Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 1085, Fri Feb 22 00:15:55 2002 UTC	revision 1455, Fri Feb 13 20:42:49 2004 UTC
#	Line 13	Line 13
13	Description:	Description:
14
15	----------------------------------------------------------------------	----------------------------------------------------------------------
16		Name: Matthias Blume (blume (at) tti - c (dot) org)
17		Date: 2004/02/13 14:50:00 CST
18		Tag: Release_110_45
19		Description:
20
21		New working version (110.45). New bootfiles.
22
23		----------------------------------------------------------------------
24		Name: Matthias Blume (blume (at) tti - c (dot) org)
25		Date: 2004/01/26 15:15:15 CST
26		Tag: blume-20040126-toplevel
27		Description:
28
29		Improved handling of exceptions at the interactive toplevel.
30
31		----------------------------------------------------------------------
32		Name: Matthias Blume (blume (at) tti - c (dot) org)
33		Date: 2004/01/26 11:25:00
34		Tag: blume-20040126-app
35		Description:
36
37		Type of top-level "app" corrected.
38		Added code for setting vp_limitPtrMask to Win32-specific runtime.
39
40		----------------------------------------------------------------------
41		Name: Matthias Blume (blume (at) tti - c (dot) org)
42		Date: 2003/11/18 17:10 CST
43		Tag: blume-20031118-basis-fiddle
44		Description:
45
46		- changed Timer interface to what might become the spec
47		- POSIX_FLAGS -> BIT_FLAGS according to spec
48		- some other minor discrepancies wrt. spec eliminated
49
50		----------------------------------------------------------------------
51		Name: Matthias Blume (blume (at) tti - c (dot) org)
52		Date: 2003/11/06 12:00:00 CST
53		Tag: Release_110_44
54		Description:
55
56		New working version (110.44). New bootfiles.
57
58		----------------------------------------------------------------------
59		Name: Matthias Blume (blume (at) tti - c (dot) org)
60		Date: 2003/11/04 11:50:00 CST
61		Tag: blume-20031104-move-libraries
62		Description:
63
64		Eliminated the "dont_move_libraries" directive in config/targets.
65		(The mechanism was broken and could not be fixed easily.  Moreover,
66		there does not seem to be any reason not to move all libraries into
67		lib during installation.  I originally implemented this directive as a
68		backward-compatibility feature when I first introduced the new CM.
69		Now that things have been stable for a long time and going back to the
70		old CM is not an option, there is no reason to keep it around.)
71
72		----------------------------------------------------------------------
73		Name: Matthias Blume (blume (at) tti - c (dot) org)
74		Date: 2003/11/03 16:00:00 CST
75		Tag: blume-20031103-installdir
76		Description:
77
78		Made installer honor INSTALLDIR variable again.  (Thanks to Chris
79		Richards for pointing out the problem and providing the solution.)
80
81		----------------------------------------------------------------------
82		Name: Matthias Blume (blume (at) tti - c (dot) org)
83		Date: 2003/10/01 17:05:00 CDT
84		Tag: blume-20031001-lal-mlrisc
85		Description:
86
87		MLRISC bug fix from Lal.
88
89		----------------------------------------------------------------------
90		Name: Matthias Blume (blume (at) tti - c (dot) org)
91		Date: 2003/09/30 16:10:00 CDT
92		Tag: blume-20030930-primio-bat
93		Description:
94
95		1. Added openVector, nullRd, and nullWr to PRIM_IO.
96		2. Improved .bat files (for Win32 port) to make things work under Win95.
97		(thanks to Aaron S. Hawley for this one)
98
99		----------------------------------------------------------------------
100		Name: Matthias Blume (blume (at) tti - c (dot) org)
101		Date: 2003/09/26 16:05:00 CDT
102		Tag: blume-20030926-wrappriv
103		Description:
104
105		Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
106
107		----------------------------------------------------------------------
108		Name: Matthias Blume (blume (at) tti - c (dot) org)
109		Date: 2003/09/26 15:00:00 CDT
110		Tag: blume-20030926-110_43_3
111		Description:
112
113		- additional cleanup
114		- version number bump, NEW BOOTFILES
115
116		----------------------------------------------------------------------
117		Name: Matthias Blume (blume (at) tti - c (dot) org)
118		Date: 2003/09/26 12:00:00 CDT
119		Tag: blume-20030926-ppautoload
120		Description:
121
122		I modified the read-eval-print loop so that the autoloader gets
123		invoked whenever the prettyprinter tries to look up a symbol that
124		is not currently defined in the toplevel environment but which
125		appears in CM's autoload registry.  As a result, we see far fewer of
126		those ?.Foo.Bar.xxx names in the prettyprinter's output.
127
128		In addition to this I tried to clean up some pieces of the Basis
129		implementation (e.g., Socket, Word8Array) in order to prevent other
130		instances of these ?.Foo.Bar.xxx names from being printed.
131
132		The mechanism that picks names for types still needs some work, though.
133		(Right now it seems that if there is a type A.t which is defined to
134		be B.u, but B is unavailable at toplevel, then A.t gets printed as
135		"?.B.u" although the perhaps more sensible solution would be to use
136		"A.t" in this case.  In other words, the prettyprinter should follow
137		a chain of DEFtycs not farther than there are corresponding toplevel
138		names in the current environment.)
139
140		----------------------------------------------------------------------
141		Name: Matthias Blume (blume (at) tti - c (dot) org)
142		Date: 2003/09/24 16:31:00 CDT
143		Tag: blume-20030924-installer
144		Description:
145
146		Another installer tweak:  All the ML code for the installer is now
147		compiled during CMB.make and put into a little library called
148		$smlnj/installer.cm.  The installation then simply invokes
149
150		sml -m $smlnj/installer.cm
151
152		and everything happens automagically.
153
154		Win32:  ML code senses value of environment variable SMLNJ_HOME.
155		Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
156		and BINDIR.
157
158		The new scheme guarantees that the ML code responsible for the installation
159		is in sync with the APIs of the main system.  Also, the installer is
160		somewhat faster because the installer script is precompiled.
161
162		----------------------------------------------------------------------
163		Name: Matthias Blume (blume (at) tti - c (dot) org)
164		Date: 2003/09/24 15:35:00 CDT
165		Tag: blume-20030924-synsock
166		Description:
167
168		Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
169		but excludes all non-blocking operations.  Defined SOCKET (in Basis)
170		and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
171		implementations of non-blocking operations from CML's Socket
172		structure.
173
174		----------------------------------------------------------------------
175		Name: Matthias Blume (blume (at) tti - c (dot) org)
176		Date: 2003/09/24 15:10:05 CDT
177		Tag: blume-20030924-sockets
178		Description:
179
180		1. Fixed SOCKET API and implementation to match Basis spec.
181		This required changing the internal representation of sockets to one
182		that remembers (for each socket file descriptor) whether it is currently
183		blocking or non-blocking.  This state is maintained lazily (i.e., a system
184		call is made only if the state actually needs to change).
185
186		2. OS-specific details of sockets were moved into separate files, thus
187		making it possible to unify the bulk of the socket implementations
188		between Unix and Win32.
189
190		3. CML's socket API changed accordingly.
191		(Note that we need to remove non-blocking functions from this API
192		since they are redundant in the case of CML!)
193
194		4. CML's socket implementation now makes use of non-blocking functions
195		provided by Basis, thus removing all OS-dependent code from this part
196		of CML.
197
198		5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
199
200		----------------------------------------------------------------------
201		Name: Matthias Blume (blume (at) tti - c (dot) org)
202		Date: 2003/09/22 12:10:00 CDT
203		Tag: blume-20030922-110_43_2
204		Description:
205
206		Made a new interim version and bootfiles for developer's bootstrapping
207		convenience.
208
209		110.43.2 -- NEW BOOTFILES
210
211		----------------------------------------------------------------------
212		Name: Matthias Blume (blume (at) tti - c (dot) org)
213		Date: 2003/09/19 15:55:00 CDT
214		Tag: blume-20030919-cmdir
215		Description:
216
217		1. new-install.sh -> install.sh
218		2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
219		3. tweaked installer so that another name instead of .cm can be chosen
220		at install time (by setting the CM_DIR_ARC environment variable
221		during installation); once installation is complete, the name is
222		fixed
223
224		----------------------------------------------------------------------
225		Name: Matthias Blume (blume (at) tti - c (dot) org)
226		Date: 2003/09/18 16:00:00 CDT
227		Tag: blume-20030918-110_43_1
228		Description:
229
230		Made a new interim version and bootfiles for developer's bootstrapping
231		convenience.
232
233		110.43.1 -- NEW BOOTFILES
234
235		----------------------------------------------------------------------
236		Name: Matthias Blume (blume (at) tti - c (dot) org)
237		Date: 2003/09/18 15:20:00 CDT
238		Tag: blume-20030918-misc
239		Description:
240
241		1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
242		this seems to be controversial at the moment) and used those in
243		Posix.ProcEnv.times.
244
245		2. Added Time.{from,to}Nanoseconds to Time.
