Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 1127, Fri Mar 8 01:35:33 2002 UTC	revision 1490, Tue May 11 19:32:39 2004 UTC
#	Line 13	Line 13
13	Description:	Description:
14
15	----------------------------------------------------------------------	----------------------------------------------------------------------
16		Name: Matthias Blume (blume (at) tti - c (dot) org)
17		Date: 2004/05/11 14:35:00 CDT
18		Tag: blume-20040511-win32sock
19		Description:
20
21		Added previously missing support for many socket-related functions
22		under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
23		for the voluminous patch!
24
25		(I have not tested this patch under win32 yet.)
26
27		Here is David's e-mail:
28
29		Hi,
30
31		Attached to this email you find a diff against sml/nj 110.45
32		that will enable socket support under Windows.
33
34		To apply the patch (using unix or cygwin)
35		1) gunzip runtime.diff.gz
36		2) "cd" into "src/runtime" in the source tree of a fresh
37		110.45 installation.
38		3) patch -p 1 < [your/path/to]runtime.diff
39
40		The code compiles fine but has NOT yet been extensively tested.
41		I only ran a few tests for basic socket client functionality
42		(which worked fine).  Especially the functions that use ioctl
43		are not tested at all and might not work (see below).
44
45		I implemented this since we want to move to a newer version of sml/nj
46		but need socket support in order to use it.  This is the first time I
47		even had a look at the sml/nj source,  so please review my changes
48		before making this part of the distribution!  Here are a few issues
49		that I think might be better for someone to solve who is more
50		familiar with the sml/nj source (and socket programming):
51
52		- getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
53		exception since I could not figure out what the windows equivalent
54		of these functions is
55
56		- In sockets-osdep.h there are a some #include statements that are
57		only used in a few files that include sockets-osdep.h
58
59		- In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
60		does not process its return value since I don't know how to
61		report an error upwards.
62
63		- It would probably be good to have a call to WSACleanup() when
64		the library is unloaded (if there is such a possibility).
65		Otherwise I think Windows will take care of this automatically
66		when the process finishes.
67
68		- I used ioctlsocket() as a replacement for ioctl() but I have
69		no idea if that is actually the proper replacement on Windows.
70
71		- All these issues are marked in the code by "FIXME" comments.
72
73		We use sml/nj extensively in our products and are quite happy
74		with it.  I hope this contribution will help you.
75
76		Keep up the good work!
77
78		David
79
80		----------------------------------------------------------------------
81		Name: Matthias Blume (blume (at) tti - c (dot) org)
82		Date: 2004/05/11 14:20:00 CDT
83		Tag: blume-20040511-installml
84		Description:
85
86		Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
87		the patch.)
88
89		----------------------------------------------------------------------
90		Name: Matthias Blume (blume (at) tti - c (dot) org)
91		Date: 2004/05/11 14:05:00 CDT
92		Tag: blume-20040511-nlffi-netbsd
93		Description:
94
95		Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
96		the patch.)
97
98		----------------------------------------------------------------------
99		Name: Matthias Blume (blume (at) tti - c (dot) org)
100		Date: 2004/05/11 12:05:00 CDT
101		Tag: blume-20040511-exports
102		Description:
103
104		As per request by Adam Chlipala <adam@hcoop.net>, extended various
105		export lists in compiler-related .cm-files.
106
107		----------------------------------------------------------------------
108		Name: Matthias Blume (blume (at) tti - c (dot) org)
109		Date: 2004/05/11 11:35:00 CDT
110		Tag: blume-20040511-allsource
111		Description:
112
113		The installer now honors the "src-smlnj" target again, although its meaning
114		has changed from "all sources required for the compiler" to "all sources
115		the installer knows about".  In other words, if you enable "src-smlnj"
116		in the "targets" file, then the installer will pull in sources for
117		everything.  (Notice that this refers to source code only.  Compiled
118		code is still only installed for modules that were requested explicitly
119		or which are required for other modules that were requested explicitly.)
120
121		----------------------------------------------------------------------
122		Name: Matthias Blume (blume (at) tti - c (dot) org)
123		Date: 2004/04/23 17:40:00 CDT
124		Tag: blume-20040423-ieee-scan
125		Description:
126
127		Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
128		in the exponent calculation we get INF or ZERO (depending on the mantissa
129		and the sign of the exponent).
