Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 1066, Thu Feb 14 16:50:02 2002 UTC	revision 1399, Tue Sep 30 21:10:05 2003 UTC
#	Line 13	Line 13
13	Description:	Description:
14
15	----------------------------------------------------------------------	----------------------------------------------------------------------
16		Name: Matthias Blume (blume (at) tti - c (dot) org)
17		Date: 2003/09/30 16:10:00 CDT
18		Tag: blume-20030930-primio-bat
19		Description:
20
21		1. Added openVector, nullRd, and nullWr to PRIM_IO.
22		2. Improved .bat files (for Win32 port) to make things work under Win95.
23		(thanks to Aaron S. Hawley for this one)
24
25		----------------------------------------------------------------------
26		Name: Matthias Blume (blume (at) tti - c (dot) org)
27		Date: 2003/09/26 16:05:00 CDT
28		Tag: blume-20030926-wrappriv
29		Description:
30
31		Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
32
33		----------------------------------------------------------------------
34		Name: Matthias Blume (blume (at) tti - c (dot) org)
35		Date: 2003/09/26 15:00:00 CDT
36		Tag: blume-20030926-110_43_3
37		Description:
38
39		- additional cleanup
40		- version number bump, NEW BOOTFILES
41
42		----------------------------------------------------------------------
43		Name: Matthias Blume (blume (at) tti - c (dot) org)
44		Date: 2003/09/26 12:00:00 CDT
45		Tag: blume-20030926-ppautoload
46		Description:
47
48		I modified the read-eval-print loop so that the autoloader gets
49		invoked whenever the prettyprinter tries to look up a symbol that
50		is not currently defined in the toplevel environment but which
51		appears in CM's autoload registry.  As a result, we see far fewer of
52		those ?.Foo.Bar.xxx names in the prettyprinter's output.
53
54		In addition to this I tried to clean up some pieces of the Basis
55		implementation (e.g., Socket, Word8Array) in order to prevent other
56		instances of these ?.Foo.Bar.xxx names from being printed.
57
58		The mechanism that picks names for types still needs some work, though.
59		(Right now it seems that if there is a type A.t which is defined to
60		be B.u, but B is unavailable at toplevel, then A.t gets printed as
61		"?.B.u" although the perhaps more sensible solution would be to use
62		"A.t" in this case.  In other words, the prettyprinter should follow
63		a chain of DEFtycs not farther than there are corresponding toplevel
64		names in the current environment.)
65
66		----------------------------------------------------------------------
67		Name: Matthias Blume (blume (at) tti - c (dot) org)
68		Date: 2003/09/24 16:31:00 CDT
69		Tag: blume-20030924-installer
70		Description:
71
72		Another installer tweak:  All the ML code for the installer is now
73		compiled during CMB.make and put into a little library called
74		$smlnj/installer.cm.  The installation then simply invokes
75
76		sml -m $smlnj/installer.cm
77
78		and everything happens automagically.
79
80		Win32:  ML code senses value of environment variable SMLNJ_HOME.
81		Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
82		and BINDIR.
83
84		The new scheme guarantees that the ML code responsible for the installation
85		is in sync with the APIs of the main system.  Also, the installer is
86		somewhat faster because the installer script is precompiled.
87
88		----------------------------------------------------------------------
89		Name: Matthias Blume (blume (at) tti - c (dot) org)
90		Date: 2003/09/24 15:35:00 CDT
91		Tag: blume-20030924-synsock
92		Description:
93
94		Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
95		but excludes all non-blocking operations.  Defined SOCKET (in Basis)
96		and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
97		implementations of non-blocking operations from CML's Socket
98		structure.
99
100		----------------------------------------------------------------------
101		Name: Matthias Blume (blume (at) tti - c (dot) org)
102		Date: 2003/09/24 15:10:05 CDT
103		Tag: blume-20030924-sockets
104		Description:
105
106		1. Fixed SOCKET API and implementation to match Basis spec.
107		This required changing the internal representation of sockets to one
108		that remembers (for each socket file descriptor) whether it is currently
109		blocking or non-blocking.  This state is maintained lazily (i.e., a system
110		call is made only if the state actually needs to change).
111
112		2. OS-specific details of sockets were moved into separate files, thus
113		making it possible to unify the bulk of the socket implementations
114		between Unix and Win32.
115
116		3. CML's socket API changed accordingly.
117		(Note that we need to remove non-blocking functions from this API
118		since they are redundant in the case of CML!)
119
120		4. CML's socket implementation now makes use of non-blocking functions
121		provided by Basis, thus removing all OS-dependent code from this part
122		of CML.
123
124		5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
125
126		----------------------------------------------------------------------
127		Name: Matthias Blume (blume (at) tti - c (dot) org)
128		Date: 2003/09/22 12:10:00 CDT
129		Tag: blume-20030922-110_43_2
130		Description:
131
132		Made a new interim version and bootfiles for developer's bootstrapping
133		convenience.
134
135		110.43.2 -- NEW BOOTFILES
136
137		----------------------------------------------------------------------
138		Name: Matthias Blume (blume (at) tti - c (dot) org)
139		Date: 2003/09/19 15:55:00 CDT
140		Tag: blume-20030919-cmdir
141		Description:
142
143		1. new-install.sh -> install.sh
144		2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
145		3. tweaked installer so that another name instead of .cm can be chosen
146		at install time (by setting the CM_DIR_ARC environment variable
147		during installation); once installation is complete, the name is
148		fixed
149
150		----------------------------------------------------------------------
151		Name: Matthias Blume (blume (at) tti - c (dot) org)
152		Date: 2003/09/18 16:00:00 CDT
153		Tag: blume-20030918-110_43_1
154		Description:
155
156		Made a new interim version and bootfiles for developer's bootstrapping
157		convenience.
158
159		110.43.1 -- NEW BOOTFILES
160
161		----------------------------------------------------------------------
162		Name: Matthias Blume (blume (at) tti - c (dot) org)
163		Date: 2003/09/18 15:20:00 CDT
164		Tag: blume-20030918-misc
165		Description:
166
167		1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
168		this seems to be controversial at the moment) and used those in
169		Posix.ProcEnv.times.
170
171		2. Added Time.{from,to}Nanoseconds to Time.
172
173		3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
174		For example, fromLargeInt never needs to look at more than 3 "big
175		digits" to get its 53 bits of precision.
176
177		----------------------------------------------------------------------
178		Name: Matthias Blume (blume (at) tti - c (dot) org)
179		Date: 2003/09/17 16:30:00 CDT
180		Tag: blume-20030917-real32-slices
181		Description:
182
183		Added an entry to the primitive environment
184		(compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
185		added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
186
187		Removed some of the "magic" constants in real64.sml and replaced them
188		with code that generates these values from their corresponding
189		integer counterparts.
190
191		Made all(?) the slice-related changes to the Basis and made everything
192		compile again...
193
194		----------------------------------------------------------------------
195		Name: Matthias Blume (blume (at) tti - c (dot) org)
196		Date: 2003/09/15 17:45:00 CDT
197		Tag: blume-20030915-rbase
198		Description:
199
200		Fixed bug in Real.fromLargeInt.
201
202		----------------------------------------------------------------------
203		Name: Matthias Blume (blume (at) tti - c (dot) org)
204		Date: 2003/09/13 18:11:00 CDT
205		Tag: blume-20030913-libinstall
206		Description:
207
208		Minor bugfix in config/libinstall (set anchor with path to
209		standalone tool after installing it, otherwise libraries that
210		need ml-lex or ml-yacc won't compile the first time the installer
211		runs).
212
213		----------------------------------------------------------------------
214		Name: Matthias Blume (blume (at) tti - c (dot) org)
215		Date: 2003/09/12 11:45:00 CDT
216		Tag: blume-20030912-various
217		Description:
218
219		- fixed bug in Real.toLargeInt
220		- fixed bug in Posix.ProcEnv.times
221		- changed inputLine functions to return an option
222		- minor installer improvements / bugfixes
223		- changed default @SMLalloc parameter for x86/celeron to 64k
224
225		----------------------------------------------------------------------
226		Name: Matthias Blume (blume (at) tti - c (dot) org)
227		Date: 2003/09/09 22:00:00 CDT
228		Tag: Release_110_43
229		Description:
230
231		New working release 110.43.  New bootfiles.
232
233		----------------------------------------------------------------------
234		Name: Matthias Blume (blume (at) tti - c (dot) org)
235		Date: 2003/09/09 19:20:00 CDT
236		Tag: blume-20030909-installer
237		Description:
238
239		Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
240		Modified config/install.bat to take advantage of it.  Also modified
241		config/install.sh (and called it config/new-install.sh) to take advantage
242		of it on Unix systems.  (The SML code is (supposed to be) platform-
243		independent.)
244
245		The installer can now install everything under Win32
246		as well as under *nix as long as it compiles.