246
247		3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
248		For example, fromLargeInt never needs to look at more than 3 "big
249		digits" to get its 53 bits of precision.
250
251		----------------------------------------------------------------------
252		Name: Matthias Blume (blume (at) tti - c (dot) org)
253		Date: 2003/09/17 16:30:00 CDT
254		Tag: blume-20030917-real32-slices
255		Description:
256
257		Added an entry to the primitive environment
258		(compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
259		added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
260
261		Removed some of the "magic" constants in real64.sml and replaced them
262		with code that generates these values from their corresponding
263		integer counterparts.
264
265		Made all(?) the slice-related changes to the Basis and made everything
266		compile again...
267
268		----------------------------------------------------------------------
269		Name: Matthias Blume (blume (at) tti - c (dot) org)
270		Date: 2003/09/15 17:45:00 CDT
271		Tag: blume-20030915-rbase
272		Description:
273
274		Fixed bug in Real.fromLargeInt.
275
276		----------------------------------------------------------------------
277		Name: Matthias Blume (blume (at) tti - c (dot) org)
278		Date: 2003/09/13 18:11:00 CDT
279		Tag: blume-20030913-libinstall
280		Description:
281
282		Minor bugfix in config/libinstall (set anchor with path to
283		standalone tool after installing it, otherwise libraries that
284		need ml-lex or ml-yacc won't compile the first time the installer
285		runs).
286
287		----------------------------------------------------------------------
288		Name: Matthias Blume (blume (at) tti - c (dot) org)
289		Date: 2003/09/12 11:45:00 CDT
290		Tag: blume-20030912-various
291		Description:
292
293		- fixed bug in Real.toLargeInt
294		- fixed bug in Posix.ProcEnv.times
295		- changed inputLine functions to return an option
296		- minor installer improvements / bugfixes
297		- changed default @SMLalloc parameter for x86/celeron to 64k
298
299		----------------------------------------------------------------------
300		Name: Matthias Blume (blume (at) tti - c (dot) org)
301		Date: 2003/09/09 22:00:00 CDT
302		Tag: Release_110_43
303		Description:
304
305		New working release 110.43.  New bootfiles.
306
307		----------------------------------------------------------------------
308		Name: Matthias Blume (blume (at) tti - c (dot) org)
309		Date: 2003/09/09 19:20:00 CDT
310		Tag: blume-20030909-installer
311		Description:
312
313		Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
314		Modified config/install.bat to take advantage of it.  Also modified
315		config/install.sh (and called it config/new-install.sh) to take advantage
316		of it on Unix systems.  (The SML code is (supposed to be) platform-
317		independent.)
318
319		The installer can now install everything under Win32
320		as well as under *nix as long as it compiles.
321
322		Other changes:
323
324		- made CML compile again under Win32
325		- made eXene compile under Win32 (by providing a fake structure UnixSock
326		and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
327		- fixed a bug in nowhere: it assumed that type OS.Process.status is the
328		same as type int;  under Win32 it isn't
329		- fixed some slice-related problems in the win32-specific parts of CML
330		- added a functor argument "sameVol" to os-path-fn.sml in the Basis
331		(under Win32, the volume name is case-insensitive, and the
332		OS.Path code compares volume names for equality)
333
334		----------------------------------------------------------------------
335		Name: Matthias Blume (blume (at) tti - c (dot) org)
336		Date: 2003/09/08 11:55:00 CDT
337		Tag: blume-20030908-fullpath
338		Description:
339
340		Made Win32 version of OS.FileSys.fullPath return current directory
341		when given an empty string.  This is what the spec says, and incidentally,
342		CM depends on it.  (CM otherwise goes into an infinite loop in certain
343		cases when presented with the name of a non-existing .cm file.)
344
345		----------------------------------------------------------------------
346		Name: Matthias Blume (blume (at) tti - c (dot) org)
347		Date: 2003/09/04 16:30:00 CDT
348		Tag: blume-20030905-slices-etc
349		Description:
350
351		1. Changed interface to vectors and arrays in Basis to match
352		(draft) Basis spec.
353		2. Added signatures and implementations of slices according to
354		Basis spec.
355		3. Edited source code throughout the system to make it compile again
356		under 1. and 2.  (In some cases code had to be added to have it
357		match the new signatures.)
358		4. MLRISC should be backward-compatible: the copies of the originals
359		of files that needed to change under 3. were retained, the .cm files
360		check the compiler version number and use old versions when
361		appropriate.
362		5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
363		dirstream -> string option (in accordance with Basis spec).
364		6. When generating code that counts lines, ml-lex used function
365		CharVector.foldli, taking advantage of its old interface.
366		This has been replaced with the corresponding code from
367		CharVectorSlice.  (html-lex must be re-lexed!)
368		7. BitArray in smlnj-lib/Util has been extended/modified to match the
369		new MONO_ARRAY signature.  (Do we need BitArraySlice?)
370		8. Removed temporary additions (fromInternal, toInternal) from the
371		(now obsolete) IntInf in smlnj-lib/Util.
372		9. Cleaned up structure Byte.
373		10. Added localOffset, scan, and fromString to Date (according to spec).
374		Cleaned/corrected implementation of Date.
375		(Still need to check for correctness; implement better canonicalizeDate.)
376		11. Added "scan" to signature IEEE_REAL.
377		12. Some improvements to IntInf [in particular: efficiency-hack for
378		mod and rem when second operand is 2 (for parity checks).]
379		13. Changed representation of type Time.time, using a single IntInf.int
380		value counting microseconds.  This considerably simplified the
381		implementation of structure Time.  We now support negative time
382		values; scan and fromString handle signs.
383		14. Functor PrimIO now takes two additional arguments (VectorSlice and
384		ArraySlice).
385
386		----------------------------------------------------------------------
387		Name: Matthias Blume (blume (at) tti - c (dot) org)
388		Date: 2003/08/28 17:00:00 CDT
389		Tag: blume-20030828-intinf
390		Description:
391
392		This is a major update which comes with a version number bump
393		(110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
394		and an implementation of IntInf in the Basis.
395
396		There are a fairly large number of related changes and updates throughout
397		the system:
398
399		Basis:
400		- Implemented IntInf.
401		- Made LargeInt a projection of IntInf (by filtering through INTEGER).
402		- Added some missing Real64 operations, most notably Real.toLargeInt.
403		- Added FixedInt as a synonym for Int32.
404
405		compiler:
406		* Added support for a built-in intinf type.
407		- literals
408		- pattern matching
409		- conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
410		- overloading on literals and operations
411
412		This required adding a primitive type intinf, some additional
413		primops, and implementations for several non-trivial intinf
414		operations in Core.  (The intinf type is completely abstract
415		to the compiler; all operations get delegated back to the Core.)
416
417		* Intinf equality is handled by polyequal.  However, the compiler
418		does not print its usual warning in this case (since polyequal
419		is the right thing to do there).
420
421		* Improved the organization of structure InlineT.
422
423		* A word about conversion primops:
424		If conversions involving intinf do not cancel out during
425		CPS contract, then the compiler must insert calls to Core functions.
426		Since all core access must be resolved already during the FLINT
427		translate phase, it would be too late a the time of CPS contract
428		to add new Core calls.  For this reason, conversion primops
429		for intinf carry two arguments:  1. the numeric argument that
430		they are supposed to convert, and 2. the Core function that
431		can help with this conversion if necessary.  If CPS contract
432		eliminates a primop, then the associated Core function becomes
433		dead and goes away.  Intinf conversion primops that do not get
434		eliminated by CPS contract get rewritten into calls of their
435		core functions by a separate, new phase.
436
437		interactive system:
438		- Control.Print.intinfDepth controls max length of intinf constants
439		being printed. (Analogous to Control.Print.stringDepth.)
440		- Cleanup in printutil and pputil: got rid of unused stuff and
441		duplicates; replaced some of the code with code that makes better
442		use of library functionality.
443
444		CM:
445		Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
446		are no longer silent.
447
448		CKIT:
449		Fixed mismatched uses of Int32 and LargeInt.  I always decided
450		in favor of LargeInt -- which is now the same as IntInf.
451		CKIT-knowledgable people should check whether this is what's
452		intended and otherwise change things back to using Int32 or
453		FixedInt.
454
455		Throughout the code:
456		Started using IntInf.int literals and built-in operations
457		(e.g., comparison with 0) where this seems appropriate.
458
459
460		----------------------------------------------------------------------
461		Name: Dave MacQueen (dbm@cs.uchicago.edu)
462		Date: 2003/08/13 11:36:00 CDT
463		Tag: dbm-20030813-mcz-merge1
464		Description:
465
466		Merging changes from the mcz-branch development branch into trunk.
467		These changes involve replacement of the emulated old prettyprinter
468		interface with direct use of the SML/NJ Lib PP library, and fixing
469		of a couple of bugs (895, 1186) relating to error messages.  A new
470		prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
471		has been added.