130
131		----------------------------------------------------------------------
132		Name: Matthias Blume (blume (at) tti - c (dot) org)
133		Date: 2004/04/23 10:40:00 CDT
134		Tag: blume-20040423-ml-build
135		Description:
136
137		The ml-build script now terminates with a non-0 status when something
138		goes wrong.
139
140		----------------------------------------------------------------------
141		Name: Matthias Blume (blume (at) tti - c (dot) org)
142		Date: 2004/04/22 16:35:00 CDT
143		Tag: blume-20040422-Option
144		Description:
145
146		Made exception Option to be the same as exception Option.Option
147		(as it should be).
148
149		----------------------------------------------------------------------
150		Name: Allen Leung (leunga (at) reservoir (dot) com)
151		Date: 2004/03/19 14:40:00 EST
152		Tag: leunga-20040319-cygwin-nlffi
153		Description:
154
155		Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
156		of SML/NJ.  The problem is that
157
158		lib = dlopen(NULL, ...)
159		f   = dlsym(lib, "malloc");
160
161		does not work on Windows unless we explicitly export symbols
162		such as 'malloc' during linking.  We fixed this by explicitly
163		exporting the required symbols with the magic gcc incantation:
164
165		-Wl,--export-all cygwin.def
166
167		where cygwin.def is a file containing all the symbols that
168		we wish to export.
169
170		I suspect this is a Windows problem and we'll have to
171		do the same (somehow with windows compilers) when
172		we build the native win32 version with the system
173		calls LoadLibrary/GetProcAddress.
174
175		----------------------------------------------------------------------
176		Name: Matthias Blume (blume (at) tti - c (dot) org)
177		Date: 2004/03/04 16:35:00 CST
178		Tag: blume-20040304-intinf-fmt
179		Description:
180
181		Fixed problem with IntInf.fmt (sign would show up on the right instead
182		of on the left for BIN, OCT, and HEX).
183
184		----------------------------------------------------------------------
185		Name: Matthias Blume (blume (at) tti - c (dot) org)
186		Date: 2004/03/04 11:25:00 CST
187		Tag: blume-20040304-symlinks
188		Description:
189
190		Fixed problem with installer script (unix only) where bin/ml-yacc and
191		friends pointed (via symlinks) to absolute locations instead of just
192		.run-sml.  This was reported by Vesa A Norrman.
193
194		----------------------------------------------------------------------
195		Name: Matthias Blume (blume (at) tti - c (dot) org)
196		Date: 2004/02/13 14:50:00 CST
197		Tag: Release_110_45
198		Description:
199
200		New working version (110.45). New bootfiles.
201
202		----------------------------------------------------------------------
203		Name: Matthias Blume (blume (at) tti - c (dot) org)
204		Date: 2004/01/26 15:15:15 CST
205		Tag: blume-20040126-toplevel
206		Description:
207
208		Improved handling of exceptions at the interactive toplevel.
209
210		----------------------------------------------------------------------
211		Name: Matthias Blume (blume (at) tti - c (dot) org)
212		Date: 2004/01/26 11:25:00
213		Tag: blume-20040126-app
214		Description:
215
216		Type of top-level "app" corrected.
217		Added code for setting vp_limitPtrMask to Win32-specific runtime.
218
219		----------------------------------------------------------------------
220		Name: Matthias Blume (blume (at) tti - c (dot) org)
221		Date: 2003/11/18 17:10 CST
222		Tag: blume-20031118-basis-fiddle
223		Description:
224
225		- changed Timer interface to what might become the spec
226		- POSIX_FLAGS -> BIT_FLAGS according to spec
227		- some other minor discrepancies wrt. spec eliminated
228
229		----------------------------------------------------------------------
230		Name: Matthias Blume (blume (at) tti - c (dot) org)
231		Date: 2003/11/06 12:00:00 CST
232		Tag: Release_110_44
233		Description:
234
235		New working version (110.44). New bootfiles.
236
237		----------------------------------------------------------------------
238		Name: Matthias Blume (blume (at) tti - c (dot) org)
239		Date: 2003/11/04 11:50:00 CST
240		Tag: blume-20031104-move-libraries
241		Description:
242
243		Eliminated the "dont_move_libraries" directive in config/targets.
244		(The mechanism was broken and could not be fixed easily.  Moreover,
245		there does not seem to be any reason not to move all libraries into
246		lib during installation.  I originally implemented this directive as a
247		backward-compatibility feature when I first introduced the new CM.