247
248		Other changes:
249
250		- made CML compile again under Win32
251		- made eXene compile under Win32 (by providing a fake structure UnixSock
252		and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
253		- fixed a bug in nowhere: it assumed that type OS.Process.status is the
254		same as type int;  under Win32 it isn't
255		- fixed some slice-related problems in the win32-specific parts of CML
256		- added a functor argument "sameVol" to os-path-fn.sml in the Basis
257		(under Win32, the volume name is case-insensitive, and the
258		OS.Path code compares volume names for equality)
259
260		----------------------------------------------------------------------
261		Name: Matthias Blume (blume (at) tti - c (dot) org)
262		Date: 2003/09/08 11:55:00 CDT
263		Tag: blume-20030908-fullpath
264		Description:
265
266		Made Win32 version of OS.FileSys.fullPath return current directory
267		when given an empty string.  This is what the spec says, and incidentally,
268		CM depends on it.  (CM otherwise goes into an infinite loop in certain
269		cases when presented with the name of a non-existing .cm file.)
270
271		----------------------------------------------------------------------
272		Name: Matthias Blume (blume (at) tti - c (dot) org)
273		Date: 2003/09/04 16:30:00 CDT
274		Tag: blume-20030905-slices-etc
275		Description:
276
277		1. Changed interface to vectors and arrays in Basis to match
278		(draft) Basis spec.
279		2. Added signatures and implementations of slices according to
280		Basis spec.
281		3. Edited source code throughout the system to make it compile again
282		under 1. and 2.  (In some cases code had to be added to have it
283		match the new signatures.)
284		4. MLRISC should be backward-compatible: the copies of the originals
285		of files that needed to change under 3. were retained, the .cm files
286		check the compiler version number and use old versions when
287		appropriate.
288		5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
289		dirstream -> string option (in accordance with Basis spec).
290		6. When generating code that counts lines, ml-lex used function
291		CharVector.foldli, taking advantage of its old interface.
292		This has been replaced with the corresponding code from
293		CharVectorSlice.  (html-lex must be re-lexed!)
294		7. BitArray in smlnj-lib/Util has been extended/modified to match the
295		new MONO_ARRAY signature.  (Do we need BitArraySlice?)
296		8. Removed temporary additions (fromInternal, toInternal) from the
297		(now obsolete) IntInf in smlnj-lib/Util.
298		9. Cleaned up structure Byte.
299		10. Added localOffset, scan, and fromString to Date (according to spec).
300		Cleaned/corrected implementation of Date.
301		(Still need to check for correctness; implement better canonicalizeDate.)
302		11. Added "scan" to signature IEEE_REAL.
303		12. Some improvements to IntInf [in particular: efficiency-hack for
304		mod and rem when second operand is 2 (for parity checks).]
305		13. Changed representation of type Time.time, using a single IntInf.int
306		value counting microseconds.  This considerably simplified the
307		implementation of structure Time.  We now support negative time
308		values; scan and fromString handle signs.
309		14. Functor PrimIO now takes two additional arguments (VectorSlice and
310		ArraySlice).
311
312		----------------------------------------------------------------------
313		Name: Matthias Blume (blume (at) tti - c (dot) org)
314		Date: 2003/08/28 17:00:00 CDT
315		Tag: blume-20030828-intinf
316		Description:
317
318		This is a major update which comes with a version number bump
319		(110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
320		and an implementation of IntInf in the Basis.
321
322		There are a fairly large number of related changes and updates throughout
323		the system:
324
325		Basis:
326		- Implemented IntInf.
327		- Made LargeInt a projection of IntInf (by filtering through INTEGER).
328		- Added some missing Real64 operations, most notably Real.toLargeInt.
329		- Added FixedInt as a synonym for Int32.
330
331		compiler:
332		* Added support for a built-in intinf type.
333		- literals
334		- pattern matching
335		- conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
336		- overloading on literals and operations
337
338		This required adding a primitive type intinf, some additional
339		primops, and implementations for several non-trivial intinf
340		operations in Core.  (The intinf type is completely abstract
341		to the compiler; all operations get delegated back to the Core.)
342
343		* Intinf equality is handled by polyequal.  However, the compiler
344		does not print its usual warning in this case (since polyequal
345		is the right thing to do there).
346
347		* Improved the organization of structure InlineT.
348
349		* A word about conversion primops:
350		If conversions involving intinf do not cancel out during
351		CPS contract, then the compiler must insert calls to Core functions.
352		Since all core access must be resolved already during the FLINT
353		translate phase, it would be too late a the time of CPS contract
354		to add new Core calls.  For this reason, conversion primops
355		for intinf carry two arguments:  1. the numeric argument that
356		they are supposed to convert, and 2. the Core function that
357		can help with this conversion if necessary.  If CPS contract
358		eliminates a primop, then the associated Core function becomes
359		dead and goes away.  Intinf conversion primops that do not get
360		eliminated by CPS contract get rewritten into calls of their
361		core functions by a separate, new phase.
362
363		interactive system:
364		- Control.Print.intinfDepth controls max length of intinf constants
365		being printed. (Analogous to Control.Print.stringDepth.)
366		- Cleanup in printutil and pputil: got rid of unused stuff and
367		duplicates; replaced some of the code with code that makes better
368		use of library functionality.
369
370		CM:
371		Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
372		are no longer silent.
373
374		CKIT:
375		Fixed mismatched uses of Int32 and LargeInt.  I always decided
376		in favor of LargeInt -- which is now the same as IntInf.
377		CKIT-knowledgable people should check whether this is what's
378		intended and otherwise change things back to using Int32 or
379		FixedInt.
380
381		Throughout the code:
382		Started using IntInf.int literals and built-in operations
383		(e.g., comparison with 0) where this seems appropriate.
384
385
386		----------------------------------------------------------------------
387		Name: Dave MacQueen (dbm@cs.uchicago.edu)
388		Date: 2003/08/13 11:36:00 CDT
389		Tag: dbm-20030813-mcz-merge1
390		Description:
391
392		Merging changes from the mcz-branch development branch into trunk.
393		These changes involve replacement of the emulated old prettyprinter
394		interface with direct use of the SML/NJ Lib PP library, and fixing
395		of a couple of bugs (895, 1186) relating to error messages.  A new
396		prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
397		has been added.
398
399		----------------------------------------------------------------------
400		Name: Matthias Blume (blume (at) tti - c (dot) org)
401		Date: 2003/08/11 15:45:00 CDT
402		Tag: blume-20030811-windows
403		Description:
404
405		Version number bumped to 110.42.9.  NEW BOOTFILES!!!
406
407		http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
408
409		This patch restores SML/NJ's ability to run under win32.  There are a
410		number of changes, including fixes for several bugs that had gone
411		unnoticed until now:
412
413		- uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
414		trying to run the win32 version from within cygwin.)
415
416		- There are a number of simple .bat scripts that substitute for their
417		corresponding Unix shell-scripts.  (See below.)
418
419		- The internals of ml-build have been modified slightly.  The main
420		difference is that instead of calling ".link-sml" (or link-sml.bat)
421		using OS.Process.system, the ML process delegates this task back
422		to the script.  Otherwise problems arise in mixed environments such
423		as Cygwin where scripts look and work like Unix scripts, but
424		where OS.Process.system cannot run them.
425
426		- In CM, the srcpath pickler used native pathname syntax -- which
427		is incorrect in the case of cross-compilation.  The new pickle format
428		is independent of platform-specific naming conventions.
429
430		- Path configuration files (such as lib/pathconfig) can now choose
431		between native and standard syntax.  Placing a line of the form
432
433		standard!
434
435		into the file causes all subsequent paths to be interpreted using
436		CM standard pathname syntax (= Unix conventions); a line
437
438		native!
439
440		switches back to native style.  This was needed so that
441		path config files can be written portably, see src/system/pathconfig.
442
443		- Runtime system:
444
445		- win32-filesys.c:  get_file_time and set_file_time now
446		access modification time, not creation time.
447
448		- I/O code made aware of new array representation.
449
450		- Bug fixes in X86.prim.masm.
451
452		- src/system/makeml made aware of win32. (For use under cygwin
453		and other Unix-environments for windows.)
454
455		- In Basis, fixed off-by-one error in win32-io.sml (function vecF)
456		which caused BinIO.inputAll to fail consistently.
457
458		.bat scripts:
459
460		Windows .bat scripts assume that SMLNJ_HOME is defined.
461
462		- sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
463		applications (sml, ml-yacc, ml-lex).
464		- ml-build.bat: analogous to ml-build.
465		- config\install.bat:  Analogous to config/install.sh.  This requires
466		that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
467		(nmake etc. must be on the path, and vcvars32 must have been run.)
468		Moreover, sources for ml-lex and ml-yacc need to exist under src,
469		and the bootfile hierarchy must have been unpacked under
470		sml.boot.x86-win32.
471		The script is very primitive and does a poor job at error checking.
472		It only installs the base system, ml-lex, and ml-yacc.  No other
473		libraries are being installed (i.e., you get only those that
474		are part of the compiler.)
475		- link-sml.bat:  analogous to .link-sml, but not currently used
476
477		Unrelated bug fixes:
478
479		- ml-nlffigen now exports structures ST_* corresponding to incomplete
480		types.
481		- Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
482		otherwise.)