472
473		----------------------------------------------------------------------
474		Name: Matthias Blume (blume (at) tti - c (dot) org)
475		Date: 2003/08/11 15:45:00 CDT
476		Tag: blume-20030811-windows
477		Description:
478
479		Version number bumped to 110.42.9.  NEW BOOTFILES!!!
480
481		http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
482
483		This patch restores SML/NJ's ability to run under win32.  There are a
484		number of changes, including fixes for several bugs that had gone
485		unnoticed until now:
486
487		- uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
488		trying to run the win32 version from within cygwin.)
489
490		- There are a number of simple .bat scripts that substitute for their
491		corresponding Unix shell-scripts.  (See below.)
492
493		- The internals of ml-build have been modified slightly.  The main
494		difference is that instead of calling ".link-sml" (or link-sml.bat)
495		using OS.Process.system, the ML process delegates this task back
496		to the script.  Otherwise problems arise in mixed environments such
497		as Cygwin where scripts look and work like Unix scripts, but
498		where OS.Process.system cannot run them.
499
500		- In CM, the srcpath pickler used native pathname syntax -- which
501		is incorrect in the case of cross-compilation.  The new pickle format
502		is independent of platform-specific naming conventions.
503
504		- Path configuration files (such as lib/pathconfig) can now choose
505		between native and standard syntax.  Placing a line of the form
506
507		standard!
508
509		into the file causes all subsequent paths to be interpreted using
510		CM standard pathname syntax (= Unix conventions); a line
511
512		native!
513
514		switches back to native style.  This was needed so that
515		path config files can be written portably, see src/system/pathconfig.
516
517		- Runtime system:
518
519		- win32-filesys.c:  get_file_time and set_file_time now
520		access modification time, not creation time.
521
522		- I/O code made aware of new array representation.
523
524		- Bug fixes in X86.prim.masm.
525
526		- src/system/makeml made aware of win32. (For use under cygwin
527		and other Unix-environments for windows.)
528
529		- In Basis, fixed off-by-one error in win32-io.sml (function vecF)
530		which caused BinIO.inputAll to fail consistently.
531
532		.bat scripts:
533
534		Windows .bat scripts assume that SMLNJ_HOME is defined.
535
536		- sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
537		applications (sml, ml-yacc, ml-lex).
538		- ml-build.bat: analogous to ml-build.
539		- config\install.bat:  Analogous to config/install.sh.  This requires
540		that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
541		(nmake etc. must be on the path, and vcvars32 must have been run.)
542		Moreover, sources for ml-lex and ml-yacc need to exist under src,
543		and the bootfile hierarchy must have been unpacked under
544		sml.boot.x86-win32.
545		The script is very primitive and does a poor job at error checking.
546		It only installs the base system, ml-lex, and ml-yacc.  No other
547		libraries are being installed (i.e., you get only those that
548		are part of the compiler.)
549		- link-sml.bat:  analogous to .link-sml, but not currently used
550
551		Unrelated bug fixes:
552
553		- ml-nlffigen now exports structures ST_* corresponding to incomplete
554		types.
555		- Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
556		otherwise.)
557
558		----------------------------------------------------------------------
559		Name: Dave MacQueen (macqueen@cs.uchicago.edu)
560		Date: 2003/06/17
561		Tag: macqueen-20030617-bug895
562		Description:
563
564		Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
565		Tag will be used for new development branch (mcz-branch) for
566		use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
567
568		----------------------------------------------------------------------
569		Name: Matthias Blume (blume (at) tti - c (dot) org)
570		Date: 2003/05/27 16:55:00 CDT
571		Tag: blume-20030527-polyeq
572		Description:
573
574		Tried to eliminated most cases of polymorphic equality.
575
576		----------------------------------------------------------------------
577		Name: Matthias Blume (blume (at) tti - c (dot) org)
578		Date: 2003/05/21 17:45:00 CDT
579		Tag: blume-20030517-complete
580		Description:
581
582		Two changes:
583
584		1. Added a flag for controlling whether non-exhaustive bindings will
585		be treated as errors (default is false).
586		2. Cleaned up the *entire* source tree so that CMB.make goes through
587		without a single non-exhaustive match- or bind warning.
588
589		----------------------------------------------------------------------
590		Name: Matthias Blume (blume (at) tti - c (dot) org)
591		Date: 2003/05/17 10:20:00 CDT
592		Tag: blume-20030517-absyn
593		Description:
594
595		1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
596
597		This mainly affects the quality of error messages.  However, some
598		of the code is now more straightforward than before. (Treatment of
599		the above four constructs in translate.sml is much simpler than
600		the "macro-expansion" that was going on before.  Plus, the mach-
601		compiler no longer gets invoked just to be able to compile an
602		if-expression.)
603
604		2. The ErrorMsg.Error exception is now caught and absorbed by the
605		interactive loop.
606
607		----------------------------------------------------------------------
608		Name: Allen Leung
609		Date: 2003/05/16 13:05:00 CDT
610		Tag: leunga-20030516-cygwin-runtime
611		Description:
612
613		Ported the runtime system to cygwin, which uses the unix
614		x86-unix bin files.  Missing/buggy features:
615
616		o getnetbyname, getnetbyaddr: these functions seem to be missing in
617		the Cygwin library.
618		o Ctrl-C handling may be flaky.
619		o Windows system calls and Windows I/O are not supported.
620
621		A new set of binfiles is located at:
622
623		http://www.dorsai.org/~leunga/boot.x86-unix.tgz
624
625		This is only needed for bootstrapping the cygwin version of smlnj.
626		Other x86 versions can use the existing binfiles.
627
628		----------------------------------------------------------------------
629		Name: Matthias Blume
630		Date: 2003/04/08 15:42:00 CDT
631		Tag: blume-20030408-listpair
632		Description:
633
634		1. Added a target 'mlrisc' to installer.
635
636		2. Added missing elements to structure ListPair.
637
638		----------------------------------------------------------------------
639		Name: Allen Leung
640		Date: 2003/01/07 10:40:00 EST
641		Tag: leunga-20030107-int-rem
642		Description:
643
644		Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
645		The arguments to the SUBL instruction were swapped.
646
647		----------------------------------------------------------------------
648		Name: Matthias Blume
649		Date: 2002/12/12 16:25:00 EST
650		Tag: blume-20021212-risc-ra
651		Description:
652
653		Fixed a serious bug in the rewrite code for FP spilling/reloading that
654		sent the RA into an infinite loop when floating point registers get
655		spilled.  (Because of this bug, e.g., nucleic stopped compiling between
656		110.37 and 110.38.)
657		There was another set of potential problems related to the handling of
658		MLRISC annotations (but those did not yet cause real problems, apparently).
659
660		----------------------------------------------------------------------
661		Name: Matthias Blume
662		Date: 2002/12/06 22:40:00 EST
663		Tag: blume-20021206-cm-fileid
664		Description:
665
666		Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
667		This fixes the problem of CM getting confused by files that suddenly
668		change their identity (e.g., by getting unlinked and recreated by some
669		text editor such as vi).  There might be a better/cheaper/cleaner way
670		of doing this, but for now this will have to do.
671
672		----------------------------------------------------------------------
673		Name: Matthias Blume
674		Date: 2002/10/28 09:50:00 EST
675		Tag: blume-20021028-typecheck
676		Description:
677
678		Exported structure Typecheck from $smlnj/viscomp/core.cm.
679
680		----------------------------------------------------------------------
681		Name: Matthias Blume
682		Date: 2002/10/17 09:10:00 EDT
683		Tag: Release_110_42
684		Description:
685
686		In good old tradition, there has been a slight hiccup so that we have
687		to patch 110.42 after the fact.   The old release tag has been replaced
688		(see below).
689
690		The change solves a problem with two competing approaches the
691		configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
692		each other's way.
693
694		This change only affects the runtime system code and the installer script.
695		(No new bootfiles.)
696
697		----------------------------------------------------------------------
698		Name: Matthias Blume
699		Date: 2002/10/16 12:00:00 EDT
700		Tag: Release_110_42_removed
701		Description:
702
703		New working release.  New bootfiles.
704
705		----------------------------------------------------------------------
706		Name: Matthias Blume
707		Date: 2002/10/10 13:10:00 EDT
708		Tag: blume-20021010-ppc-divs
709		Description:
710
711		The mltree operator DIVS must be implemented with an overflow check on
712		the PPC because the hardware indicates divide-by-zero using "overflow" as
713		well.
714
715		----------------------------------------------------------------------
716		Name: Matthias Blume
717		Date: 2002/07/23 11:45:00
718		Tag: blume-20020723-smlnj-home
719		Description:
720
721		Sml now senses the SMLNJ_HOME environment variable.  If this is set,
722		then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
723		CM_PATHCONFIG is also set), the path configuration file is assumed
724		to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
725		the entire tree to some other place and everything will "just work".