248		Now that things have been stable for a long time and going back to the
249		old CM is not an option, there is no reason to keep it around.)
250
251		----------------------------------------------------------------------
252		Name: Matthias Blume (blume (at) tti - c (dot) org)
253		Date: 2003/11/03 16:00:00 CST
254		Tag: blume-20031103-installdir
255		Description:
256
257		Made installer honor INSTALLDIR variable again.  (Thanks to Chris
258		Richards for pointing out the problem and providing the solution.)
259
260		----------------------------------------------------------------------
261		Name: Matthias Blume (blume (at) tti - c (dot) org)
262		Date: 2003/10/01 17:05:00 CDT
263		Tag: blume-20031001-lal-mlrisc
264		Description:
265
266		MLRISC bug fix from Lal.
267
268		----------------------------------------------------------------------
269		Name: Matthias Blume (blume (at) tti - c (dot) org)
270		Date: 2003/09/30 16:10:00 CDT
271		Tag: blume-20030930-primio-bat
272		Description:
273
274		1. Added openVector, nullRd, and nullWr to PRIM_IO.
275		2. Improved .bat files (for Win32 port) to make things work under Win95.
276		(thanks to Aaron S. Hawley for this one)
277
278		----------------------------------------------------------------------
279		Name: Matthias Blume (blume (at) tti - c (dot) org)
280		Date: 2003/09/26 16:05:00 CDT
281		Tag: blume-20030926-wrappriv
282		Description:
283
284		Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
285
286		----------------------------------------------------------------------
287		Name: Matthias Blume (blume (at) tti - c (dot) org)
288		Date: 2003/09/26 15:00:00 CDT
289		Tag: blume-20030926-110_43_3
290		Description:
291
292		- additional cleanup
293		- version number bump, NEW BOOTFILES
294
295		----------------------------------------------------------------------
296		Name: Matthias Blume (blume (at) tti - c (dot) org)
297		Date: 2003/09/26 12:00:00 CDT
298		Tag: blume-20030926-ppautoload
299		Description:
300
301		I modified the read-eval-print loop so that the autoloader gets
302		invoked whenever the prettyprinter tries to look up a symbol that
303		is not currently defined in the toplevel environment but which
304		appears in CM's autoload registry.  As a result, we see far fewer of
305		those ?.Foo.Bar.xxx names in the prettyprinter's output.
306
307		In addition to this I tried to clean up some pieces of the Basis
308		implementation (e.g., Socket, Word8Array) in order to prevent other
309		instances of these ?.Foo.Bar.xxx names from being printed.
310
311		The mechanism that picks names for types still needs some work, though.
312		(Right now it seems that if there is a type A.t which is defined to
313		be B.u, but B is unavailable at toplevel, then A.t gets printed as
314		"?.B.u" although the perhaps more sensible solution would be to use
315		"A.t" in this case.  In other words, the prettyprinter should follow
316		a chain of DEFtycs not farther than there are corresponding toplevel
317		names in the current environment.)
318
319		----------------------------------------------------------------------
320		Name: Matthias Blume (blume (at) tti - c (dot) org)
321		Date: 2003/09/24 16:31:00 CDT
322		Tag: blume-20030924-installer
323		Description:
324
325		Another installer tweak:  All the ML code for the installer is now
326		compiled during CMB.make and put into a little library called
327		$smlnj/installer.cm.  The installation then simply invokes
328
329		sml -m $smlnj/installer.cm
330
331		and everything happens automagically.
332
333		Win32:  ML code senses value of environment variable SMLNJ_HOME.
334		Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
335		and BINDIR.
336
337		The new scheme guarantees that the ML code responsible for the installation
338		is in sync with the APIs of the main system.  Also, the installer is
339		somewhat faster because the installer script is precompiled.
340
341		----------------------------------------------------------------------
342		Name: Matthias Blume (blume (at) tti - c (dot) org)
343		Date: 2003/09/24 15:35:00 CDT
344		Tag: blume-20030924-synsock
345		Description:
346
347		Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
348		but excludes all non-blocking operations.  Defined SOCKET (in Basis)
349		and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
350		implementations of non-blocking operations from CML's Socket
351		structure.
352
353		----------------------------------------------------------------------
354		Name: Matthias Blume (blume (at) tti - c (dot) org)
355		Date: 2003/09/24 15:10:05 CDT
356		Tag: blume-20030924-sockets
357		Description:
358
359		1. Fixed SOCKET API and implementation to match Basis spec.