483
484		----------------------------------------------------------------------
485		Name: Dave MacQueen (macqueen@cs.uchicago.edu)
486		Date: 2003/06/17
487		Tag: macqueen-20030617-bug895
488		Description:
489
490		Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
491		Tag will be used for new development branch (mcz-branch) for
492		use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
493
494		----------------------------------------------------------------------
495		Name: Matthias Blume (blume (at) tti - c (dot) org)
496		Date: 2003/05/27 16:55:00 CDT
497		Tag: blume-20030527-polyeq
498		Description:
499
500		Tried to eliminated most cases of polymorphic equality.
501
502		----------------------------------------------------------------------
503		Name: Matthias Blume (blume (at) tti - c (dot) org)
504		Date: 2003/05/21 17:45:00 CDT
505		Tag: blume-20030517-complete
506		Description:
507
508		Two changes:
509
510		1. Added a flag for controlling whether non-exhaustive bindings will
511		be treated as errors (default is false).
512		2. Cleaned up the *entire* source tree so that CMB.make goes through
513		without a single non-exhaustive match- or bind warning.
514
515		----------------------------------------------------------------------
516		Name: Matthias Blume (blume (at) tti - c (dot) org)
517		Date: 2003/05/17 10:20:00 CDT
518		Tag: blume-20030517-absyn
519		Description:
520
521		1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
522
523		This mainly affects the quality of error messages.  However, some
524		of the code is now more straightforward than before. (Treatment of
525		the above four constructs in translate.sml is much simpler than
526		the "macro-expansion" that was going on before.  Plus, the mach-
527		compiler no longer gets invoked just to be able to compile an
528		if-expression.)
529
530		2. The ErrorMsg.Error exception is now caught and absorbed by the
531		interactive loop.
532
533		----------------------------------------------------------------------
534		Name: Allen Leung
535		Date: 2003/05/16 13:05:00 CDT
536		Tag: leunga-20030516-cygwin-runtime
537		Description:
538
539		Ported the runtime system to cygwin, which uses the unix
540		x86-unix bin files.  Missing/buggy features:
541
542		o getnetbyname, getnetbyaddr: these functions seem to be missing in
543		the Cygwin library.
544		o Ctrl-C handling may be flaky.
545		o Windows system calls and Windows I/O are not supported.
546
547		A new set of binfiles is located at:
548
549		http://www.dorsai.org/~leunga/boot.x86-unix.tgz
550
551		This is only needed for bootstrapping the cygwin version of smlnj.
552		Other x86 versions can use the existing binfiles.
553
554		----------------------------------------------------------------------
555		Name: Matthias Blume
556		Date: 2003/04/08 15:42:00 CDT
557		Tag: blume-20030408-listpair
558		Description:
559
560		1. Added a target 'mlrisc' to installer.
561
562		2. Added missing elements to structure ListPair.
563
564		----------------------------------------------------------------------
565		Name: Allen Leung
566		Date: 2003/01/07 10:40:00 EST
567		Tag: leunga-20030107-int-rem
568		Description:
569
570		Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
571		The arguments to the SUBL instruction were swapped.
572
573		----------------------------------------------------------------------
574		Name: Matthias Blume
575		Date: 2002/12/12 16:25:00 EST
576		Tag: blume-20021212-risc-ra
577		Description:
578
579		Fixed a serious bug in the rewrite code for FP spilling/reloading that
580		sent the RA into an infinite loop when floating point registers get
581		spilled.  (Because of this bug, e.g., nucleic stopped compiling between
582		110.37 and 110.38.)
583		There was another set of potential problems related to the handling of
584		MLRISC annotations (but those did not yet cause real problems, apparently).
585
586		----------------------------------------------------------------------
587		Name: Matthias Blume
588		Date: 2002/12/06 22:40:00 EST
589		Tag: blume-20021206-cm-fileid
590		Description:
591
592		Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
593		This fixes the problem of CM getting confused by files that suddenly
594		change their identity (e.g., by getting unlinked and recreated by some
595		text editor such as vi).  There might be a better/cheaper/cleaner way
596		of doing this, but for now this will have to do.
597
598		----------------------------------------------------------------------
599		Name: Matthias Blume
600		Date: 2002/10/28 09:50:00 EST
601		Tag: blume-20021028-typecheck
602		Description:
603
604		Exported structure Typecheck from $smlnj/viscomp/core.cm.
605
606		----------------------------------------------------------------------
607		Name: Matthias Blume
608		Date: 2002/10/17 09:10:00 EDT
609		Tag: Release_110_42
610		Description:
611
612		In good old tradition, there has been a slight hiccup so that we have
613		to patch 110.42 after the fact.   The old release tag has been replaced
614		(see below).
615
616		The change solves a problem with two competing approaches the
617		configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
618		each other's way.
619
620		This change only affects the runtime system code and the installer script.
621		(No new bootfiles.)
622
623		----------------------------------------------------------------------
624		Name: Matthias Blume
625		Date: 2002/10/16 12:00:00 EDT
626		Tag: Release_110_42_removed
627		Description:
628
629		New working release.  New bootfiles.
630
631		----------------------------------------------------------------------
632		Name: Matthias Blume
633		Date: 2002/10/10 13:10:00 EDT
634		Tag: blume-20021010-ppc-divs
635		Description:
636
637		The mltree operator DIVS must be implemented with an overflow check on
638		the PPC because the hardware indicates divide-by-zero using "overflow" as
639		well.
640
641		----------------------------------------------------------------------
642		Name: Matthias Blume
643		Date: 2002/07/23 11:45:00
644		Tag: blume-20020723-smlnj-home
645		Description:
646
647		Sml now senses the SMLNJ_HOME environment variable.  If this is set,
648		then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
649		CM_PATHCONFIG is also set), the path configuration file is assumed
650		to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
651		the entire tree to some other place and everything will "just work".
652
653		(Companion commands such as ml-build and ml-makedepend also sense this
654		variable.)
655
656		----------------------------------------------------------------------
657		Name: Matthias Blume
658		Date: 2002/07/12 21:19:00 EDT
659		Tag: blume-20020712-liveness
660		Description:
661
662		Exported two useful "step" functions from liveness module (MLRISC).
663
664		----------------------------------------------------------------------
665		Name: Matthias Blume
666		Date: 2002/07/05 16:00 EDT
667		Tag: Release_110_41
668		Description:
669
670		New working release.  New bootfiles.
671
672		----------------------------------------------------------------------
673		Name: Matthias Blume
674		Date: 2002/07/05 10:25:00 EDT
675		Tag: blume-20020705-btimp
676		Description:
677
678		Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
679		other clients can set up backtracing support.
680
681		----------------------------------------------------------------------
682		Name: Matthias Blume
683		Date: 2002/06/25 17:23:00 EDT
684		Tag: blume-20020625-fpmax
685		Description:
686
687		Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
688		case.  (The sense of the isNaN test was reversed -- which made min and
689		max always return their first argument.)
690
691		----------------------------------------------------------------------
692		Name: Matthias Blume
693		Date: 2002/06/11
694		Tag: blume-20020611-unixpath
695		Description:
696
697		Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
698
699		----------------------------------------------------------------------
700		Name: Matthias Blume
701		Date: 2002/06/10 16:35:00 EDT
702		Tag: blume-20020610-ieeereal
703		Description:
704
705		I back-ported my implementation of IEEEReal.fromString from the
706		idlbasis-devel branch so that we can test it.
707
708		Another small change is that ppDec tries to give more information
709		than just "<sig>" in the case of functors.  However, this code is
710		broken in some mysterious way if the functor's body's signature
711		has not been declared by ascription but gets inferred from the
712		implementation.  This needs fixing...
713
714		----------------------------------------------------------------------
715		Name: Matthias Blume
716		Date: 2002/05/31
717		Tag: blume-20020531-btrace-mode
718		Description:
719
720		Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
721		the wayside when I switched over to using Controls everywhere.)
722
723		----------------------------------------------------------------------
724		Name: Lal George
725		Date: 2002/05/23 12:21:40 EDT
726		Tag: george-20020523-visual-labels
727		Description:
728
729		Labels are now displayed in the graphical output to make
730		the fall-through and target blocks obvious.
731
732		----------------------------------------------------------------------
733		Name: Matthias Blume
734		Date: 2002/05/22 11:03:00 EDT
735		Tag: blume-20020522-shrink
736		Description:
737
738		John tweaked yesterday's fix for 1131 to handle an out-of-memory
739		situation that comes up when allocating huge arrays.
740
741		----------------------------------------------------------------------
742		Name: Matthias Blume
743		Date: 2002/05/21 16:00:00 EDT
744		Tag: Release_110_40
745		Description:
746
747		New working release (110.40).  New bootfiles.
748
749		[Also: John Reppy fixed GC bug 1131.]
750
751		----------------------------------------------------------------------
752		Name: Matthias Blume
753		Date: 2002/05/21 12:35:00 EDT
754		Tag: blume-20020521-cmdoc
755		Description:
756
757		CM documentation update.
758
759		----------------------------------------------------------------------
760		Name: Matthias Blume
761		Date: 2002/05/21 10:55:00 EDT
762		Tag: blume-20020521-misc
763		Description:
764
765		- John tweaked runtime to be silent on heap export (except when
766		GC messages are on).
767		- I added a few more things (cross-compiling versions of CMB) to
768		config/preloads (as suggestions).