726
727		(Companion commands such as ml-build and ml-makedepend also sense this
728		variable.)
729
730		----------------------------------------------------------------------
731		Name: Matthias Blume
732		Date: 2002/07/12 21:19:00 EDT
733		Tag: blume-20020712-liveness
734		Description:
735
736		Exported two useful "step" functions from liveness module (MLRISC).
737
738		----------------------------------------------------------------------
739		Name: Matthias Blume
740		Date: 2002/07/05 16:00 EDT
741		Tag: Release_110_41
742		Description:
743
744		New working release.  New bootfiles.
745
746		----------------------------------------------------------------------
747		Name: Matthias Blume
748		Date: 2002/07/05 10:25:00 EDT
749		Tag: blume-20020705-btimp
750		Description:
751
752		Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
753		other clients can set up backtracing support.
754
755		----------------------------------------------------------------------
756		Name: Matthias Blume
757		Date: 2002/06/25 17:23:00 EDT
758		Tag: blume-20020625-fpmax
759		Description:
760
761		Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
762		case.  (The sense of the isNaN test was reversed -- which made min and
763		max always return their first argument.)
764
765		----------------------------------------------------------------------
766		Name: Matthias Blume
767		Date: 2002/06/11
768		Tag: blume-20020611-unixpath
769		Description:
770
771		Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
772
773		----------------------------------------------------------------------
774		Name: Matthias Blume
775		Date: 2002/06/10 16:35:00 EDT
776		Tag: blume-20020610-ieeereal
777		Description:
778
779		I back-ported my implementation of IEEEReal.fromString from the
780		idlbasis-devel branch so that we can test it.
781
782		Another small change is that ppDec tries to give more information
783		than just "<sig>" in the case of functors.  However, this code is
784		broken in some mysterious way if the functor's body's signature
785		has not been declared by ascription but gets inferred from the
786		implementation.  This needs fixing...
787
788		----------------------------------------------------------------------
789		Name: Matthias Blume
790		Date: 2002/05/31
791		Tag: blume-20020531-btrace-mode
792		Description:
793
794		Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
795		the wayside when I switched over to using Controls everywhere.)
796
797		----------------------------------------------------------------------
798		Name: Lal George
799		Date: 2002/05/23 12:21:40 EDT
800		Tag: george-20020523-visual-labels
801		Description:
802
803		Labels are now displayed in the graphical output to make
804		the fall-through and target blocks obvious.
805
806		----------------------------------------------------------------------
807		Name: Matthias Blume
808		Date: 2002/05/22 11:03:00 EDT
809		Tag: blume-20020522-shrink
810		Description:
811
812		John tweaked yesterday's fix for 1131 to handle an out-of-memory
813		situation that comes up when allocating huge arrays.
814
815		----------------------------------------------------------------------
816		Name: Matthias Blume
817		Date: 2002/05/21 16:00:00 EDT
818		Tag: Release_110_40
819		Description:
820
821		New working release (110.40).  New bootfiles.
822
823		[Also: John Reppy fixed GC bug 1131.]
824
825		----------------------------------------------------------------------
826		Name: Matthias Blume
827		Date: 2002/05/21 12:35:00 EDT
828		Tag: blume-20020521-cmdoc
829		Description:
830
831		CM documentation update.
832
833		----------------------------------------------------------------------
834		Name: Matthias Blume
835		Date: 2002/05/21 10:55:00 EDT
836		Tag: blume-20020521-misc
837		Description:
838
839		- John tweaked runtime to be silent on heap export (except when
840		GC messages are on).
841		- I added a few more things (cross-compiling versions of CMB) to
842		config/preloads (as suggestions).
843
844		----------------------------------------------------------------------
845		Name: Matthias Blume
846		Date: 2002/05/20 22:25:00 EDT
847		Tag: blume-20020520-controls
848		Description:
849
850		- Added ControlUtil structure to control-lib.cm.
851		- Use it throughout.
852		- Used Controls facility to define MLRISC controls (as opposed to
853		registering MLRISC control ref cells with Controls after the fact)
854		- Fixed messed-up controls priorities.
855
856		* Removed again all the stuff from config/preloads that one wouldn't
857		be able to preload at the time the initial heap image is built.
858		(Many libraries, e.g., CML, do not exist yet at this time.  The
859		only libraries that can be preloaded via config/preloads are those
860		that come bundled with the bootfiles.)
861
862		----------------------------------------------------------------------
863		Name: Matthias Blume
864		Date: 2002/05/20 10:59:00 EDT
865		Tag: blume-20020520-preloads
866		Description:
867
868		Added a lot of commented-out suggestions for things to be included
869		in config/preloads.
870
871		----------------------------------------------------------------------
872		Name: Allen Leung
873		Date: 2002/05/18 14:20:00 EDT
874		Tag: leunga-20020518-mdl
875		Description:
876
877		o Made the mdl tool stuff compile and run again.
878		o I've disabled all the stuff that depends on RTL specifications; they
879		are all badly broken anyway.
880
881		----------------------------------------------------------------------
882		Name: Matthias Blume
883		Date: 2002/05/17 16:49:00 EDT
884		Tag: blume-20020517-controls
885		Description:
886
887		0. John Reppy made several modifications to the SML/NJ library.
888		In particular, there is a shiny new controls-lib.cm.
889
890		1. Pushed new controls interface through compiler so that everything
891		compiles again.
892
893		2. Added FormatComb and FORMAT_COMB to the CML version of the
894		SML/NJ library (so that CML compiles again).
895
896		3. Modified init scripts because XXX_DEFAULT environment variables
897		are no longer with us.  (Boot-time initialization is now done
898		using the same environment variables that are also used for
899		startup-time initialization of controls.)
900
901		----------------------------------------------------------------------
902		Name: Lal George
903		Date: 2002/05/15 09:20:10 EDT
904		Tag: george-20020515-pseudo-op-decls
905		Description:
906
907		All pseudo-ops emitted before the first segment declaration
908		such as TEXT, DATA, and BSS directives are assumed to be global
909		declarations and are emitted first in the assembly file. This is
910		useful in a number of situations where one has pseudo-ops that are not
911		specific to any segment, and also works around the constraint that one
912		cannot have client pseudo-ops in the TEXT segment.
913
914		Because no segment is associated with these declarations it is
915		an error to allocate any space or objects before the first segment
916		directive and an exception will be raised. However, we cannot make
917		this check for client pseudo-ops.
918
919		These top level declarations are a field in the CFG graph_info.
920		In theory you can continue to add to this field after the CFG has been
921		built -- provided you know what you are doing;-)
922
923		----------------------------------------------------------------------
924		Name: Matthias Blume
925		Date: 2002/05/13 16:40:00 EDT
926		Tag: blume-20020513-pp-etc
927		Description:
928
929		A few minor bugfixes:
930
931		- Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
932		(Bogus printouts for redefined bindings still occur.  Compiler
933		bug should no longer occur now.  We need to redo the prettyprinter
934		from scratch.)
935
936		- CM pathname printer now also adds escape sequences for ( and )
937
938		- commend and docu fixes for ml-nlffi
939
940		----------------------------------------------------------------------
941		Name: Matthias Blume
942		Date: 2002/05/10 16:40:00 EDT
943		Tag: blume-20020510-erg-textio
944		Description:
945
946		Applied the following bugfix provided by Emden Gansner:
947
948		Output is corrupted when outputSubstr is used rather than output.
949
950		The problem occurs when a substring
951
952		ss = (s, dataStart, dataLen)
953
954		where dataStart > 0, fills a stream buffer with avail bytes left.
955		avail bytes of s, starting at index dataStart, are copied into the
956		buffer, the buffer is flushed, and then the remaining dataLen-avail
957		bytes of ss are copied into the beginning of the buffer. Instead of
958		starting this copy at index dataStart+avail in s, the current code
959		starts the copy at index avail.
960
961		Fix:
962		In text-io-fn.sml, change line 695 from
963		val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
964		to
965		val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
966
967		----------------------------------------------------------------------
968		Name: Matthias Blume
969		Date: 2002/04/12 13:55:00 EDT
970		Tag: blume-20020412-assyntax
971		Description:
972
973		1. Grabbed newer assyntax.h from the XFree86 project.
974		2. Fiddled with how to compile X86.prim.asm without warnings.
975		3. (Very) Minor cleanup in CM.
976
977		----------------------------------------------------------------------
978		Name: Matthias Blume
979		Date: 2002/04/01 (no joke!) 17:07:00 EST
980		Tag: blume-20020401-x86div
981		Description:
982
983		Added full support for div/mod/rem/quot on the x86, using the machine
984		instruction's two results (without clumsily recomputing the remainder)
985		directly where appropriate.
986
987		Some more extensive power-of-two support was added to the x86 instruction
988		selector (avoiding expensive divs, mods, and muls where they can be
989		replaced with cheaper shifts and masks).  However, this sort of thing
990		ought to be done earlier, e.g., within the CPS optimizer so that
991		all architectures benefit from it.