360		This required changing the internal representation of sockets to one
361		that remembers (for each socket file descriptor) whether it is currently
362		blocking or non-blocking.  This state is maintained lazily (i.e., a system
363		call is made only if the state actually needs to change).
364
365		2. OS-specific details of sockets were moved into separate files, thus
366		making it possible to unify the bulk of the socket implementations
367		between Unix and Win32.
368
369		3. CML's socket API changed accordingly.
370		(Note that we need to remove non-blocking functions from this API
371		since they are redundant in the case of CML!)
372
373		4. CML's socket implementation now makes use of non-blocking functions
374		provided by Basis, thus removing all OS-dependent code from this part
375		of CML.
376
377		5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
378
379		----------------------------------------------------------------------
380		Name: Matthias Blume (blume (at) tti - c (dot) org)
381		Date: 2003/09/22 12:10:00 CDT
382		Tag: blume-20030922-110_43_2
383		Description:
384
385		Made a new interim version and bootfiles for developer's bootstrapping
386		convenience.
387
388		110.43.2 -- NEW BOOTFILES
389
390		----------------------------------------------------------------------
391		Name: Matthias Blume (blume (at) tti - c (dot) org)
392		Date: 2003/09/19 15:55:00 CDT
393		Tag: blume-20030919-cmdir
394		Description:
395
396		1. new-install.sh -> install.sh
397		2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
398		3. tweaked installer so that another name instead of .cm can be chosen
399		at install time (by setting the CM_DIR_ARC environment variable
400		during installation); once installation is complete, the name is
401		fixed
402
403		----------------------------------------------------------------------
404		Name: Matthias Blume (blume (at) tti - c (dot) org)
405		Date: 2003/09/18 16:00:00 CDT
406		Tag: blume-20030918-110_43_1
407		Description:
408
409		Made a new interim version and bootfiles for developer's bootstrapping
410		convenience.
411
412		110.43.1 -- NEW BOOTFILES
413
414		----------------------------------------------------------------------
415		Name: Matthias Blume (blume (at) tti - c (dot) org)
416		Date: 2003/09/18 15:20:00 CDT
417		Tag: blume-20030918-misc
418		Description:
419
420		1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
421		this seems to be controversial at the moment) and used those in
422		Posix.ProcEnv.times.
423
424		2. Added Time.{from,to}Nanoseconds to Time.
425
426		3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
427		For example, fromLargeInt never needs to look at more than 3 "big
428		digits" to get its 53 bits of precision.
429
430		----------------------------------------------------------------------
431		Name: Matthias Blume (blume (at) tti - c (dot) org)
432		Date: 2003/09/17 16:30:00 CDT
433		Tag: blume-20030917-real32-slices
434		Description:
435
436		Added an entry to the primitive environment
437		(compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
438		added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
439
440		Removed some of the "magic" constants in real64.sml and replaced them
441		with code that generates these values from their corresponding
442		integer counterparts.
443
444		Made all(?) the slice-related changes to the Basis and made everything
445		compile again...
446
447		----------------------------------------------------------------------
448		Name: Matthias Blume (blume (at) tti - c (dot) org)
449		Date: 2003/09/15 17:45:00 CDT
450		Tag: blume-20030915-rbase
451		Description:
452
453		Fixed bug in Real.fromLargeInt.
454
455		----------------------------------------------------------------------
456		Name: Matthias Blume (blume (at) tti - c (dot) org)
457		Date: 2003/09/13 18:11:00 CDT
458		Tag: blume-20030913-libinstall
459		Description:
460
461		Minor bugfix in config/libinstall (set anchor with path to
462		standalone tool after installing it, otherwise libraries that
463		need ml-lex or ml-yacc won't compile the first time the installer
464		runs).
465
466		----------------------------------------------------------------------
467		Name: Matthias Blume (blume (at) tti - c (dot) org)
468		Date: 2003/09/12 11:45:00 CDT
469		Tag: blume-20030912-various
470		Description:
471
472		- fixed bug in Real.toLargeInt
473		- fixed bug in Posix.ProcEnv.times
474		- changed inputLine functions to return an option
475		- minor installer improvements / bugfixes
476		- changed default @SMLalloc parameter for x86/celeron to 64k
477
478		----------------------------------------------------------------------
479		Name: Matthias Blume (blume (at) tti - c (dot) org)
480		Date: 2003/09/09 22:00:00 CDT
481		Tag: Release_110_43
482		Description:
483
484		New working release 110.43.  New bootfiles.
485
486		----------------------------------------------------------------------
487		Name: Matthias Blume (blume (at) tti - c (dot) org)
488		Date: 2003/09/09 19:20:00 CDT
489		Tag: blume-20030909-installer
490		Description:
491
492		Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
493		Modified config/install.bat to take advantage of it.  Also modified
494		config/install.sh (and called it config/new-install.sh) to take advantage
495		of it on Unix systems.  (The SML code is (supposed to be) platform-
496		independent.)