769
770		----------------------------------------------------------------------
771		Name: Matthias Blume
772		Date: 2002/05/20 22:25:00 EDT
773		Tag: blume-20020520-controls
774		Description:
775
776		- Added ControlUtil structure to control-lib.cm.
777		- Use it throughout.
778		- Used Controls facility to define MLRISC controls (as opposed to
779		registering MLRISC control ref cells with Controls after the fact)
780		- Fixed messed-up controls priorities.
781
782		* Removed again all the stuff from config/preloads that one wouldn't
783		be able to preload at the time the initial heap image is built.
784		(Many libraries, e.g., CML, do not exist yet at this time.  The
785		only libraries that can be preloaded via config/preloads are those
786		that come bundled with the bootfiles.)
787
788		----------------------------------------------------------------------
789		Name: Matthias Blume
790		Date: 2002/05/20 10:59:00 EDT
791		Tag: blume-20020520-preloads
792		Description:
793
794		Added a lot of commented-out suggestions for things to be included
795		in config/preloads.
796
797		----------------------------------------------------------------------
798		Name: Allen Leung
799		Date: 2002/05/18 14:20:00 EDT
800		Tag: leunga-20020518-mdl
801		Description:
802
803		o Made the mdl tool stuff compile and run again.
804		o I've disabled all the stuff that depends on RTL specifications; they
805		are all badly broken anyway.
806
807		----------------------------------------------------------------------
808		Name: Matthias Blume
809		Date: 2002/05/17 16:49:00 EDT
810		Tag: blume-20020517-controls
811		Description:
812
813		0. John Reppy made several modifications to the SML/NJ library.
814		In particular, there is a shiny new controls-lib.cm.
815
816		1. Pushed new controls interface through compiler so that everything
817		compiles again.
818
819		2. Added FormatComb and FORMAT_COMB to the CML version of the
820		SML/NJ library (so that CML compiles again).
821
822		3. Modified init scripts because XXX_DEFAULT environment variables
823		are no longer with us.  (Boot-time initialization is now done
824		using the same environment variables that are also used for
825		startup-time initialization of controls.)
826
827		----------------------------------------------------------------------
828		Name: Lal George
829		Date: 2002/05/15 09:20:10 EDT
830		Tag: george-20020515-pseudo-op-decls
831		Description:
832
833		All pseudo-ops emitted before the first segment declaration
834		such as TEXT, DATA, and BSS directives are assumed to be global
835		declarations and are emitted first in the assembly file. This is
836		useful in a number of situations where one has pseudo-ops that are not
837		specific to any segment, and also works around the constraint that one
838		cannot have client pseudo-ops in the TEXT segment.
839
840		Because no segment is associated with these declarations it is
841		an error to allocate any space or objects before the first segment
842		directive and an exception will be raised. However, we cannot make
843		this check for client pseudo-ops.
844
845		These top level declarations are a field in the CFG graph_info.
846		In theory you can continue to add to this field after the CFG has been
847		built -- provided you know what you are doing;-)
848
849		----------------------------------------------------------------------
850		Name: Matthias Blume
851		Date: 2002/05/13 16:40:00 EDT
852		Tag: blume-20020513-pp-etc
853		Description:
854
855		A few minor bugfixes:
856
857		- Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
858		(Bogus printouts for redefined bindings still occur.  Compiler
859		bug should no longer occur now.  We need to redo the prettyprinter
860		from scratch.)
861
862		- CM pathname printer now also adds escape sequences for ( and )
863
864		- commend and docu fixes for ml-nlffi
865
866		----------------------------------------------------------------------
867		Name: Matthias Blume
868		Date: 2002/05/10 16:40:00 EDT
869		Tag: blume-20020510-erg-textio
870		Description:
871
872		Applied the following bugfix provided by Emden Gansner:
873
874		Output is corrupted when outputSubstr is used rather than output.
875
876		The problem occurs when a substring
877
878		ss = (s, dataStart, dataLen)
879
880		where dataStart > 0, fills a stream buffer with avail bytes left.
881		avail bytes of s, starting at index dataStart, are copied into the
882		buffer, the buffer is flushed, and then the remaining dataLen-avail
883		bytes of ss are copied into the beginning of the buffer. Instead of
884		starting this copy at index dataStart+avail in s, the current code
885		starts the copy at index avail.
886
887		Fix:
888		In text-io-fn.sml, change line 695 from
889		val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
890		to
891		val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
892
893		----------------------------------------------------------------------
894		Name: Matthias Blume
895		Date: 2002/04/12 13:55:00 EDT
896		Tag: blume-20020412-assyntax
897		Description:
898
899		1. Grabbed newer assyntax.h from the XFree86 project.
900		2. Fiddled with how to compile X86.prim.asm without warnings.
901		3. (Very) Minor cleanup in CM.
902
903		----------------------------------------------------------------------
904		Name: Matthias Blume
905		Date: 2002/04/01 (no joke!) 17:07:00 EST
906		Tag: blume-20020401-x86div
907		Description:
908
909		Added full support for div/mod/rem/quot on the x86, using the machine
910		instruction's two results (without clumsily recomputing the remainder)
911		directly where appropriate.
912
913		Some more extensive power-of-two support was added to the x86 instruction
914		selector (avoiding expensive divs, mods, and muls where they can be
915		replaced with cheaper shifts and masks).  However, this sort of thing
916		ought to be done earlier, e.g., within the CPS optimizer so that
917		all architectures benefit from it.
918
919		The compiler compiles to a fixed point, but changes might be somewhat
920		fragile nevertheless.  Please, report any strange things that you might
921		see wrt. div/mod/quot/rem...
922
923		----------------------------------------------------------------------
924		Name: Matthias Blume
925		Date: 2002/03/29 17:22:00
926		Tag: blume-20020329-div
927		Description:
928
929		Fixed my broken div/mod logic.  Unfortunately, this means that the
930		inline code for div/mod now has one more comparison than before.
931		Fast paths (quotient > 0 or remainder = 0) are not affected, though.
932		The problem was with quotient = 0, because that alone does not tell
933		us which way the rounding went.  One then has to look at whether
934		remainder and divisor have the same sign...  :(
935
936		Anyway, I replaced the bootfiles with fresh ones...
937
938		----------------------------------------------------------------------
939		Name: Matthias Blume
940		Date: 2002/03/29 14:10:00 EST
941		Tag: blume-20020329-inlprims
942		Description:
943
944		NEW BOOTFILES!!!    Version number bumped to 110.39.3.
945
946		Primops have changed. This means that the bin/boot-file formats have
947		changed as well.
948
949		To make sure that there is no confusion, I made a new version.
950
951
952		CHANGES:
953
954		* removed REMT from mltree (remainder should never overflow).
955
956		* added primops to deal with divisions of all flavors to the frontend
957
958		* handled these primops all the way through so they map to their respective
959		MLRISC support
960
961		* used these primops in the implementation of Int, Int32, Word, Word32
962
963		* removed INLDIV, INLMOD, and INLREM as they are no longer necessary
964
965		* parameterized INLMIN, INLMAX, and INLABS by a numkind
966
967		* translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
968		floating point
969
970		* used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
971		and Real (but Real.abs maps to a separate floating-point-only primop)
972
973
974		TODO items:
975
976		* Hacked Alpha32 instruction selection, disabling the selection of REMx
977		instructions because the machine instruction encoder cannot handle
978		them.  (Hppa, PPC, and Sparc instruction selection did not handle
979		REM in the first place, and REM is supported by the x86 machine coder.)
980
981		* Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
982		selection phase.  (The two can be streamlined because the hardware
983		delivers both quotient and remainder at the same time anyway.)
984
985		* Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
986		(Currently the behavior is inconsistent both across architectures and
987		wrt. the draft Basis spec.)
988
989		* Word8 should eventually be handled natively, too.
990
991		* There seems to be one serious bug in mltree-gen.sml.  It appears, though,
992		as if there currently is no execution path that could trigger it in
993		SML/NJ.  (The assumptions underlying functions arith and promotable do not
994		hold for things like multiplication and division.)
995
996		----------------------------------------------------------------------
997		Name: Matthias Blume
998		Date: 2002/03/27 16:27:00 EST
999		Tag: blume-20020327-mlrisc-divisions
1000		Description:
1001
1002		Added support for all four division operations (ML's div, mod, quot,
1003		and rem) to MLRISC.  In the course of doing so, I also rationalized
1004		the naming (no more annoying switch-around of DIV and QUOT), by
1005		parameterizing the operation by div_rounding_mode (which can be either
1006		DIV_TO_ZERO or DIV_TO_NEGINF).
1007
1008		The generic MLTreeGen functor takes care of compiling all four
1009		operations down to only round-to-zero div.
1010
1011		Missing pieces:
1012
1013		* Doing something smarter than relying on MLTreeGen on architectures
1014		like, e.g., the x86 where hardware division delivers both quotient and
1015		remainder at the same time.  With this, the implementation of the
1016		round-to-neginf operations could be further streamlined.
1017
1018		* Remove inlining support for div/mod/rem from the frontend and replace it
1019		with primops that get carried through to the backend.  Do this for all
1020		int and word types.