992
993		The compiler compiles to a fixed point, but changes might be somewhat
994		fragile nevertheless.  Please, report any strange things that you might
995		see wrt. div/mod/quot/rem...
996
997		----------------------------------------------------------------------
998		Name: Matthias Blume
999		Date: 2002/03/29 17:22:00
1000		Tag: blume-20020329-div
1001		Description:
1002
1003		Fixed my broken div/mod logic.  Unfortunately, this means that the
1004		inline code for div/mod now has one more comparison than before.
1005		Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1006		The problem was with quotient = 0, because that alone does not tell
1007		us which way the rounding went.  One then has to look at whether
1008		remainder and divisor have the same sign...  :(
1009
1010		Anyway, I replaced the bootfiles with fresh ones...
1011
1012		----------------------------------------------------------------------
1013		Name: Matthias Blume
1014		Date: 2002/03/29 14:10:00 EST
1015		Tag: blume-20020329-inlprims
1016		Description:
1017
1018		NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1019
1020		Primops have changed. This means that the bin/boot-file formats have
1021		changed as well.
1022
1023		To make sure that there is no confusion, I made a new version.
1024
1025
1026		CHANGES:
1027
1028		* removed REMT from mltree (remainder should never overflow).
1029
1030		* added primops to deal with divisions of all flavors to the frontend
1031
1032		* handled these primops all the way through so they map to their respective
1033		MLRISC support
1034
1035		* used these primops in the implementation of Int, Int32, Word, Word32
1036
1037		* removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1038
1039		* parameterized INLMIN, INLMAX, and INLABS by a numkind
1040
1041		* translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1042		floating point
1043
1044		* used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1045		and Real (but Real.abs maps to a separate floating-point-only primop)
1046
1047
1048		TODO items:
1049
1050		* Hacked Alpha32 instruction selection, disabling the selection of REMx
1051		instructions because the machine instruction encoder cannot handle
1052		them.  (Hppa, PPC, and Sparc instruction selection did not handle
1053		REM in the first place, and REM is supported by the x86 machine coder.)
1054
1055		* Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1056		selection phase.  (The two can be streamlined because the hardware
1057		delivers both quotient and remainder at the same time anyway.)
1058
1059		* Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1060		(Currently the behavior is inconsistent both across architectures and
1061		wrt. the draft Basis spec.)
1062
1063		* Word8 should eventually be handled natively, too.
1064
1065		* There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1066		as if there currently is no execution path that could trigger it in
1067		SML/NJ.  (The assumptions underlying functions arith and promotable do not
1068		hold for things like multiplication and division.)
1069
1070		----------------------------------------------------------------------
1071		Name: Matthias Blume
1072		Date: 2002/03/27 16:27:00 EST
1073		Tag: blume-20020327-mlrisc-divisions
1074		Description:
1075
1076		Added support for all four division operations (ML's div, mod, quot,
1077		and rem) to MLRISC.  In the course of doing so, I also rationalized
1078		the naming (no more annoying switch-around of DIV and QUOT), by
1079		parameterizing the operation by div_rounding_mode (which can be either
1080		DIV_TO_ZERO or DIV_TO_NEGINF).
1081
1082		The generic MLTreeGen functor takes care of compiling all four
1083		operations down to only round-to-zero div.
1084
1085		Missing pieces:
1086
1087		* Doing something smarter than relying on MLTreeGen on architectures
1088		like, e.g., the x86 where hardware division delivers both quotient and
1089		remainder at the same time.  With this, the implementation of the
1090		round-to-neginf operations could be further streamlined.
1091
1092		* Remove inlining support for div/mod/rem from the frontend and replace it
1093		with primops that get carried through to the backend.  Do this for all
1094		int and word types.
1095
1096		----------------------------------------------------------------------
1097		Name: Matthias Blume
1098		Date: 2002/03/25 17:25:00 EST
1099		Tag: blume-20020325-divmod
1100		Description:
1101
1102		I improved (hopefully without breaking them) the implementation of Int.div,
1103		Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1104		advantage of the following observations:
1105
1106		Let  q = x quot y      r = x rem y
1107		d = x div  y      m = x mod y
1108
1109		where "quot" is the round-to-zero version of integer division that
1110		hardware usually provides.  Then we have:
1111
1112		r = x - q * y        where neither the * nor the - will overflow
1113		d = if q >= 0 orelse x = q * y then q else q - 1
1114		where neither the * nor the - will overflow
1115		m = if q >= 0 orelse r = 0 then r else r + y
1116		where the + will not overflow
1117
1118		This results in substantial simplification of the generated code.
1119		The following table shows the number of CFG nodes and edges generated
1120		for
1121		fun f (x, y) = x OPER y
1122		(* with OPER \in div, mod, quot, rem *)
1123
1124
1125		OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1126		--------------------------------------------------------
1127		div |         24 |         39 |         12 |         16
1128		mod |         41 |         71 |         12 |         16
1129		quot |          8 |         10 |          8 |         10
1130		rem |         10 |         14 |          8 |         10
1131
1132
1133		----------------------------------------------------------------------
1134		Name: Matthias Blume
1135		Date: 2002/03/25 22:06:00 EST
1136		Tag: blume-20020325-cprotobug
1137		Description:
1138
1139		Fixed a bug in cproto (c prototype decoder).
1140
1141		----------------------------------------------------------------------
1142		Name: Matthias Blume
1143		Date: 2002/03/25 16:00:00 EST
1144		Tag: blume-20020325-raw-primops
1145		Description:
1146
1147		I did some cleanup to Allen's new primop code and
1148		replaced yesterday's bootfiles with new ones.
1149		(But they are stored in the same place.)
1150
1151		----------------------------------------------------------------------
1152		Name: Matthias Blume
1153		Date: 2002/03/24 22:40:00 EST
1154		Tag: blume-20020324-bootfiles
1155		Description:
1156
1157		Made the bootfiles that Allen asked for.
1158
1159		----------------------------------------------------------------------
1160		Name: Allen Leung
1161		Date: 2002/03/23 15:50:00 EST
1162		Tag: leunga-20020323-flint-cps-rcc-primops
1163		Description:
1164
1165		1. Changes to FLINT primops:
1166
1167		(* make a call to a C-function;
1168		* The primop carries C function prototype information and specifies
1169		* which of its (ML-) arguments are floating point. C prototype
1170		* information is for use by the backend, ML information is for
1171		* use by the CPS converter. *)
1172		| RAW_CCALL of { c_proto: CTypes.c_proto,
1173		ml_args: ccall_type list,
1174		ml_res_opt: ccall_type option,
1175		reentrant : bool
1176		} option
1177		(* Allocate uninitialized storage on the heap.
1178		* The record is meant to hold short-lived C objects, i.e., they
1179		* are not ML pointers.  With the tag, the representation is
1180		* the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1181		*)
1182		| RAW_RECORD of {tag:bool,sz:int}
1183		and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1184
1185		2.  These CPS primops are now overloaded:
1186
1187		rawload of {kind:numkind}
1188		rawstore of {kind:numkind}
1189
1190		The one argument form is:
1191
1192		rawload {kind} address
1193
1194		The two argument form is:
1195
1196		rawload {kind} [ml object, byte-offset]
1197
1198		3. RAW_CCALL/RCC now takes two extra arguments:
1199
1200		a. The first is whether the C call is reentrant, i.e., whether
1201		ML state should be saved and restored.
1202		b. The second argument is a string argument specifying the name of
1203		library and the C function.
1204
1205		These things are currently not handled in the code generator, yet.
1206
1207		4. In CProto,
1208
1209		An encoding type of "bool" means "ml object" and is mapped into
1210		C prototype of PTR.  Note that "bool" is different than "string",
1211		even though "string" is also mapped into PTR, because "bool"
1212		is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1213
1214		5. Pickler/unpicker
1215
1216		Changed to handle RAW_RECORD and newest RAW_CCALL
1217
1218		6. MLRiscGen,
1219
1220		1. Changed to handle the new rawload/rawstore/rawrecord operators.
1221		2. Code for handling C Calls has been moved to a new module CPSCCalls,
1222		in the file CodeGen/cpscompile/cps-c-calls.sml
1223
1224		7. Added the conditional move operator
1225
1226		condmove of branch
1227
1228		to cps.  Generation of this is still buggy so it is currently
1229		disabled.
1230
1231		----------------------------------------------------------------------
1232		Name: Lal George
1233		Date: 2002/03/22 14:18:25 EST
1234		Tag: george-20020322-cps-branch-prob
1235		Description:
1236
1237		Implemented the Ball-Larus branch prediction-heuristics, and
1238		incorporated graphical viewers for control flow graphs.