497
498		The installer can now install everything under Win32
499		as well as under *nix as long as it compiles.
500
501		Other changes:
502
503		- made CML compile again under Win32
504		- made eXene compile under Win32 (by providing a fake structure UnixSock
505		and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
506		- fixed a bug in nowhere: it assumed that type OS.Process.status is the
507		same as type int;  under Win32 it isn't
508		- fixed some slice-related problems in the win32-specific parts of CML
509		- added a functor argument "sameVol" to os-path-fn.sml in the Basis
510		(under Win32, the volume name is case-insensitive, and the
511		OS.Path code compares volume names for equality)
512
513		----------------------------------------------------------------------
514		Name: Matthias Blume (blume (at) tti - c (dot) org)
515		Date: 2003/09/08 11:55:00 CDT
516		Tag: blume-20030908-fullpath
517		Description:
518
519		Made Win32 version of OS.FileSys.fullPath return current directory
520		when given an empty string.  This is what the spec says, and incidentally,
521		CM depends on it.  (CM otherwise goes into an infinite loop in certain
522		cases when presented with the name of a non-existing .cm file.)
523
524		----------------------------------------------------------------------
525		Name: Matthias Blume (blume (at) tti - c (dot) org)
526		Date: 2003/09/04 16:30:00 CDT
527		Tag: blume-20030905-slices-etc
528		Description:
529
530		1. Changed interface to vectors and arrays in Basis to match
531		(draft) Basis spec.
532		2. Added signatures and implementations of slices according to
533		Basis spec.
534		3. Edited source code throughout the system to make it compile again
535		under 1. and 2.  (In some cases code had to be added to have it
536		match the new signatures.)
537		4. MLRISC should be backward-compatible: the copies of the originals
538		of files that needed to change under 3. were retained, the .cm files
539		check the compiler version number and use old versions when
540		appropriate.
541		5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
542		dirstream -> string option (in accordance with Basis spec).
543		6. When generating code that counts lines, ml-lex used function
544		CharVector.foldli, taking advantage of its old interface.
545		This has been replaced with the corresponding code from
546		CharVectorSlice.  (html-lex must be re-lexed!)
547		7. BitArray in smlnj-lib/Util has been extended/modified to match the
548		new MONO_ARRAY signature.  (Do we need BitArraySlice?)
549		8. Removed temporary additions (fromInternal, toInternal) from the
550		(now obsolete) IntInf in smlnj-lib/Util.
551		9. Cleaned up structure Byte.
552		10. Added localOffset, scan, and fromString to Date (according to spec).
553		Cleaned/corrected implementation of Date.
554		(Still need to check for correctness; implement better canonicalizeDate.)
555		11. Added "scan" to signature IEEE_REAL.
556		12. Some improvements to IntInf [in particular: efficiency-hack for
557		mod and rem when second operand is 2 (for parity checks).]
558		13. Changed representation of type Time.time, using a single IntInf.int
559		value counting microseconds.  This considerably simplified the
560		implementation of structure Time.  We now support negative time
561		values; scan and fromString handle signs.
562		14. Functor PrimIO now takes two additional arguments (VectorSlice and
563		ArraySlice).
564
565		----------------------------------------------------------------------
566		Name: Matthias Blume (blume (at) tti - c (dot) org)
567		Date: 2003/08/28 17:00:00 CDT
568		Tag: blume-20030828-intinf
569		Description:
570
571		This is a major update which comes with a version number bump
572		(110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
573		and an implementation of IntInf in the Basis.
574
575		There are a fairly large number of related changes and updates throughout
576		the system:
577
578		Basis:
579		- Implemented IntInf.
580		- Made LargeInt a projection of IntInf (by filtering through INTEGER).
581		- Added some missing Real64 operations, most notably Real.toLargeInt.
582		- Added FixedInt as a synonym for Int32.
