixes for ml-nlffi
754
755		----------------------------------------------------------------------
756		Name: Matthias Blume
757		Date: 2002/05/10 16:40:00 EDT
758		Tag: blume-20020510-erg-textio
759		Description:
760
761		Applied the following bugfix provided by Emden Gansner:
762
763		Output is corrupted when outputSubstr is used rather than output.
764
765		The problem occurs when a substring
766
767		ss = (s, dataStart, dataLen)
768
769		where dataStart > 0, fills a stream buffer with avail bytes left.
770		avail bytes of s, starting at index dataStart, are copied into the
771		buffer, the buffer is flushed, and then the remaining dataLen-avail
772		bytes of ss are copied into the beginning of the buffer. Instead of
773		starting this copy at index dataStart+avail in s, the current code
774		starts the copy at index avail.
775
776		Fix:
777		In text-io-fn.sml, change line 695 from
778		val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
779		to
780		val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
781
782		----------------------------------------------------------------------
783		Name: Matthias Blume
784		Date: 2002/04/12 13:55:00 EDT
785		Tag: blume-20020412-assyntax
786		Description:
787
788		1. Grabbed newer assyntax.h from the XFree86 project.
789		2. Fiddled with how to compile X86.prim.asm without warnings.
790		3. (Very) Minor cleanup in CM.
791
792		----------------------------------------------------------------------
793		Name: Matthias Blume
794		Date: 2002/04/01 (no joke!) 17:07:00 EST
795		Tag: blume-20020401-x86div
796		Description:
797
798		Added full support for div/mod/rem/quot on the x86, using the machine
799		instruction's two results (without clumsily recomputing the remainder)
800		directly where appropriate.
801
802		Some more extensive power-of-two support was added to the x86 instruction
803		selector (avoiding expensive divs, mods, and muls where they can be
804		replaced with cheaper shifts and masks).  However, this sort of thing
805		ought to be done earlier, e.g., within the CPS optimizer so that
806		all architectures benefit from it.
807
808		The compiler compiles to a fixed point, but changes might be somewhat
809		fragile nevertheless.  Please, report any strange things that you might
810		see wrt. div/mod/quot/rem...
811
812		----------------------------------------------------------------------
813		Name: Matthias Blume
814		Date: 2002/03/29 17:22:00
815		Tag: blume-20020329-div
816		Description:
817
818		Fixed my broken div/mod logic.  Unfortunately, this means that the
819		inline code for div/mod now has one more comparison than before.
820		Fast paths (quotient > 0 or remainder = 0) are not affected, though.
821		The problem was with quotient = 0, because that alone does not tell
822		us which way the rounding went.  One then has to look at whether
823		remainder and divisor have the same sign...  :(
824
825		Anyway, I replaced the bootfiles with fresh ones...
826
827		----------------------------------------------------------------------
828		Name: Matthias Blume
829		Date: 2002/03/29 14:10:00 EST
830		Tag: blume-20020329-inlprims
831		Description:
832
833		NEW BOOTFILES!!!    Version number bumped to 110.39.3.
834
835		Primops have changed. This means that the bin/boot-file formats have
836		changed as well.
837
838		To make sure that there is no confusion, I made a new version.
839
840
841		CHANGES:
842
843		* removed REMT from mltree (remainder should never overflow).
844
845		* added primops to deal with divisions of all flavors to the frontend
846
847		* handled these primops all the way through so they map to their respective
848		MLRISC support
849
850		* used these primops in the implementation of Int, Int32, Word, Word32
851
852		* removed INLDIV, INLMOD, and INLREM as they are no longer necessary
853
854		* parameterized INLMIN, INLMAX, and INLABS by a numkind
855
856		* translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
857		floating point
858
859		* used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
860		and Real (but Real.abs maps to a separate floating-point-only primop)
861
862
863		TODO items:
864
865		* Hacked Alpha32 instruction selection, disabling the selection of REMx
866		instructions because the machine instruction encoder cannot handle
867		them.  (Hppa, PPC, and Sparc instruction selection did not handle
868		REM in the first place, and REM is supported by the x86 machine coder.)
869
870		* Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
871		selection phase.  (The two can be streamlined because the hardware
872		delivers both quotient and remainder at the same time anyway.)
873
874		* Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
875		(Currently the behavior is inconsistent both across architectures and
876		wrt. the draft Basis spec.)
877
878		* Word8 should eventually be handled natively, too.
879
880		* There seems to be one serious bug in mltree-gen.sml.  It appears, though,
881		as if there currently is no execution path that could trigger it in
882		SML/NJ.  (The assumptions underlying functions arith and promotable do not
883		hold for things like multiplication and division.)