1239
1240		Ball-Larus Heuristics:
1241		---------------------
1242		See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1243
1244		By design it uses the Dempster-Shafer theory for combining
1245		probabilities.  For example, in the function:
1246
1247		fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1248
1249		the ball-larus heuristics predicts that the n=0 is unlikely
1250		(OH-heuristic), and the 'then' branch is unlikely because of the
1251		RH-heuristic -- giving the 'then' branch an even lower combined
1252		probability using the Dempster-Shafer theory.
1253
1254		Finally, John Reppy's loop analysis in MLRISC, further lowers the
1255		probability of the 'then' branch because of the loop in the else
1256		branch.
1257
1258
1259		Graphical Viewing:
1260		------------------
1261		I merely plugged in Allen's graphical viewers into the compiler. The
1262		additional code is not much. At the top level, saying:
1263
1264		Control.MLRISC.getFlag "cfg-graphical-view" := true;
1265
1266		will display the graphical view of the control flow graph just before
1267		back-patching.  daVinci must be in your path for this to work. If
1268		daVinci is not available, then the default viewer can be changed
1269		using:
1270
1271		Control.MLRISC.getString "viewer"
1272
1273		which can be set to "dot" or "vcg" for the corresponding viewers. Of
1274		course, these viewers must be in your path.
1275
1276		The above will display the compilation unit at the level of clusters,
1277		many of which are small, boring, and un-interesting. Also setting:
1278
1279		Control.MLRISC.getInt "cfg-graphical-view_size"
1280
1281		will display clusters that are larger than the value set by the above.
1282
1283
1284		----------------------------------------------------------------------
1285		Name: Matthias Blume
1286		Date: 2002/03/21 22:20:00 EST
1287		Tag: blume-20020321-kmp-bugfix
1288		Description:
1289
1290		Changed the interface to the KMP routine in PreString and fixed
1291		a minor bug in one place where it was used.
1292
1293		----------------------------------------------------------------------
1294		Name: Allen Leung
1295		Date: 2002/03/21 20:30:00 EST
1296		Tag: leunga-20020321-cfg
1297		Description:
1298
1299		Fixed a potential problem in cfg edge splitting.
1300
1301		----------------------------------------------------------------------
1302		Name: Allen Leung
1303		Date: 2002/03/21 17:15:00 EST
1304		Tag: leunga-20020321-x86-fp-cfg
1305		Description:
1306
1307		1. Recoded the buggy parts of x86-fp.
1308
1309		a. All the block reordering code has been removed.
1310		We now depend on the block placement phases to do this work.
1311
1312		b. Critical edge splitting code has been simplified and moved into the
1313		CFG modules, as where they belong.
1314
1315		Both of these were quite buggy and complex.  The code is now much, much
1316		simpler.
1317
1318		2. X86 backend.
1319
1320		a. Added instructions for 64-bit support.  Instruction selection for
1321		64-bit has not been committed, however, since that
1322		requires changes to MLTREE which haven't been approved by
1323		Lal and John.
1324
1325		b. Added support for FUCOMI and FUCOMIP when generating code for
1326		PentiumPro and above.  We only generate these instructions in
1327		the fast-fp mode.
1328
1329		c. Added cases for JP and JNP in X86FreqProps.
1330
1331		3. CFG
1332
1333		CFG now has a bunch of methods for edge splitting and merging.
1334
1335		4. Machine description.
1336
1337		John's simplification of MLTREE_BASIS.fcond broke a few machine
1338		description things:
1339
1340		rtl-build.{sig,sml} and hppa.mdl fixed.
1341
1342		NOTE: the machine description stuff in the repository is still broken.
1343		Again, I can't put my fixes in because that involves
1344		changes to MLTREE.
1345
1346		----------------------------------------------------------------------
1347		Name: Matthias Blume
1348		Date: 2002/03/20 15:55:00 EST
1349		Tag: blume-20020320-kmp
1350		Description:
1351
1352		Implemented Knuth-Morris-Pratt string matching in PreString and used
1353		it for String.isSubstring, Substring.isSubstring, and
1354		Substring.position.
1355
1356		(Might need some stress-testing.  Simple examples worked fine.)
1357
1358		----------------------------------------------------------------------
1359		Name: Matthias Blume
1360		Date: 2002/03/19 16:37:00 EST
1361		Tag: blume-20020319-witnesses
1362		Description:
1363
1364		Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1365
1366		This implements a generic mechanism for changing constness qualifiers
1367		anywhere within big C types without resorting to outright "casts".
1368		(So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1369		modify the constness at the outermost level.)
1370		The implementation of "convert" is based on the idea of "witness"
1371		values -- values that are not used by the operation but whose types
1372		"testify" to their applicability.  On the implementation side, "convert"
1373		is simply a projection (returning its second curried argument).  With
1374		cross-module inlining, it should not result in any machine code being
1375		generated.
1376
1377		----------------------------------------------------------------------
1378		Name: Matthias Blume
1379		Date: 2002/03/15 16:40:00 EST
1380		Tag: blume-20020315-basis
1381		Description:
1382
1383		Provided (preliminary?) implementations for
1384
1385		{String,Substring}.{concatWith,isSuffix,isSubstring}
1386
1387		and
1388
1389		Substring.full
1390
1391		Those are in the Basis spec but they were missing in SML/NJ.
1392
1393		----------------------------------------------------------------------
1394		Name: Matthias Blume
1395		Date: 2002/03/14 21:30:00 EST
1396		Tag: blume-20020314-controls
1397		Description:
1398
1399		Controls:
1400		---------
1401
1402		1. Factored out the recently-added Controls : CONTROLS stuff and put
1403		it into its own library $/controls-lib.cm.  The source tree for
1404		this is under src/smlnj-lib/Controls.
1405
1406		2. Changed the names of types and functions in this interface, so they
1407		make a bit more "sense":
1408
1409		module -> registry
1410		'a registry -> 'a group
1411
1412		3. The interface now deals in ref cells only.  The getter/setter interface
1413		is (mostly) gone.
1414
1415		4. Added a function that lets one register an already-existing ref cell.
1416
1417		5. Made the corresponding modifications to the rest of the code so that
1418		everything compiles again.
1419
1420		6. Changed the implementation of Controls.MLRISC back to something closer
1421		to the original.  In particular, this module (and therefore MLRISC)
1422		does not depend on Controls.  There now is some link-time code in
1423		int-sys.sml that registers the MLRISC controls with the Controls
1424		module.
1425
1426		CM:
1427		---
1428
1429		* One can now specify the lambda-split aggressiveness in init.cmi.
1430
1431		----------------------------------------------------------------------
1432		Name: Allen Leung
1433		Date: 2002/03/13 17:30:00 EST
1434		Tag: leunga-20020313-x86-fp-unary
1435		Description:
1436
1437		Bug fix for:
1438
1439		> leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1440		> Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1441		> - fun f(x,(y,z)) = Real.~ y;
1442		> [autoloading]
1443		> [autoloading done]
1444		>       fchsl   (%eax), 184(%esp)
1445		> Error: MLRisc bug: X86MCEmitter.emitInstr
1446		>
1447		> uncaught exception Error
1448		>   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1449
1450		The problem was that the code generator did not generate any fp registers
1451		in this case, and the ra didn't know that it needed to run the X86FP phase to
1452		translate the pseudo fp instruction.   This only happened with unary fp
1453		operators in certain situations.
1454
1455		----------------------------------------------------------------------
1456		Name: Matthias Blume
1457		Date: 2002/03/13 14:00:00 EST
1458		Tag: blume-20020313-overload-etc
1459		Description:
1460
1461		1. Added _overload as a synonym for overload for backward compatibility.
1462		(Control.overloadKW must be true for either version to be accepted.)
1463
1464		2. Fixed bug in install script that caused more things to be installed
1465		than what was requested in config/targets.
1466
1467		3. Made CM aware of the (_)overload construct so that autoloading
1468		works.
1469
1470		----------------------------------------------------------------------
1471		Name: Matthias Blume
1472		Date: 2002/03/12 22:03:00 EST
1473		Tag: blume-20020312-url
1474		Description:
1475
1476		Forgot to update BOOT and srcarchiveurl.
1477
1478		----------------------------------------------------------------------
1479		Name: Matthias Blume
1480		Date: 2002/03/12 17:30:00 EST
1481		Tag: blume-20020312-version110392
1482		Description:
1483
1484		Yet another version number bump (because of small changes to the
1485		binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1486
1487		Changes:
1488
1489		The new pid generation scheme described a few weeks ago was overly
1490		complicated.  I implemented a new mechanism that is simpler and
1491		provides a bit more "stability":  Once CM has seen a compilation
1492		unit, it keeps its identity constant (as long as you do not delete
1493		those crucial CM/GUID/* files).  This means that when you change
1494		an interface, compile, then go back to the old interface, and
1495		compile again, you arrive at the original pid.
1496
1497		There now also is a mechanism that instructs CM to use the plain
1498		environment hash as a module's pid (effectively making its GUID
1499		the empty string).  For this, "noguid" must be specified as an
1500		option to the .sml file in question within its .cm file.