583
584		compiler:
585		* Added support for a built-in intinf type.
586		- literals
587		- pattern matching
588		- conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
589		- overloading on literals and operations
590
591		This required adding a primitive type intinf, some additional
592		primops, and implementations for several non-trivial intinf
593		operations in Core.  (The intinf type is completely abstract
594		to the compiler; all operations get delegated back to the Core.)
595
596		* Intinf equality is handled by polyequal.  However, the compiler
597		does not print its usual warning in this case (since polyequal
598		is the right thing to do there).
599
600		* Improved the organization of structure InlineT.
601
602		* A word about conversion primops:
603		If conversions involving intinf do not cancel out during
604		CPS contract, then the compiler must insert calls to Core functions.
605		Since all core access must be resolved already during the FLINT
606		translate phase, it would be too late a the time of CPS contract
607		to add new Core calls.  For this reason, conversion primops
608		for intinf carry two arguments:  1. the numeric argument that
609		they are supposed to convert, and 2. the Core function that
610		can help with this conversion if necessary.  If CPS contract
611		eliminates a primop, then the associated Core function becomes
612		dead and goes away.  Intinf conversion primops that do not get
613		eliminated by CPS contract get rewritten into calls of their
614		core functions by a separate, new phase.
615
616		interactive system:
617		- Control.Print.intinfDepth controls max length of intinf constants
618		being printed. (Analogous to Control.Print.stringDepth.)
619		- Cleanup in printutil and pputil: got rid of unused stuff and
620		duplicates; replaced some of the code with code that makes better
621		use of library functionality.
622
623		CM:
624		Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
625		are no longer silent.
626
627		CKIT:
628		Fixed mismatched uses of Int32 and LargeInt.  I always decided
629		in favor of LargeInt -- which is now the same as IntInf.
630		CKIT-knowledgable people should check whether this is what's
631		intended and otherwise change things back to using Int32 or
632		FixedInt.
633
634		Throughout the code:
635		Started using IntInf.int literals and built-in operations
636		(e.g., comparison with 0) where this seems appropriate.
637
638
639		----------------------------------------------------------------------
640		Name: Dave MacQueen (dbm@cs.uchicago.edu)
641		Date: 2003/08/13 11:36:00 CDT
642		Tag: dbm-20030813-mcz-merge1
643		Description:
644
645		Merging changes from the mcz-branch development branch into trunk.
646		These changes involve replacement of the emulated old prettyprinter
647		interface with direct use of the SML/NJ Lib PP library, and fixing
648		of a couple of bugs (895, 1186) relating to error messages.  A new
649		prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
650		has been added.
651
652		----------------------------------------------------------------------
653		Name: Matthias Blume (blume (at) tti - c (dot) org)
654		Date: 2003/08/11 15:45:00 CDT
655		Tag: blume-20030811-windows
656		Description:
657
658		Version number bumped to 110.42.9.  NEW BOOTFILES!!!
659
660		http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
661
662		This patch restores SML/NJ's ability to run under win32.  There are a
663		number of changes, including fixes for several bugs that had gone
664		unnoticed until now:
665
666		- uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
667		trying to run the win32 version from within cygwin.)
668
669		- There are a number of simple .bat scripts that substitute for their
670		corresponding Unix shell-scripts.  (See below.)
671
672		- The internals of ml-build have been modified slightly.  The main
673		difference is that instead of calling ".link-sml" (or link-sml.bat)
674		using OS.Process.system, the ML process delegates this task back
675		to the script.  Otherwise problems arise in mixed environments such
676		as Cygwin where scripts look and work like Unix scripts, but
677		where OS.Process.system cannot run them.
678
679		- In CM, the srcpath pickler used native pathname syntax -- which
680		is incorrect in the case of cross-compilation.  The new pickle format
681		is independent of platform-specific naming conventions.
682
683		- Path configuration files (such as lib/pathconfig) can now choose
684		between native and standard syntax.  Placing a line of the form
685
686		standard!
687
688		into the file causes all subsequent paths to be interpreted using
689		CM standard pathname syntax (= Unix conventions); a line
690
691		native!
692
693		switches back to native style.  This was needed so that
694		path config files can be written portably, see src/system/pathconfig.
695
696		- Runtime system:
697
698		- win32-filesys.c:  get_file_time and set_file_time now
699		access modification time, not creation time.
700
701		- I/O code made aware of new array representation.
702
703		- Bug fixes in X86.prim.masm.