1021
1022		----------------------------------------------------------------------
1023		Name: Matthias Blume
1024		Date: 2002/03/25 17:25:00 EST
1025		Tag: blume-20020325-divmod
1026		Description:
1027
1028		I improved (hopefully without breaking them) the implementation of Int.div,
1029		Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1030		advantage of the following observations:
1031
1032		Let  q = x quot y      r = x rem y
1033		d = x div  y      m = x mod y
1034
1035		where "quot" is the round-to-zero version of integer division that
1036		hardware usually provides.  Then we have:
1037
1038		r = x - q * y        where neither the * nor the - will overflow
1039		d = if q >= 0 orelse x = q * y then q else q - 1
1040		where neither the * nor the - will overflow
1041		m = if q >= 0 orelse r = 0 then r else r + y
1042		where the + will not overflow
1043
1044		This results in substantial simplification of the generated code.
1045		The following table shows the number of CFG nodes and edges generated
1046		for
1047		fun f (x, y) = x OPER y
1048		(* with OPER \in div, mod, quot, rem *)
1049
1050
1051		OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1052		--------------------------------------------------------
1053		div |         24 |         39 |         12 |         16
1054		mod |         41 |         71 |         12 |         16
1055		quot |          8 |         10 |          8 |         10
1056		rem |         10 |         14 |          8 |         10
1057
1058
1059		----------------------------------------------------------------------
1060		Name: Matthias Blume
1061		Date: 2002/03/25 22:06:00 EST
1062		Tag: blume-20020325-cprotobug
1063		Description:
1064
1065		Fixed a bug in cproto (c prototype decoder).
1066
1067		----------------------------------------------------------------------
1068		Name: Matthias Blume
1069		Date: 2002/03/25 16:00:00 EST
1070		Tag: blume-20020325-raw-primops
1071		Description:
1072
1073		I did some cleanup to Allen's new primop code and
1074		replaced yesterday's bootfiles with new ones.
1075		(But they are stored in the same place.)
1076
1077		----------------------------------------------------------------------
1078		Name: Matthias Blume
1079		Date: 2002/03/24 22:40:00 EST
1080		Tag: blume-20020324-bootfiles
1081		Description:
1082
1083		Made the bootfiles that Allen asked for.
1084
1085		----------------------------------------------------------------------
1086		Name: Allen Leung
1087		Date: 2002/03/23 15:50:00 EST
1088		Tag: leunga-20020323-flint-cps-rcc-primops
1089		Description:
1090
1091		1. Changes to FLINT primops:
1092
1093		(* make a call to a C-function;
1094		* The primop carries C function prototype information and specifies
1095		* which of its (ML-) arguments are floating point. C prototype
1096		* information is for use by the backend, ML information is for
1097		* use by the CPS converter. *)
1098		| RAW_CCALL of { c_proto: CTypes.c_proto,
1099		ml_args: ccall_type list,
1100		ml_res_opt: ccall_type option,
1101		reentrant : bool
1102		} option
1103		(* Allocate uninitialized storage on the heap.
1104		* The record is meant to hold short-lived C objects, i.e., they
1105		* are not ML pointers.  With the tag, the representation is
1106		* the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1107		*)
1108		| RAW_RECORD of {tag:bool,sz:int}
1109		and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1110
1111		2.  These CPS primops are now overloaded:
1112
1113		rawload of {kind:numkind}
1114		rawstore of {kind:numkind}
1115
1116		The one argument form is:
1117
1118		rawload {kind} address
1119
1120		The two argument form is:
1121
1122		rawload {kind} [ml object, byte-offset]
1123
1124		3. RAW_CCALL/RCC now takes two extra arguments:
1125
1126		a. The first is whether the C call is reentrant, i.e., whether
1127		ML state should be saved and restored.
1128		b. The second argument is a string argument specifying the name of
1129		library and the C function.
1130
1131		These things are currently not handled in the code generator, yet.
1132
1133		4. In CProto,
1134
1135		An encoding type of "bool" means "ml object" and is mapped into
1136		C prototype of PTR.  Note that "bool" is different than "string",
1137		even though "string" is also mapped into PTR, because "bool"
1138		is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1139
1140		5. Pickler/unpicker
1141
1142		Changed to handle RAW_RECORD and newest RAW_CCALL
1143
1144		6. MLRiscGen,
1145
1146		1. Changed to handle the new rawload/rawstore/rawrecord operators.
1147		2. Code for handling C Calls has been moved to a new module CPSCCalls,
1148		in the file CodeGen/cpscompile/cps-c-calls.sml
1149
1150		7. Added the conditional move operator
1151
1152		condmove of branch
1153
1154		to cps.  Generation of this is still buggy so it is currently
1155		disabled.
1156
1157		----------------------------------------------------------------------
1158		Name: Lal George
1159		Date: 2002/03/22 14:18:25 EST
1160		Tag: george-20020322-cps-branch-prob
1161		Description:
1162
1163		Implemented the Ball-Larus branch prediction-heuristics, and
1164		incorporated graphical viewers for control flow graphs.
1165
1166		Ball-Larus Heuristics:
1167		---------------------
1168		See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1169
1170		By design it uses the Dempster-Shafer theory for combining
1171		probabilities.  For example, in the function:
1172
1173		fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1174
1175		the ball-larus heuristics predicts that the n=0 is unlikely
1176		(OH-heuristic), and the 'then' branch is unlikely because of the
1177		RH-heuristic -- giving the 'then' branch an even lower combined
1178		probability using the Dempster-Shafer theory.
1179
1180		Finally, John Reppy's loop analysis in MLRISC, further lowers the
1181		probability of the 'then' branch because of the loop in the else
1182		branch.
1183
1184
1185		Graphical Viewing:
1186		------------------
1187		I merely plugged in Allen's graphical viewers into the compiler. The
1188		additional code is not much. At the top level, saying:
1189
1190		Control.MLRISC.getFlag "cfg-graphical-view" := true;
1191
1192		will display the graphical view of the control flow graph just before
1193		back-patching.  daVinci must be in your path for this to work. If
1194		daVinci is not available, then the default viewer can be changed
1195		using:
1196
1197		Control.MLRISC.getString "viewer"
1198
1199		which can be set to "dot" or "vcg" for the corresponding viewers. Of
1200		course, these viewers must be in your path.
1201
1202		The above will display the compilation unit at the level of clusters,
1203		many of which are small, boring, and un-interesting. Also setting:
1204
1205		Control.MLRISC.getInt "cfg-graphical-view_size"
1206
1207		will display clusters that are larger than the value set by the above.
1208
1209
1210		----------------------------------------------------------------------
1211		Name: Matthias Blume
1212		Date: 2002/03/21 22:20:00 EST
1213		Tag: blume-20020321-kmp-bugfix
1214		Description:
1215
1216		Changed the interface to the KMP routine in PreString and fixed
1217		a minor bug in one place where it was used.
1218
1219		----------------------------------------------------------------------
1220		Name: Allen Leung
1221		Date: 2002/03/21 20:30:00 EST
1222		Tag: leunga-20020321-cfg
1223		Description:
1224
1225		Fixed a potential problem in cfg edge splitting.
1226
1227		----------------------------------------------------------------------
1228		Name: Allen Leung
1229		Date: 2002/03/21 17:15:00 EST
1230		Tag: leunga-20020321-x86-fp-cfg
1231		Description:
1232
1233		1. Recoded the buggy parts of x86-fp.
1234
1235		a. All the block reordering code has been removed.
1236		We now depend on the block placement phases to do this work.
1237
1238		b. Critical edge splitting code has been simplified and moved into the
1239		CFG modules, as where they belong.
1240
1241		Both of these were quite buggy and complex.  The code is now much, much
1242		simpler.
1243
1244		2. X86 backend.
1245
1246		a. Added instructions for 64-bit support.  Instruction selection for
1247		64-bit has not been committed, however, since that
1248		requires changes to MLTREE which haven't been approved by
1249		Lal and John.
1250
1251		b. Added support for FUCOMI and FUCOMIP when generating code for
1252		PentiumPro and above.  We only generate these instructions in
1253		the fast-fp mode.
1254
1255		c. Added cases for JP and JNP in X86FreqProps.
1256
1257		3. CFG
1258
1259		CFG now has a bunch of methods for edge splitting and merging.
1260
1261		4. Machine description.
1262
1263		John's simplification of MLTREE_BASIS.fcond broke a few machine
1264		description things:
1265
1266		rtl-build.{sig,sml} and hppa.mdl fixed.
1267
1268		NOTE: the machine description stuff in the repository is still broken.
1269		Again, I can't put my fixes in because that involves
1270		changes to MLTREE.
1271
1272		----------------------------------------------------------------------
1273		Name: Matthias Blume
1274		Date: 2002/03/20 15:55:00 EST
1275		Tag: blume-20020320-kmp
1276		Description:
1277
1278		Implemented Knuth-Morris-Pratt string matching in PreString and used
1279		it for String.isSubstring, Substring.isSubstring, and
1280		Substring.position.
1281
1282		(Might need some stress-testing.  Simple examples worked fine.)
1283
1284		----------------------------------------------------------------------
1285		Name: Matthias Blume
1286		Date: 2002/03/19 16:37:00 EST
1287		Tag: blume-20020319-witnesses
1288		Description:
1289
1290		Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1291
1292		This implements a generic mechanism for changing constness qualifiers
1293		anywhere within big C types without resorting to outright "casts".