1501		This is most useful for code that is being generated by tools such
1502		as ml-nlffigen (because during development programmers tend to
1503		erase the tool's entire output directory tree including CM's cached
1504		GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1505		revert to the old, broken behavior of SML/NJ, but in specific cases
1506		where there is no danger of interface confusion, its use is ok
1507		(I think).
1508
1509		ml-nlffigen by default generates "noguid" annotations.  They can be
1510		turned off by specifying -guid in its command line.
1511
1512		----------------------------------------------------------------------
1513		Name: Lal George
1514		Date: 2002/03/12 12 14:42:36 EST
1515		Tag: george-20020312-frequency-computation
1516		Description:
1517
1518		Integrated jump chaining and static block frequency into the
1519		compiler. More details and numbers later.
1520
1521		----------------------------------------------------------------------
1522		Name: Lal George
1523		Date: 2002/03/11 11 22:38:53 EST
1524		Tag: george-20020311-jump-chain-elim
1525		Description:
1526
1527		Tested the jump chain elimination on all architectures (except the
1528		hppa).  This is on by default right now and is profitable for the
1529		alpha and x86, however, it may not be profitable for the sparc and ppc
1530		when compiling the compiler.
1531
1532		The gc test will typically jump to a label at the end of the cluster,
1533		where there is another jump to an external cluster containing the actual
1534		code to invoke gc. This is to allow factoring of common gc invocation
1535		sequences. That is to say, we generate:
1536
1537		f:
1538		testgc
1539		ja   L1      % jump if above to L1
1540
1541		L1:
1542		jmp L2
1543
1544
1545		After jump chain elimination the 'ja L1' instructions is converted to
1546		'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1547		up being implemented in their long form (if L2 is far away) using:
1548
1549		jbe     L3      % jump if below or equal to L3
1550		jmp     L2
1551		L3:
1552		...
1553
1554
1555		For large compilation units L2  may be far away.
1556
1557
1558		----------------------------------------------------------------------
1559		Name: Matthias Blume
1560		Date: 2002/03/11 13:30:00 EST
1561		Tag: blume-20020311-mltreeeval
1562		Description:
1563
1564		A functor parameter was missing.
1565
1566		----------------------------------------------------------------------
1567		Name: Allen Leung
1568		Date: 2002/03/11 10:30:00 EST
1569		Tag: leunga-20020311-runtime-string0
1570		Description:
1571
1572		The representation of the empty string now points to a
1573		legal null terminated C string instead of unit.  It is now possible
1574		to convert an ML string into C string with InlineT.CharVector.getData.
1575		This compiles into one single machine instruction.
1576
1577		----------------------------------------------------------------------
1578		Name: Allen Leung
1579		Date: 2002/03/10 23:55:00 EST
1580		Tag: leunga-20020310-x86-call
1581		Description:
1582
1583		Added machine generation for CALL instruction (relative displacement mode)
1584
1585		----------------------------------------------------------------------
1586		Name: Matthias Blume
1587		Date: 2002/03/08 16:05:00
1588		Tag: blume-20020308-entrypoints
1589		Description:
1590
1591		Version number bumped to 110.39.1.  NEW BOOTFILES!
1592
1593		Entrypoints: non-zero offset into a code object where execution should begin.
1594
1595		- Added the notion of an entrypoint to CodeObj.
1596		- Added reading/writing of entrypoint info to Binfile.
1597		- Made runtime system bootloader aware of entrypoints.
1598		- Use the address of the label of the first function given to mlriscGen
1599		as the entrypoint.  This address is currently always 0, but it will
1600		not be 0 once we turn on block placement.
1601		- Removed the linkage cluster code (which was The Other Way(tm) of dealing
1602		with entry points) from mlriscGen.
1603
1604		----------------------------------------------------------------------
1605		Name: Allen Leung
1606		Date: 2002/03/07 20:45:00 EST
1607		Tag: leunga-20020307-x86-cmov
1608		Description:
1609
1610		Bug fixes for CMOVcc on x86.
1611
1612		1. Added machine code generation for CMOVcc
1613		2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1614		3. CMOVcc cannot have an immediate operand as argument.
1615
1616		----------------------------------------------------------------------
1617		Name: Matthias Blume
1618		Date: 2002/03/07 16:15:00 EST
1619		Tag: blume-20020307-controls
1620		Description:
1621
1622		This is a very large but mostly boring patch which makes (almost)
1623		every tuneable compiler knob (i.e., pretty much everything under
1624		Control.* plus a few other things) configurable via both the command
1625		line and environment variables in the style CM did its configuration
1626		until now.
1627
1628		Try starting sml with '-h' (or, if you are brave, '-H')
1629
1630		To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1631		implements the underlying generic mechanism.
1632
1633		The interface to some of the existing such facilities has changed somewhat.
1634		For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1635		(The getFoo interface is still there for backward-compatibility, but its
1636		use is deprecated.)
1637
1638		The ml-build script passes -Cxxx=yyy command-line arguments through so
1639		that one can now twiddle the compiler settings when using this "batch"
1640		compiler.
1641
1642		TODO items:
1643
1644		We should go through and throw out all controls that are no longer
1645		connected to anything.  Moreover, we should go through and provide
1646		meaningful (and correct!) documentation strings for those controls
1647		that still are connected.
1648
1649		Currently, multiple calls to Controls.new are accepted (only the first
1650		has any effect).  Eventually we should make sure that every control
1651		is being made (via Controls.new) exactly once.  Future access can then
1652		be done using Controls.acc.
1653
1654		Finally, it would probably be a good idea to use the getter-setter
1655		interface to controls rather than ref cells.  For the time being, both
1656		styles are provided by the Controls module, but getter-setter pairs are
1657		better if thread-safety is of any concern because they can be wrapped.
1658
1659		*****************************************
1660
1661		One bug fix: The function blockPlacement in three of the MLRISC
1662		backpatch files used to be hard-wired to one of two possibilities at
1663		link time (according to the value of the placementFlag).  But (I
1664		think) it should rather sense the flag every time.
1665
1666		*****************************************
1667
1668		Other assorted changes (by other people who did not supply a HISTORY entry):
1669
1670		1. the cross-module inliner now works much better (Monnier)
1671		2. representation of weights, frequencies, and probabilities in MLRISC
1672		changed in preparation of using those for weighted block placement
1673		(Reppy, George)
1674
1675		----------------------------------------------------------------------
1676		Name: Lal George
1677		Date: 2002/03/07 14:44:24 EST 2002
1678		Tag: george-20020307-weighted-block-placement
1679
1680		Tested the weighted block placement optimization on all architectures
1681		(except the hppa) using AMPL to generate the block and edge frequencies.
1682		Changes were required in the machine properties to correctly
1683		categorize trap instructions. There is an MLRISC flag
1684		"weighted-block-placement" that can be used to enable weighted block
1685		placement, but this will be ineffective without block/edge
1686		frequencies (coming soon).
1687
1688
1689		----------------------------------------------------------------------
1690		Name: Lal George
1691		Date: 2002/03/05 17:24:48 EST
1692		Tag: george-20020305-linkage-cluster
1693
1694		In order to support the block placement optimization, a new cluster
1695		is generated as the very first cluster (called the linkage cluster).
1696		It contains a single jump to the 'real' entry point for the compilation
1697		unit. Block placement has no effect on the linkage cluster itself, but
1698		all the other clusters  have full freedom in the manner in which they
1699		reorder blocks or functions.
1700
1701		On the x86 the typical linkage code that is generated is:
1702		----------------------
1703		.align 2
1704		L0:
1705		addl    $L1-L0, 72(%esp)
1706		jmp     L1
1707
1708
1709		.align  2
1710		L1:
1711		----------------------
1712
1713		72(%esp) is the memory location for the stdlink register. This
1714		must contain the address of the CPS function being called. In the
1715		above example, it contains the address of  L0; before
1716		calling L1 (the real entry point for the compilation unit), it
1717		must contain the address for L1, and hence
1718
1719		addl $L1-L0, 72(%esp)
1720
1721		I have tested this on all architectures except the hppa.The increase
1722		in code size is of course negligible
1723
1724		----------------------------------------------------------------------
1725		Name: Allen Leung
1726		Date: 2002/03/03 13:20:00 EST
1727		Tag: leunga-20020303-mlrisc-tools
1728
1729		Added #[ ... ] expressions to mlrisc tools
1730
1731		----------------------------------------------------------------------
1732		Name: Matthias Blume
1733		Date: 2002/02/27 12:29:00 EST
1734		Tag: blume-20020227-cdebug
1735		Description:
1736
1737		- made types in structure C and C_Debug to be equal
1738		- got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1739		- there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1740
1741		----------------------------------------------------------------------
1742		Name: Matthias Blume
1743		Date: 2002/02/26 12:00:00 EST
1744		Tag: blume-20020226-ffi
1745		Description:
1746
1747		1. Fixed a minor bug in CM's "noweb" tool:
1748		If numbering is turned off, then truly don't number (i.e., do not
1749		supply the -L option to noweb).  The previous behavior was to supply
1750		-L'' -- which caused noweb to use the "default" line numbering scheme.