884
885		----------------------------------------------------------------------
886		Name: Matthias Blume
887		Date: 2002/03/27 16:27:00 EST
888		Tag: blume-20020327-mlrisc-divisions
889		Description:
890
891		Added support for all four division operations (ML's div, mod, quot,
892		and rem) to MLRISC.  In the course of doing so, I also rationalized
893		the naming (no more annoying switch-around of DIV and QUOT), by
894		parameterizing the operation by div_rounding_mode (which can be either
895		DIV_TO_ZERO or DIV_TO_NEGINF).
896
897		The generic MLTreeGen functor takes care of compiling all four
898		operations down to only round-to-zero div.
899
900		Missing pieces:
901
902		* Doing something smarter than relying on MLTreeGen on architectures
903		like, e.g., the x86 where hardware division delivers both quotient and
904		remainder at the same time.  With this, the implementation of the
905		round-to-neginf operations could be further streamlined.
906
907		* Remove inlining support for div/mod/rem from the frontend and replace it
908		with primops that get carried through to the backend.  Do this for all
909		int and word types.
910
911		----------------------------------------------------------------------
912		Name: Matthias Blume
913		Date: 2002/03/25 17:25:00 EST
914		Tag: blume-20020325-divmod
915		Description:
916
917		I improved (hopefully without breaking them) the implementation of Int.div,
918		Int.mod, and Int.rem.   For this, the code in translate.sml now takes
919		advantage of the following observations:
920
921		Let  q = x quot y      r = x rem y
922		d = x div  y      m = x mod y
923
924		where "quot" is the round-to-zero version of integer division that
925		hardware usually provides.  Then we have:
926
927		r = x - q * y        where neither the * nor the - will overflow
928		d = if q >= 0 orelse x = q * y then q else q - 1
929		where neither the * nor the - will overflow
930		m = if q >= 0 orelse r = 0 then r else r + y
931		where the + will not overflow
932
933		This results in substantial simplification of the generated code.
934		The following table shows the number of CFG nodes and edges generated
935		for
936		fun f (x, y) = x OPER y
937		(* with OPER \in div, mod, quot, rem *)
938
939
940		OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
941		--------------------------------------------------------
942		div |         24 |         39 |         12 |         16
943		mod |         41 |         71 |         12 |         16
944		quot |          8 |         10 |          8 |         10
945		rem |         10 |         14 |          8 |         10
946
947
948		----------------------------------------------------------------------
949		Name: Matthias Blume
950		Date: 2002/03/25 22:06:00 EST
951		Tag: blume-20020325-cprotobug
952		Description:
953
954		Fixed a bug in cproto (c prototype decoder).
955
956		----------------------------------------------------------------------
957		Name: Matthias Blume
958		Date: 2002/03/25 16:00:00 EST
959		Tag: blume-20020325-raw-primops
960		Description:
961
962		I did some cleanup to Allen's new primop code and
963		replaced yesterday's bootfiles with new ones.
964		(But they are stored in the same place.)
965
966		----------------------------------------------------------------------
967		Name: Matthias Blume
968		Date: 2002/03/24 22:40:00 EST
969		Tag: blume-20020324-bootfiles
970		Description:
971
972		Made the bootfiles that Allen asked for.
973
974		----------------------------------------------------------------------
975		Name: Allen Leung
976		Date: 2002/03/23 15:50:00 EST
977		Tag: leunga-20020323-flint-cps-rcc-primops
978		Description:
979
980		1. Changes to FLINT primops:
981
982		(* make a call to a C-function;
983		* The primop carries C function prototype information and specifies
984		* which of its (ML-) arguments are floating point. C prototype
985		* information is for use by the backend, ML information is for
986		* use by the CPS converter. *)
987		| RAW_CCALL of { c_proto: CTypes.c_proto,
988		ml_args: ccall_type list,
989		ml_res_opt: ccall_type option,
990		reentrant : bool
991		} option
992		(* Allocate uninitialized storage on the heap.
993		* The record is meant to hold short-lived C objects, i.e., they
994		* are not ML pointers.  With the tag, the representation is
995		* the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
996		*)
997		| RAW_RECORD of {tag:bool,sz:int}
998		and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
999
1000		2.  These CPS primops are now overloaded:
1001
1002		rawload of {kind:numkind}
1003		rawstore of {kind:numkind}
1004
1005		The one argument form is:
1006
1007		rawload {kind} address
1008
1009		The two argument form is:
1010
1011		rawload {kind} [ml object, byte-offset]
1012
1013		3. RAW_CCALL/RCC now takes two extra arguments:
1014
1015		a. The first is whether the C call is reentrant, i.e., whether
1016		ML state should be saved and restored.
1017		b. The second argument is a string argument specifying the name of
1018		library and the C function.