1294		(So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1295		modify the constness at the outermost level.)
1296		The implementation of "convert" is based on the idea of "witness"
1297		values -- values that are not used by the operation but whose types
1298		"testify" to their applicability.  On the implementation side, "convert"
1299		is simply a projection (returning its second curried argument).  With
1300		cross-module inlining, it should not result in any machine code being
1301		generated.
1302
1303		----------------------------------------------------------------------
1304		Name: Matthias Blume
1305		Date: 2002/03/15 16:40:00 EST
1306		Tag: blume-20020315-basis
1307		Description:
1308
1309		Provided (preliminary?) implementations for
1310
1311		{String,Substring}.{concatWith,isSuffix,isSubstring}
1312
1313		and
1314
1315		Substring.full
1316
1317		Those are in the Basis spec but they were missing in SML/NJ.
1318
1319		----------------------------------------------------------------------
1320		Name: Matthias Blume
1321		Date: 2002/03/14 21:30:00 EST
1322		Tag: blume-20020314-controls
1323		Description:
1324
1325		Controls:
1326		---------
1327
1328		1. Factored out the recently-added Controls : CONTROLS stuff and put
1329		it into its own library $/controls-lib.cm.  The source tree for
1330		this is under src/smlnj-lib/Controls.
1331
1332		2. Changed the names of types and functions in this interface, so they
1333		make a bit more "sense":
1334
1335		module -> registry
1336		'a registry -> 'a group
1337
1338		3. The interface now deals in ref cells only.  The getter/setter interface
1339		is (mostly) gone.
1340
1341		4. Added a function that lets one register an already-existing ref cell.
1342
1343		5. Made the corresponding modifications to the rest of the code so that
1344		everything compiles again.
1345
1346		6. Changed the implementation of Controls.MLRISC back to something closer
1347		to the original.  In particular, this module (and therefore MLRISC)
1348		does not depend on Controls.  There now is some link-time code in
1349		int-sys.sml that registers the MLRISC controls with the Controls
1350		module.
1351
1352		CM:
1353		---
1354
1355		* One can now specify the lambda-split aggressiveness in init.cmi.
1356
1357		----------------------------------------------------------------------
1358		Name: Allen Leung
1359		Date: 2002/03/13 17:30:00 EST
1360		Tag: leunga-20020313-x86-fp-unary
1361		Description:
1362
1363		Bug fix for:
1364
1365		> leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1366		> Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1367		> - fun f(x,(y,z)) = Real.~ y;
1368		> [autoloading]
1369		> [autoloading done]
1370		>       fchsl   (%eax), 184(%esp)
1371		> Error: MLRisc bug: X86MCEmitter.emitInstr
1372		>
1373		> uncaught exception Error
1374		>   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1375
1376		The problem was that the code generator did not generate any fp registers
1377		in this case, and the ra didn't know that it needed to run the X86FP phase to
1378		translate the pseudo fp instruction.   This only happened with unary fp
1379		operators in certain situations.
1380
1381		----------------------------------------------------------------------
1382		Name: Matthias Blume
1383		Date: 2002/03/13 14:00:00 EST
1384		Tag: blume-20020313-overload-etc
1385		Description:
1386
1387		1. Added _overload as a synonym for overload for backward compatibility.
1388		(Control.overloadKW must be true for either version to be accepted.)
1389
1390		2. Fixed bug in install script that caused more things to be installed
1391		than what was requested in config/targets.
1392
1393		3. Made CM aware of the (_)overload construct so that autoloading
1394		works.
1395
1396		----------------------------------------------------------------------
1397		Name: Matthias Blume
1398		Date: 2002/03/12 22:03:00 EST
1399		Tag: blume-20020312-url
1400		Description:
1401
1402		Forgot to update BOOT and srcarchiveurl.
1403
1404		----------------------------------------------------------------------
1405		Name: Matthias Blume
1406		Date: 2002/03/12 17:30:00 EST
1407		Tag: blume-20020312-version110392
1408		Description:
1409
1410		Yet another version number bump (because of small changes to the
1411		binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1412
1413		Changes:
1414
1415		The new pid generation scheme described a few weeks ago was overly
1416		complicated.  I implemented a new mechanism that is simpler and
1417		provides a bit more "stability":  Once CM has seen a compilation
1418		unit, it keeps its identity constant (as long as you do not delete
1419		those crucial CM/GUID/* files).  This means that when you change
1420		an interface, compile, then go back to the old interface, and
1421		compile again, you arrive at the original pid.
1422
1423		There now also is a mechanism that instructs CM to use the plain
1424		environment hash as a module's pid (effectively making its GUID
1425		the empty string).  For this, "noguid" must be specified as an
1426		option to the .sml file in question within its .cm file.
1427		This is most useful for code that is being generated by tools such
1428		as ml-nlffigen (because during development programmers tend to
1429		erase the tool's entire output directory tree including CM's cached
1430		GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1431		revert to the old, broken behavior of SML/NJ, but in specific cases
1432		where there is no danger of interface confusion, its use is ok
1433		(I think).
1434
1435		ml-nlffigen by default generates "noguid" annotations.  They can be
1436		turned off by specifying -guid in its command line.
1437
1438		----------------------------------------------------------------------
1439		Name: Lal George
1440		Date: 2002/03/12 12 14:42:36 EST
1441		Tag: george-20020312-frequency-computation
1442		Description:
1443
1444		Integrated jump chaining and static block frequency into the
1445		compiler. More details and numbers later.
1446
1447		----------------------------------------------------------------------
1448		Name: Lal George
1449		Date: 2002/03/11 11 22:38:53 EST
1450		Tag: george-20020311-jump-chain-elim
1451		Description:
1452
1453		Tested the jump chain elimination on all architectures (except the
1454		hppa).  This is on by default right now and is profitable for the
1455		alpha and x86, however, it may not be profitable for the sparc and ppc
1456		when compiling the compiler.
1457
1458		The gc test will typically jump to a label at the end of the cluster,
1459		where there is another jump to an external cluster containing the actual
1460		code to invoke gc. This is to allow factoring of common gc invocation
1461		sequences. That is to say, we generate:
1462
1463		f:
1464		testgc
1465		ja   L1      % jump if above to L1
1466
1467		L1:
1468		jmp L2
1469
1470
1471		After jump chain elimination the 'ja L1' instructions is converted to
1472		'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1473		up being implemented in their long form (if L2 is far away) using:
1474
1475		jbe     L3      % jump if below or equal to L3
1476		jmp     L2
1477		L3:
1478		...
1479
1480
1481		For large compilation units L2  may be far away.
1482
1483
1484		----------------------------------------------------------------------
1485		Name: Matthias Blume
1486		Date: 2002/03/11 13:30:00 EST
1487		Tag: blume-20020311-mltreeeval
1488		Description:
1489
1490		A functor parameter was missing.
1491
1492		----------------------------------------------------------------------
1493		Name: Allen Leung
1494		Date: 2002/03/11 10:30:00 EST
1495		Tag: leunga-20020311-runtime-string0
1496		Description:
1497
1498		The representation of the empty string now points to a
1499		legal null terminated C string instead of unit.  It is now possible
1500		to convert an ML string into C string with InlineT.CharVector.getData.
1501		This compiles into one single machine instruction.
1502
1503		----------------------------------------------------------------------
1504		Name: Allen Leung
1505		Date: 2002/03/10 23:55:00 EST
1506		Tag: leunga-20020310-x86-call
1507		Description:
1508
1509		Added machine generation for CALL instruction (relative displacement mode)
1510
1511		----------------------------------------------------------------------
1512		Name: Matthias Blume
1513		Date: 2002/03/08 16:05:00
1514		Tag: blume-20020308-entrypoints
1515		Description:
1516
1517		Version number bumped to 110.39.1.  NEW BOOTFILES!
1518
1519		Entrypoints: non-zero offset into a code object where execution should begin.
1520
1521		- Added the notion of an entrypoint to CodeObj.
1522		- Added reading/writing of entrypoint info to Binfile.
1523		- Made runtime system bootloader aware of entrypoints.
1524		- Use the address of the label of the first function given to mlriscGen
1525		as the entrypoint.  This address is currently always 0, but it will
1526		not be 0 once we turn on block placement.
1527		- Removed the linkage cluster code (which was The Other Way(tm) of dealing
1528		with entry points) from mlriscGen.
1529
1530		----------------------------------------------------------------------
1531		Name: Allen Leung
1532		Date: 2002/03/07 20:45:00 EST
1533		Tag: leunga-20020307-x86-cmov
1534		Description:
1535
1536		Bug fixes for CMOVcc on x86.
1537
1538		1. Added machine code generation for CMOVcc
1539		2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1540		3. CMOVcc cannot have an immediate operand as argument.
1541
1542		----------------------------------------------------------------------
1543		Name: Matthias Blume
1544		Date: 2002/03/07 16:15:00 EST
1545		Tag: blume-20020307-controls
1546		Description:
1547
1548		This is a very large but mostly boring patch which makes (almost)
1549		every tuneable compiler knob (i.e., pretty much everything under
1550		Control.* plus a few other things) configurable via both the command
1551		line and environment variables in the style CM did its configuration
1552		until now.