1751		Thanks to Chris Richards for pointing this out (and supplying the fix).
1752
1753		2. Once again, I reworked some aspects of the FFI:
1754
1755		A. The incomplete/complete type business:
1756
1757		- Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
1758		gone!
1759		- ML types representing an incomplete type are now *equal* to
1760		ML types representing their corresponding complete types (just like
1761		in C).  This is still safe because ml-nlffigen will not generate
1762		RTTI for incomplete types, nor will it generate functions that
1763		require access to such RTTI.   But when ML code generated from both
1764		incomplete and complete versions of the C type meet, the ML types
1765		are trivially interoperable.
1766
1767		NOTE:  These changes restore the full generality of the translation
1768		(which was previously lost when I eliminated functorization)!
1769
1770		B. Enum types:
1771
1772		- Structure C now has a type constructor "enum" that is similar to
1773		how the "su" constructor works.  However, "enum" is not a phantom
1774		type because each "T enum" has values (and is isomorphic to
1775		MLRep.Signed.int).
1776		- There are generic access operations for enum objects (using
1777		MLRep.Signed.int).
1778		- ml-nlffigen will generate a structure E_foo for each "enum foo".
1779		* The structure contains the definition of type "mlrep" (the ML-side
1780		representation type of the enum).  Normally, mlrep is the same
1781		as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
1782		then mlrep will be defined as a datatype -- thus facilitating
1783		pattern matching on mlrep values.
1784		("-ec" will be suppressed if there are duplicate values in an
1785		enumeration.)
1786		* Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
1787		will be generated for each C enum constant xxx.
1788		* Conversion functions m2i and i2m convert between mlrep and
1789		MLRep.Signed.int.  (Without "-ec", these functions are identities.)
1790		* Coversion functions c and ml convert between mlrep and "tag enum".
1791		* Access functions (get/set) fetch and store mlrep values.
1792		- By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
1793		enumerations are merged into one single enumeration represented by
1794		structure E_'.
1795
1796		----------------------------------------------------------------------
1797		Name: Allen Leung
1798		Date: 2002/02/25 04:45:00 EST
1799		Tag: leunga-20020225-cps-spill
1800
1801		This is a new implementation of the CPS spill phase.
1802		The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
1803		In case of problems, replace it with the old file spill.sml
1804
1805		The current compiler runs into some serious performance problems when
1806		constructing a large record.  This can happen when we try to compile a
1807		structure with many items.  Even a very simple structure like the following
1808		makes the compiler slow down.
1809
1810		structure Foo = struct
1811		val x_1 = 0w1 : Word32.int
1812		val x_2 = 0w2 : Word32.int
1813		val x_3 = 0w3 : Word32.int
1814		...
1815		val x_N = 0wN : Word32.int
1816		end
1817
1818		The following table shows the compile time, from N=1000 to N=4000,
1819		with the old compiler:
1820
1821		N
1822		1000   CPS 100 spill                           0.04u  0.00s  0.00g
1823		MLRISC ra                               0.06u  0.00s  0.05g
1824		(spills = 0 reloads = 0)
1825		TOTAL                                   0.63u  0.07s  0.21g
1826
1827		1100   CPS 100 spill                           8.25u  0.32s  0.64g
1828		MLRISC ra                               5.68u  0.59s  3.93g
1829		(spills = 0 reloads = 0)
1830		TOTAL                                   14.71u  0.99s  4.81g
1831
1832		1500   CPS 100 spill                           58.55u  2.34s  1.74g
1833		MLRISC ra                               5.54u  0.65s  3.91g
1834		(spills = 543 reloads = 1082)
1835		TOTAL                                   65.40u  3.13s  6.00g
1836
1837		2000   CPS 100 spill                           126.69u  4.84s  3.08g
1838		MLRISC ra                               0.80u  0.10s  0.55g
1839		(spills = 42 reloads = 84)
1840		TOTAL                                   129.42u  5.10s  4.13g
1841
1842		3000   CPS 100 spill                           675.59u  19.03s  11.64g
1843		MLRISC ra                               2.69u  0.27s  1.38g
1844		(spills = 62 reloads = 124)
1845		TOTAL                                   682.48u  19.61s  13.99g
1846
1847		4000   CPS 100 spill                           2362.82u  56.28s  43.60g
1848		MLRISC ra                               4.96u  0.27s  2.72g
1849		(spills = 85 reloads = 170)
1850		TOTAL                                   2375.26u  57.21s  48.00g
1851
1852		As you can see the old cps spill module suffers from some serious
1853		performance problem.  But since I cannot decipher the old code fully,
1854		instead of patching the problems up, I'm reimplementing it
1855		with a different algorithm.  The new code is more modular,
1856		smaller when compiled, and substantially faster
1857		(O(n log n) time and O(n) space).  Timing of the new spill module:
1858
1859		4000  CPS 100 spill                           0.02u  0.00s  0.00g
1860		MLRISC ra                               0.25u  0.02s  0.15g
1861		(spills=1 reloads=3)
1862		TOTAL                                   7.74u  0.34s  1.62g
1863
1864		Implementation details:
1865
1866		As far as I can tell, the purpose of the CPS spill module is to make sure the
1867		number of live variables at any program point (the bandwidth)
1868		does not exceed a certain limit, which is determined by the
1869		size of the spill area.
1870
1871		When the bandwidth is too large, we decrease the register pressure by
1872		packing live variables into spill records.  How we achieve this is
1873		completely different than what we did in the old code.
1874
1875		First, there is something about the MLRiscGen code generator
1876		that we should be aware of:
1877
1878		o MLRiscGen performs code motion!
1879
1880		In particular, it will move floating point computations and
1881		address computations involving only the heap pointer to
1882		their use sites (if there is only a single use).
1883		What this means is that if we have a CPS record construction
1884		statement
1885
1886		RECORD(k,vl,w,e)
1887
1888		we should never count the new record address w as live if w
1889		has only one use (which is often the case).
1890
1891		We should do something similar to floating point, but the transformation
1892		there is much more complex, so I won't deal with that.
1893
1894		Secondly, there are now two new cps primops at our disposal:
1895
1896		1. rawrecord of record_kind option
1897		This pure operator allocates some uninitialized storage from the heap.
1898		There are two forms:
1899
1900		rawrecord NONE [INT n]  allocates a tagless record of length n
1901		rawrecord (SOME rk) [INT n] allocates a tagged record of length n
1902		and initializes the tag.
1903
1904		2. rawupdate of cty
1905		rawupdate cty (v,i,x)
1906		Assigns to x to the ith component of record v.
1907		The storelist is not updated.
1908
1909		We use these new primops for both spilling and increment record construction.
1910
1911		1. Spilling.
1912
1913		This is implemented with a linear scan algorithm (but generalized
1914		to trees).  The algorithm will create a single spill record at the
1915		beginning of the cps function and use rawupdate to spill to it,
1916		and SELECT or SELp to reload from it.  So both spills and reloads
1917		are fine-grain operations.  In contrast, in the old algorithm
1918		"spills" have to be bundled together in records.
1919
1920		Ideally, we should sink the spill record construction to where
1921		it is needed.  We can even split the spill record into multiple ones
1922		at the places where they are needed.  But CPS is not a good
1923		representation for global code motion, so I'll keep it simple and
1924		am not attempting this.
1925
1926		2. Incremental record construction (aka record splitting).
1927
1928		Long records with many component values which are simulatenously live
1929		(recall that single use record addresses are not considered to
1930		be live) are constructed with rawrecord and rawupdate.
1931		We allocate space on the heap with rawrecord first, then gradually
1932		fill it in with rawupdate.  This is the technique suggested to me
1933		by Matthias.
1934
1935		Some restrictions on when this is applicable:
1936		1. It is not a VECTOR record.  The code generator currently does not handle
1937		this case. VECTOR record uses double indirection like arrays.
1938		2. All the record component values are defined in the same "basic block"
1939		as the record constructor.  This is to prevent speculative
1940		record construction.
1941
1942		----------------------------------------------------------------------
1943		Name: Allen Leung
1944		Date: 2002/02/22 01:02:00 EST
1945		Tag: leunga-20020222-mlrisc-tools
1946
1947		Minor bug fixes in the parser and rewriter
1948
1949		----------------------------------------------------------------------
1950	Name: Allen Leung	Name: Allen Leung
1951	Date: 2002/02/21 20:20:00 EST	Date: 2002/02/21 20:20:00 EST
1952	Tag: leunga-20020221-peephole	Tag: leunga-20020221-peephole

---

Colored Diff Long Colored Diff Unidiff Context Diff Side by Side

Legend: Removed from v.1085   changed lines   Added in v.1455

---

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