1019
1020		These things are currently not handled in the code generator, yet.
1021
1022		4. In CProto,
1023
1024		An encoding type of "bool" means "ml object" and is mapped into
1025		C prototype of PTR.  Note that "bool" is different than "string",
1026		even though "string" is also mapped into PTR, because "bool"
1027		is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1028
1029		5. Pickler/unpicker
1030
1031		Changed to handle RAW_RECORD and newest RAW_CCALL
1032
1033		6. MLRiscGen,
1034
1035		1. Changed to handle the new rawload/rawstore/rawrecord operators.
1036		2. Code for handling C Calls has been moved to a new module CPSCCalls,
1037		in the file CodeGen/cpscompile/cps-c-calls.sml
1038
1039		7. Added the conditional move operator
1040
1041		condmove of branch
1042
1043		to cps.  Generation of this is still buggy so it is currently
1044		disabled.
1045
1046		----------------------------------------------------------------------
1047		Name: Lal George
1048		Date: 2002/03/22 14:18:25 EST
1049		Tag: george-20020322-cps-branch-prob
1050		Description:
1051
1052		Implemented the Ball-Larus branch prediction-heuristics, and
1053		incorporated graphical viewers for control flow graphs.
1054
1055		Ball-Larus Heuristics:
1056		---------------------
1057		See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1058
1059		By design it uses the Dempster-Shafer theory for combining
1060		probabilities.  For example, in the function:
1061
1062		fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1063
1064		the ball-larus heuristics predicts that the n=0 is unlikely
1065		(OH-heuristic), and the 'then' branch is unlikely because of the
1066		RH-heuristic -- giving the 'then' branch an even lower combined
1067		probability using the Dempster-Shafer theory.
1068
1069		Finally, John Reppy's loop analysis in MLRISC, further lowers the
1070		probability of the 'then' branch because of the loop in the else
1071		branch.
1072
1073
1074		Graphical Viewing:
1075		------------------
1076		I merely plugged in Allen's graphical viewers into the compiler. The
1077		additional code is not much. At the top level, saying:
1078
1079		Control.MLRISC.getFlag "cfg-graphical-view" := true;
1080
1081		will display the graphical view of the control flow graph just before
1082		back-patching.  daVinci must be in your path for this to work. If
1083		daVinci is not available, then the default viewer can be changed
1084		using:
1085
1086		Control.MLRISC.getString "viewer"
1087
1088		which can be set to "dot" or "vcg" for the corresponding viewers. Of
1089		course, these viewers must be in your path.
1090
1091		The above will display the compilation unit at the level of clusters,
1092		many of which are small, boring, and un-interesting. Also setting:
1093
1094		Control.MLRISC.getInt "cfg-graphical-view_size"
1095
1096		will display clusters that are larger than the value set by the above.
1097
1098
1099		----------------------------------------------------------------------
1100		Name: Matthias Blume
1101		Date: 2002/03/21 22:20:00 EST
1102		Tag: blume-20020321-kmp-bugfix
1103		Description:
1104
1105		Changed the interface to the KMP routine in PreString and fixed
1106		a minor bug in one place where it was used.
1107
1108		----------------------------------------------------------------------
1109		Name: Allen Leung
1110		Date: 2002/03/21 20:30:00 EST
1111		Tag: leunga-20020321-cfg
1112		Description:
1113
1114		Fixed a potential problem in cfg edge splitting.
1115
1116		----------------------------------------------------------------------
1117		Name: Allen Leung
1118		Date: 2002/03/21 17:15:00 EST
1119		Tag: leunga-20020321-x86-fp-cfg
1120		Description:
1121
1122		1. Recoded the buggy parts of x86-fp.
1123
1124		a. All the block reordering code has been removed.
1125		We now depend on the block placement phases to do this work.
1126
1127		b. Critical edge splitting code has been simplified and moved into the
1128		CFG modules, as where they belong.
1129
1130		Both of these were quite buggy and complex.  The code is now much, much
1131		simpler.
1132
1133		2. X86 backend.
1134
1135		a. Added instructions for 64-bit support.  Instruction selection for
1136		64-bit has not been committed, however, since that
1137		requires changes to MLTREE which haven't been approved by
1138		Lal and John.
1139
1140		b. Added support for FUCOMI and FUCOMIP when generating code for
1141		PentiumPro and above.  We only generate these instructions in
1142		the fast-fp mode.