1553
1554		Try starting sml with '-h' (or, if you are brave, '-H')
1555
1556		To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1557		implements the underlying generic mechanism.
1558
1559		The interface to some of the existing such facilities has changed somewhat.
1560		For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1561		(The getFoo interface is still there for backward-compatibility, but its
1562		use is deprecated.)
1563
1564		The ml-build script passes -Cxxx=yyy command-line arguments through so
1565		that one can now twiddle the compiler settings when using this "batch"
1566		compiler.
1567
1568		TODO items:
1569
1570		We should go through and throw out all controls that are no longer
1571		connected to anything.  Moreover, we should go through and provide
1572		meaningful (and correct!) documentation strings for those controls
1573		that still are connected.
1574
1575		Currently, multiple calls to Controls.new are accepted (only the first
1576		has any effect).  Eventually we should make sure that every control
1577		is being made (via Controls.new) exactly once.  Future access can then
1578		be done using Controls.acc.
1579
1580		Finally, it would probably be a good idea to use the getter-setter
1581		interface to controls rather than ref cells.  For the time being, both
1582		styles are provided by the Controls module, but getter-setter pairs are
1583		better if thread-safety is of any concern because they can be wrapped.
1584
1585		*****************************************
1586
1587		One bug fix: The function blockPlacement in three of the MLRISC
1588		backpatch files used to be hard-wired to one of two possibilities at
1589		link time (according to the value of the placementFlag).  But (I
1590		think) it should rather sense the flag every time.
1591
1592		*****************************************
1593
1594		Other assorted changes (by other people who did not supply a HISTORY entry):
1595
1596		1. the cross-module inliner now works much better (Monnier)
1597		2. representation of weights, frequencies, and probabilities in MLRISC
1598		changed in preparation of using those for weighted block placement
1599		(Reppy, George)
1600
1601		----------------------------------------------------------------------
1602		Name: Lal George
1603		Date: 2002/03/07 14:44:24 EST 2002
1604		Tag: george-20020307-weighted-block-placement
1605
1606		Tested the weighted block placement optimization on all architectures
1607		(except the hppa) using AMPL to generate the block and edge frequencies.
1608		Changes were required in the machine properties to correctly
1609		categorize trap instructions. There is an MLRISC flag
1610		"weighted-block-placement" that can be used to enable weighted block
1611		placement, but this will be ineffective without block/edge
1612		frequencies (coming soon).
1613
1614
1615		----------------------------------------------------------------------
1616		Name: Lal George
1617		Date: 2002/03/05 17:24:48 EST
1618		Tag: george-20020305-linkage-cluster
1619
1620		In order to support the block placement optimization, a new cluster
1621		is generated as the very first cluster (called the linkage cluster).
1622		It contains a single jump to the 'real' entry point for the compilation
1623		unit. Block placement has no effect on the linkage cluster itself, but
1624		all the other clusters  have full freedom in the manner in which they
1625		reorder blocks or functions.
1626
1627		On the x86 the typical linkage code that is generated is:
1628		----------------------
1629		.align 2
1630		L0:
1631		addl    $L1-L0, 72(%esp)
1632		jmp     L1
1633
1634
1635		.align  2
1636		L1:
1637		----------------------
1638
1639		72(%esp) is the memory location for the stdlink register. This
1640		must contain the address of the CPS function being called. In the
1641		above example, it contains the address of  L0; before
1642		calling L1 (the real entry point for the compilation unit), it
1643		must contain the address for L1, and hence
1644
1645		addl $L1-L0, 72(%esp)
1646
1647		I have tested this on all architectures except the hppa.The increase
1648		in code size is of course negligible
1649
1650		----------------------------------------------------------------------
1651		Name: Allen Leung
1652		Date: 2002/03/03 13:20:00 EST
1653		Tag: leunga-20020303-mlrisc-tools
1654
1655		Added #[ ... ] expressions to mlrisc tools
1656
1657		----------------------------------------------------------------------
1658		Name: Matthias Blume
1659		Date: 2002/02/27 12:29:00 EST
1660		Tag: blume-20020227-cdebug
1661		Description:
1662
1663		- made types in structure C and C_Debug to be equal
1664		- got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1665		- there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1666
1667		----------------------------------------------------------------------
1668		Name: Matthias Blume
1669		Date: 2002/02/26 12:00:00 EST
1670		Tag: blume-20020226-ffi
1671		Description:
1672
1673		1. Fixed a minor bug in CM's "noweb" tool:
1674		If numbering is turned off, then truly don't number (i.e., do not
1675		supply the -L option to noweb).  The previous behavior was to supply
1676		-L'' -- which caused noweb to use the "default" line numbering scheme.
1677		Thanks to Chris Richards for pointing this out (and supplying the fix).
1678
1679		2. Once again, I reworked some aspects of the FFI:
1680
1681		A. The incomplete/complete type business:
1682
1683		- Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
1684		gone!
1685		- ML types representing an incomplete type are now *equal* to
1686		ML types representing their corresponding complete types (just like
1687		in C).  This is still safe because ml-nlffigen will not generate
1688		RTTI for incomplete types, nor will it generate functions that
1689		require access to such RTTI.   But when ML code generated from both
1690		incomplete and complete versions of the C type meet, the ML types
1691		are trivially interoperable.
1692
1693		NOTE:  These changes restore the full generality of the translation
1694		(which was previously lost when I eliminated functorization)!
1695
1696		B. Enum types:
1697
1698		- Structure C now has a type constructor "enum" that is similar to
1699		how the "su" constructor works.  However, "enum" is not a phantom
1700		type because each "T enum" has values (and is isomorphic to
1701		MLRep.Signed.int).
1702		- There are generic access operations for enum objects (using
1703		MLRep.Signed.int).
1704		- ml-nlffigen will generate a structure E_foo for each "enum foo".
1705		* The structure contains the definition of type "mlrep" (the ML-side
1706		representation type of the enum).  Normally, mlrep is the same
1707		as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
1708		then mlrep will be defined as a datatype -- thus facilitating
1709		pattern matching on mlrep values.
1710		("-ec" will be suppressed if there are duplicate values in an
1711		enumeration.)
1712		* Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
1713		will be generated for each C enum constant xxx.
1714		* Conversion functions m2i and i2m convert between mlrep and
1715		MLRep.Signed.int.  (Without "-ec", these functions are identities.)
1716		* Coversion functions c and ml convert between mlrep and "tag enum".
1717		* Access functions (get/set) fetch and store mlrep values.
1718		- By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
1719		enumerations are merged into one single enumeration represented by
1720		structure E_'.
1721
1722		----------------------------------------------------------------------
1723		Name: Allen Leung
1724		Date: 2002/02/25 04:45:00 EST
1725		Tag: leunga-20020225-cps-spill
1726
1727		This is a new implementation of the CPS spill phase.
1728		The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
1729		In case of problems, replace it with the old file spill.sml
1730
1731		The current compiler runs into some serious performance problems when
1732		constructing a large record.  This can happen when we try to compile a
1733		structure with many items.  Even a very simple structure like the following
1734		makes the compiler slow down.
1735
1736		structure Foo = struct
1737		val x_1 = 0w1 : Word32.int
1738		val x_2 = 0w2 : Word32.int
1739		val x_3 = 0w3 : Word32.int
1740		...
1741		val x_N = 0wN : Word32.int
1742		end
1743
1744		The following table shows the compile time, from N=1000 to N=4000,
1745		with the old compiler:
1746
1747		N
1748		1000   CPS 100 spill                           0.04u  0.00s  0.00g
1749		MLRISC ra                               0.06u  0.00s  0.05g
1750		(spills = 0 reloads = 0)
1751		TOTAL                                   0.63u  0.07s  0.21g
1752
1753		1100   CPS 100 spill                           8.25u  0.32s  0.64g
1754		MLRISC ra                               5.68u  0.59s  3.93g
1755		(spills = 0 reloads = 0)
1756		TOTAL                                   14.71u  0.99s  4.81g
1757
1758		1500   CPS 100 spill                           58.55u  2.34s  1.74g
1759		MLRISC ra                               5.54u  0.65s  3.91g
1760		(spills = 543 reloads = 1082)
1761		TOTAL                                   65.40u  3.13s  6.00g
1762
1763		2000   CPS 100 spill                           126.69u  4.84s  3.08g
1764		MLRISC ra                               0.80u  0.10s  0.55g
1765		(spills = 42 reloads = 84)
1766		TOTAL                                   129.42u  5.10s  4.13g
1767
1768		3000   CPS 100 spill                           675.59u  19.03s  11.64g
1769		MLRISC ra                               2.69u  0.27s  1.38g
1770		(spills = 62 reloads = 124)
1771		TOTAL                                   682.48u  19.61s  13.99g
1772
1773		4000   CPS 100 spill                           2362.82u  56.28s  43.60g
1774		MLRISC ra                               4.96u  0.27s  2.72g
1775		(spills = 85 reloads = 170)
1776		TOTAL                                   2375.26u  57.21s  48.00g
1777
1778		As you can see the old cps spill module suffers from some serious
1779		performance problem.  But since I cannot decipher the old code fully,
1780		instead of patching the problems up, I'm reimplementing it
1781		with a different algorithm.  The new code is more modular,
1782		smaller when compiled, and substantially faster
1783		(O(n log n) time and O(n) space).  Timing of the new spill module:
1784
1785		4000  CPS 100 spill                           0.02u  0.00s  0.00g
1786		MLRISC ra                               0.25u  0.02s  0.15g
1787		(spills=1 reloads=3)
1788		TOTAL                                   7.74u  0.34s  1.62g
1789
1790		Implementation details:
1791
1792		As far as I can tell, the purpose of the CPS spill module is to make sure the
1793		number of live variables at any program point (the bandwidth)
1794		does not exceed a certain limit, which is determined by the
1795		size of the spill area.