704
705		- src/system/makeml made aware of win32. (For use under cygwin
706		and other Unix-environments for windows.)
707
708		- In Basis, fixed off-by-one error in win32-io.sml (function vecF)
709		which caused BinIO.inputAll to fail consistently.
710
711		.bat scripts:
712
713		Windows .bat scripts assume that SMLNJ_HOME is defined.
714
715		- sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
716		applications (sml, ml-yacc, ml-lex).
717		- ml-build.bat: analogous to ml-build.
718		- config\install.bat:  Analogous to config/install.sh.  This requires
719		that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
720		(nmake etc. must be on the path, and vcvars32 must have been run.)
721		Moreover, sources for ml-lex and ml-yacc need to exist under src,
722		and the bootfile hierarchy must have been unpacked under
723		sml.boot.x86-win32.
724		The script is very primitive and does a poor job at error checking.
725		It only installs the base system, ml-lex, and ml-yacc.  No other
726		libraries are being installed (i.e., you get only those that
727		are part of the compiler.)
728		- link-sml.bat:  analogous to .link-sml, but not currently used
729
730		Unrelated bug fixes:
731
732		- ml-nlffigen now exports structures ST_* corresponding to incomplete
733		types.
734		- Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
735		otherwise.)
736
737		----------------------------------------------------------------------
738		Name: Dave MacQueen (macqueen@cs.uchicago.edu)
739		Date: 2003/06/17
740		Tag: macqueen-20030617-bug895
741		Description:
742
743		Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
744		Tag will be used for new development branch (mcz-branch) for
745		use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
746
747		----------------------------------------------------------------------
748		Name: Matthias Blume (blume (at) tti - c (dot) org)
749		Date: 2003/05/27 16:55:00 CDT
750		Tag: blume-20030527-polyeq
751		Description:
752
753		Tried to eliminated most cases of polymorphic equality.
754
755		----------------------------------------------------------------------
756		Name: Matthias Blume (blume (at) tti - c (dot) org)
757		Date: 2003/05/21 17:45:00 CDT
758		Tag: blume-20030517-complete
759		Description:
760
761		Two changes:
762
763		1. Added a flag for controlling whether non-exhaustive bindings will
764		be treated as errors (default is false).
765		2. Cleaned up the *entire* source tree so that CMB.make goes through
766		without a single non-exhaustive match- or bind warning.
767
768		----------------------------------------------------------------------
769		Name: Matthias Blume (blume (at) tti - c (dot) org)
770		Date: 2003/05/17 10:20:00 CDT
771		Tag: blume-20030517-absyn
772		Description:
773
774		1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
775
776		This mainly affects the quality of error messages.  However, some
777		of the code is now more straightforward than before. (Treatment of
778		the above four constructs in translate.sml is much simpler than
779		the "macro-expansion" that was going on before.  Plus, the mach-
780		compiler no longer gets invoked just to be able to compile an
781		if-expression.)
782
783		2. The ErrorMsg.Error exception is now caught and absorbed by the
784		interactive loop.
785
786		----------------------------------------------------------------------
787		Name: Allen Leung
788		Date: 2003/05/16 13:05:00 CDT
789		Tag: leunga-20030516-cygwin-runtime
790		Description:
791
792		Ported the runtime system to cygwin, which uses the unix
793		x86-unix bin files.  Missing/buggy features:
794
795		o getnetbyname, getnetbyaddr: these functions seem to be missing in
796		the Cygwin library.
797		o Ctrl-C handling may be flaky.
798		o Windows system calls and Windows I/O are not supported.
799
800		A new set of binfiles is located at:
801
802		http://www.dorsai.org/~leunga/boot.x86-unix.tgz
803
804		This is only needed for bootstrapping the cygwin version of smlnj.
805		Other x86 versions can use the existing binfiles.
806
807		----------------------------------------------------------------------
808		Name: Matthias Blume
809		Date: 2003/04/08 15:42:00 CDT
810		Tag: blume-20030408-listpair
811		Description:
812
813		1. Added a target 'mlrisc' to installer.
814
815		2. Added missing elements to structure ListPair.