1143
1144		c. Added cases for JP and JNP in X86FreqProps.
1145
1146		3. CFG
1147
1148		CFG now has a bunch of methods for edge splitting and merging.
1149
1150		4. Machine description.
1151
1152		John's simplification of MLTREE_BASIS.fcond broke a few machine
1153		description things:
1154
1155		rtl-build.{sig,sml} and hppa.mdl fixed.
1156
1157		NOTE: the machine description stuff in the repository is still broken.
1158		Again, I can't put my fixes in because that involves
1159		changes to MLTREE.
1160
1161		----------------------------------------------------------------------
1162		Name: Matthias Blume
1163		Date: 2002/03/20 15:55:00 EST
1164		Tag: blume-20020320-kmp
1165		Description:
1166
1167		Implemented Knuth-Morris-Pratt string matching in PreString and used
1168		it for String.isSubstring, Substring.isSubstring, and
1169		Substring.position.
1170
1171		(Might need some stress-testing.  Simple examples worked fine.)
1172
1173		----------------------------------------------------------------------
1174		Name: Matthias Blume
1175		Date: 2002/03/19 16:37:00 EST
1176		Tag: blume-20020319-witnesses
1177		Description:
1178
1179		Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1180
1181		This implements a generic mechanism for changing constness qualifiers
1182		anywhere within big C types without resorting to outright "casts".
1183		(So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1184		modify the constness at the outermost level.)
1185		The implementation of "convert" is based on the idea of "witness"
1186		values -- values that are not used by the operation but whose types
1187		"testify" to their applicability.  On the implementation side, "convert"
1188		is simply a projection (returning its second curried argument).  With
1189		cross-module inlining, it should not result in any machine code being
1190		generated.
1191
1192		----------------------------------------------------------------------
1193		Name: Matthias Blume
1194		Date: 2002/03/15 16:40:00 EST
1195		Tag: blume-20020315-basis
1196		Description:
1197
1198		Provided (preliminary?) implementations for
1199
1200		{String,Substring}.{concatWith,isSuffix,isSubstring}
1201
1202		and
1203
1204		Substring.full
1205
1206		Those are in the Basis spec but they were missing in SML/NJ.
1207
1208		----------------------------------------------------------------------
1209		Name: Matthias Blume
1210		Date: 2002/03/14 21:30:00 EST
1211		Tag: blume-20020314-controls
1212		Description:
1213
1214		Controls:
1215		---------
1216
1217		1. Factored out the recently-added Controls : CONTROLS stuff and put
1218		it into its own library $/controls-lib.cm.  The source tree for
1219		this is under src/smlnj-lib/Controls.
1220
1221		2. Changed the names of types and functions in this interface, so they
1222		make a bit more "sense":
1223
1224		module -> registry
1225		'a registry -> 'a group
1226
1227		3. The interface now deals in ref cells only.  The getter/setter interface
1228		is (mostly) gone.
1229
1230		4. Added a function that lets one register an already-existing ref cell.
1231
1232		5. Made the corresponding modifications to the rest of the code so that
1233		everything compiles again.
1234
1235		6. Changed the implementation of Controls.MLRISC back to something closer
1236		to the original.  In particular, this module (and therefore MLRISC)
1237		does not depend on Controls.  There now is some link-time code in
1238		int-sys.sml that registers the MLRISC controls with the Controls
1239		module.
1240
1241		CM:
1242		---
1243
1244		* One can now specify the lambda-split aggressiveness in init.cmi.
1245
1246		----------------------------------------------------------------------
1247		Name: Allen Leung
1248		Date: 2002/03/13 17:30:00 EST
1249		Tag: leunga-20020313-x86-fp-unary
1250		Description:
1251
1252		Bug fix for:
1253
1254		> leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1255		> Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1256		> - fun f(x,(y,z)) = Real.~ y;
1257		> [autoloading]
1258		> [autoloading done]
1259		>       fchsl   (%eax), 184(%esp)
1260		> Error: MLRisc bug: X86MCEmitter.emitInstr
1261		>
1262		> uncaught exception Error
1263		>   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1264
1265		The problem was that the code generator did not generate any fp registers
1266		in this case, and the ra didn't know that it needed to run the X86FP phase to
1267		translate the pseudo fp instruction.   This only happened with unary fp
1268		operators in certain situations.
1269
1270		----------------------------------------------------------------------
1271		Name: Matthias Blume
1272		Date: 2002/03/13 14:00:00 EST
1273		Tag: blume-20020313-overload-etc
1274		Description:
1275
1276		1. Added _overload as a synonym for overload for backward compatibility.