1796
1797		When the bandwidth is too large, we decrease the register pressure by
1798		packing live variables into spill records.  How we achieve this is
1799		completely different than what we did in the old code.
1800
1801		First, there is something about the MLRiscGen code generator
1802		that we should be aware of:
1803
1804		o MLRiscGen performs code motion!
1805
1806		In particular, it will move floating point computations and
1807		address computations involving only the heap pointer to
1808		their use sites (if there is only a single use).
1809		What this means is that if we have a CPS record construction
1810		statement
1811
1812		RECORD(k,vl,w,e)
1813
1814		we should never count the new record address w as live if w
1815		has only one use (which is often the case).
1816
1817		We should do something similar to floating point, but the transformation
1818		there is much more complex, so I won't deal with that.
1819
1820		Secondly, there are now two new cps primops at our disposal:
1821
1822		1. rawrecord of record_kind option
1823		This pure operator allocates some uninitialized storage from the heap.
1824		There are two forms:
1825
1826		rawrecord NONE [INT n]  allocates a tagless record of length n
1827		rawrecord (SOME rk) [INT n] allocates a tagged record of length n
1828		and initializes the tag.
1829
1830		2. rawupdate of cty
1831		rawupdate cty (v,i,x)
1832		Assigns to x to the ith component of record v.
1833		The storelist is not updated.
1834
1835		We use these new primops for both spilling and increment record construction.
1836
1837		1. Spilling.
1838
1839		This is implemented with a linear scan algorithm (but generalized
1840		to trees).  The algorithm will create a single spill record at the
1841		beginning of the cps function and use rawupdate to spill to it,
1842		and SELECT or SELp to reload from it.  So both spills and reloads
1843		are fine-grain operations.  In contrast, in the old algorithm
1844		"spills" have to be bundled together in records.
1845
1846		Ideally, we should sink the spill record construction to where
1847		it is needed.  We can even split the spill record into multiple ones
1848		at the places where they are needed.  But CPS is not a good
1849		representation for global code motion, so I'll keep it simple and
1850		am not attempting this.
1851
1852		2. Incremental record construction (aka record splitting).
1853
1854		Long records with many component values which are simulatenously live
1855		(recall that single use record addresses are not considered to
1856		be live) are constructed with rawrecord and rawupdate.
1857		We allocate space on the heap with rawrecord first, then gradually
1858		fill it in with rawupdate.  This is the technique suggested to me
1859		by Matthias.
1860
1861		Some restrictions on when this is applicable:
1862		1. It is not a VECTOR record.  The code generator currently does not handle
1863		this case. VECTOR record uses double indirection like arrays.
1864		2. All the record component values are defined in the same "basic block"
1865		as the record constructor.  This is to prevent speculative
1866		record construction.
1867
1868		----------------------------------------------------------------------
1869		Name: Allen Leung
1870		Date: 2002/02/22 01:02:00 EST
1871		Tag: leunga-20020222-mlrisc-tools
1872
1873		Minor bug fixes in the parser and rewriter
1874
1875		----------------------------------------------------------------------
1876		Name: Allen Leung
1877		Date: 2002/02/21 20:20:00 EST
1878		Tag: leunga-20020221-peephole
1879
1880		Regenerated the peephole files.  Some contained typos in the specification
1881		and some didn't compile because of pretty printing bugs in the old version
1882		of 'nowhere'.
1883
1884		----------------------------------------------------------------------
1885		Name: Allen Leung
1886		Date: 2002/02/19 20:20:00 EST
1887		Tag: leunga-20020219-mlrisc-tools
1888		Description:
1889
1890		Minor bug fixes to the mlrisc-tools library:
1891
1892		1.  Fixed up parsing colon suffixed keywords
1893		2.  Added the ability to shut the error messages up
1894		3.  Reimplemented the pretty printer and fixed up/improved
1895		the pretty printing of handle and -> types.
1896		4.  Fixed up generation of literal symbols in the nowhere tool.
1897		5.  Added some SML keywords to to sml.sty
1898
1899		----------------------------------------------------------------------
1900		Name: Matthias Blume
1901		Date: 2002/02/19 16:20:00 EST
1902		Tag: blume-20020219-cmffi
1903		Description:
1904
1905		A wild mix of changes, some minor, some major:
1906
1907		* All C FFI-related libraries are now anchored under $c:
1908		$/c.cm      --> $c/c.cm
1909		$/c-int.cm  --> $c/internals/c-int.cm
1910		$/memory.cm --> $c/memory/memory.cm
1911
1912		* "make" tool (in CM) now treats its argument pathname slightly
1913		differently:
1914		1. If the native expansion is an absolute name, then before invoking
1915		the "make" command on it, CM will apply OS.Path.mkRelative
1916		(with relativeTo = OS.FileSys.getDir()) to it.
1917		2. The argument will be passed through to subsequent phases of CM
1918		processing without "going native".  In particular, if the argument
1919		was an anchored path, then "make" will not lose track of that anchor.
1920
1921		* Compiler backends now "know" their respective C calling conventions
1922		instead of having to be told about it by ml-nlffigen.  This relieves
1923		ml-nlffigen from one of its burdens.
1924
1925		* The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
1926
1927		* Export C_DEBUG and C_Debug from $c/c.cm.
1928
1929		* C type encoding in ml-nlffi-lib has been improved to model the conceptual
1930		subtyping relationship between incomplete pointers and their complete
1931		counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
1932		with the convention of instantiating 'o with ('t, 'c) obj whenever
1933		the pointer target type is complete.  In the incomplete case, 'o
1934		will be instantiated with some "'c iobj" -- a type obtained by
1935		using one of the functors PointerToIncompleteType or PointerToCompleteType.
1936
1937		Operations that work on both incomplete and complete pointer types are
1938		typed as taking an 'o ptr while operations that require the target to
1939		be known are typed as taking some ('t, 'c) obj ptr.
1940
1941		voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
1942		where void is an eqtype without any values.  This makes it possible
1943		to work on voidptr values using functions meant to operate on light
1944		incomplete pointers.
1945
1946		* As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
1947		been vastly simplified.
1948
1949		----------------------------------------------------------------------
1950		Name: Matthias Blume
1951		Date: 2002/02/19 10:48:00 EST
1952		Tag: blume-20020219-pqfix
1953		Description:
1954
1955		Applied Chris Okasaki's bug fix for priority queues.
1956
1957		----------------------------------------------------------------------
1958		Name: Matthias Blume
1959		Date: 2002/02/15 17:05:00
1960		Tag: Release_110_39
1961		Description:
1962
1963		Last-minute retagging is becoming a tradition... :-(
1964
1965		This is the working release 110.39.
1966
1967		----------------------------------------------------------------------
1968		Name: Matthias Blume
1969		Date: 2002/02/15 16:00:00 EST
1970		Tag: Release_110_39-orig
1971		Description:
1972
1973		Working release 110.39.  New bootfiles.
1974
1975		(Update: There was a small bug in the installer so it wouldn't work
1976		with all shells.  So I retagged. -Matthias)
1977
1978		----------------------------------------------------------------------
1979		Name: Matthias Blume
1980		Date: 2002/02/15 14:17:00 EST
1981		Tag: blume-20020215-showbindings
1982		Description:
1983
1984		Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
1985		the latter can be useful for exploring what bindings are available at
1986		the interactive prompt.  (The first function returns only the list
1987		of symbols that are really bound, the second prints those but also the
1988		ones that CM's autoloading mechanism knows about.)
1989
1990		----------------------------------------------------------------------
1991		Name: Matthias Blume
1992		Date: 2002/02/15 12:08:00 EST
1993		Tag: blume-20020215-iptrs
1994		Description:
1995
1996		Two improvements to ml-nlffigen:
1997
1998		1. Write files only if they do not exist or if their current contents
1999		do not coincide with what's being written.  (That is, avoid messing
2000		with the time stamps unless absolutely necessary.)
2001
2002		2. Implement a "repository" mechanism for generated files related
2003		to "incomplete pointer types".   See the README file for details.
2004
2005		----------------------------------------------------------------------
2006	Name: Matthias Blume	Name: Matthias Blume
2007	Date: 2002/02/14 11:50:00 EST	Date: 2002/02/14 11:50:00 EST
2008	Tag: blume-20020214-quote	Tag: blume-20020214-quote

---

Colored Diff Long Colored Diff Unidiff Context Diff Side by Side

Legend: Removed from v.1066   changed lines   Added in v.1399

---

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