1277		(Control.overloadKW must be true for either version to be accepted.)
1278
1279		2. Fixed bug in install script that caused more things to be installed
1280		than what was requested in config/targets.
1281
1282		3. Made CM aware of the (_)overload construct so that autoloading
1283		works.
1284
1285		----------------------------------------------------------------------
1286		Name: Matthias Blume
1287		Date: 2002/03/12 22:03:00 EST
1288		Tag: blume-20020312-url
1289		Description:
1290
1291		Forgot to update BOOT and srcarchiveurl.
1292
1293		----------------------------------------------------------------------
1294		Name: Matthias Blume
1295		Date: 2002/03/12 17:30:00 EST
1296		Tag: blume-20020312-version110392
1297		Description:
1298
1299		Yet another version number bump (because of small changes to the
1300		binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1301
1302		Changes:
1303
1304		The new pid generation scheme described a few weeks ago was overly
1305		complicated.  I implemented a new mechanism that is simpler and
1306		provides a bit more "stability":  Once CM has seen a compilation
1307		unit, it keeps its identity constant (as long as you do not delete
1308		those crucial CM/GUID/* files).  This means that when you change
1309		an interface, compile, then go back to the old interface, and
1310		compile again, you arrive at the original pid.
1311
1312		There now also is a mechanism that instructs CM to use the plain
1313		environment hash as a module's pid (effectively making its GUID
1314		the empty string).  For this, "noguid" must be specified as an
1315		option to the .sml file in question within its .cm file.
1316		This is most useful for code that is being generated by tools such
1317		as ml-nlffigen (because during development programmers tend to
1318		erase the tool's entire output directory tree including CM's cached
1319		GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1320		revert to the old, broken behavior of SML/NJ, but in specific cases
1321		where there is no danger of interface confusion, its use is ok
1322		(I think).
1323
1324		ml-nlffigen by default generates "noguid" annotations.  They can be
1325		turned off by specifying -guid in its command line.
1326
1327		----------------------------------------------------------------------
1328		Name: Lal George
1329		Date: 2002/03/12 12 14:42:36 EST
1330		Tag: george-20020312-frequency-computation
1331		Description:
1332
1333		Integrated jump chaining and static block frequency into the
1334		compiler. More details and numbers later.
1335
1336		----------------------------------------------------------------------
1337		Name: Lal George
1338		Date: 2002/03/11 11 22:38:53 EST
1339		Tag: george-20020311-jump-chain-elim
1340		Description:
1341
1342		Tested the jump chain elimination on all architectures (except the
1343		hppa).  This is on by default right now and is profitable for the
1344		alpha and x86, however, it may not be profitable for the sparc and ppc
1345		when compiling the compiler.
1346
1347		The gc test will typically jump to a label at the end of the cluster,
1348		where there is another jump to an external cluster containing the actual
1349		code to invoke gc. This is to allow factoring of common gc invocation
1350		sequences. That is to say, we generate:
1351
1352		f:
1353		testgc
1354		ja   L1      % jump if above to L1
1355
1356		L1:
1357		jmp L2
1358
1359
1360		After jump chain elimination the 'ja L1' instructions is converted to
1361		'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1362		up being implemented in their long form (if L2 is far away) using:
1363
1364		jbe     L3      % jump if below or equal to L3
1365		jmp     L2
1366		L3:
1367		...
1368
1369
1370		For large compilation units L2  may be far away.
1371
1372
1373		----------------------------------------------------------------------
1374		Name: Matthias Blume
1375		Date: 2002/03/11 13:30:00 EST
1376		Tag: blume-20020311-mltreeeval
1377		Description:
1378
1379		A functor parameter was missing.
1380
1381		----------------------------------------------------------------------
1382		Name: Allen Leung
1383		Date: 2002/03/11 10:30:00 EST
1384		Tag: leunga-20020311-runtime-string0
1385		Description:
1386
1387		The representation of the empty string now points to a
1388		legal null terminated C string instead of unit.  It is now possible
1389		to convert an ML string into C string with InlineT.CharVector.getData.
1390		This compiles into one single machine instruction.
1391
1392		----------------------------------------------------------------------
1393	Name: Allen Leung	Name: Allen Leung
1394	Date: 2002/03/10 23:55:00 EST	Date: 2002/03/10 23:55:00 EST
1395	Tag: leunga-20020310-x86-call	Tag: leunga-20020310-x86-call

---

Colored Diff Long Colored Diff Unidiff Context Diff Side by Side

Legend: Removed from v.1130   changed lines   Added in v.1387

---

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