816
817		----------------------------------------------------------------------
818		Name: Allen Leung
819		Date: 2003/01/07 10:40:00 EST
820		Tag: leunga-20030107-int-rem
821		Description:
822
823		Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
824		The arguments to the SUBL instruction were swapped.
825
826		----------------------------------------------------------------------
827		Name: Matthias Blume
828		Date: 2002/12/12 16:25:00 EST
829		Tag: blume-20021212-risc-ra
830		Description:
831
832		Fixed a serious bug in the rewrite code for FP spilling/reloading that
833		sent the RA into an infinite loop when floating point registers get
834		spilled.  (Because of this bug, e.g., nucleic stopped compiling between
835		110.37 and 110.38.)
836		There was another set of potential problems related to the handling of
837		MLRISC annotations (but those did not yet cause real problems, apparently).
838
839		----------------------------------------------------------------------
840		Name: Matthias Blume
841		Date: 2002/12/06 22:40:00 EST
842		Tag: blume-20021206-cm-fileid
843		Description:
844
845		Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
846		This fixes the problem of CM getting confused by files that suddenly
847		change their identity (e.g., by getting unlinked and recreated by some
848		text editor such as vi).  There might be a better/cheaper/cleaner way
849		of doing this, but for now this will have to do.
850
851		----------------------------------------------------------------------
852		Name: Matthias Blume
853		Date: 2002/10/28 09:50:00 EST
854		Tag: blume-20021028-typecheck
855		Description:
856
857		Exported structure Typecheck from $smlnj/viscomp/core.cm.
858
859		----------------------------------------------------------------------
860		Name: Matthias Blume
861		Date: 2002/10/17 09:10:00 EDT
862		Tag: Release_110_42
863		Description:
864
865		In good old tradition, there has been a slight hiccup so that we have
866		to patch 110.42 after the fact.   The old release tag has been replaced
867		(see below).
868
869		The change solves a problem with two competing approaches the
870		configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
871		each other's way.
872
873		This change only affects the runtime system code and the installer script.
874		(No new bootfiles.)
875
876		----------------------------------------------------------------------
877		Name: Matthias Blume
878		Date: 2002/10/16 12:00:00 EDT
879		Tag: Release_110_42_removed
880		Description:
881
882		New working release.  New bootfiles.
883
884		----------------------------------------------------------------------
885		Name: Matthias Blume
886		Date: 2002/10/10 13:10:00 EDT
887		Tag: blume-20021010-ppc-divs
888		Description:
889
890		The mltree operator DIVS must be implemented with an overflow check on
891		the PPC because the hardware indicates divide-by-zero using "overflow" as
892		well.
893
894		----------------------------------------------------------------------
895		Name: Matthias Blume
896		Date: 2002/07/23 11:45:00
897		Tag: blume-20020723-smlnj-home
898		Description:
899
900		Sml now senses the SMLNJ_HOME environment variable.  If this is set,
901		then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
902		CM_PATHCONFIG is also set), the path configuration file is assumed
903		to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
904		the entire tree to some other place and everything will "just work".
905
906		(Companion commands such as ml-build and ml-makedepend also sense this
907		variable.)
908
909		----------------------------------------------------------------------
910		Name: Matthias Blume
911		Date: 2002/07/12 21:19:00 EDT
912		Tag: blume-20020712-liveness
913		Description:
914
915		Exported two useful "step" functions from liveness module (MLRISC).
916
917		----------------------------------------------------------------------
918		Name: Matthias Blume
919		Date: 2002/07/05 16:00 EDT
920		Tag: Release_110_41
921		Description:
922
923		New working release.  New bootfiles.
924
925		----------------------------------------------------------------------
926		Name: Matthias Blume
927		Date: 2002/07/05 10:25:00 EDT
928		Tag: blume-20020705-btimp
929		Description:
930
931		Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
932		other clients can set up backtracing support.
933
934		----------------------------------------------------------------------
935		Name: Matthias Blume
936		Date: 2002/06/25 17:23:00 EDT
937		Tag: blume-20020625-fpmax
938		Description:
939
940		Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
941		case.  (The sense of the isNaN test was reversed -- which made min and
942		max always return their first argument.)
943
944		----------------------------------------------------------------------
945		Name: Matthias Blume
946		Date: 2002/06/11
947		Tag: blume-20020611-unixpath
948		Description:
949
950		Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
951
952		----------------------------------------------------------------------
953		Name: Matthias Blume
954		Date: 2002/06/10 16:35:00 EDT
955		Tag: blume-20020610-ieeereal
956		Description:
957
958		I back-ported my implementation of IEEEReal.fromString from the
959		idlbasis-devel branch so that we can test it.
960
961		Another small change is that ppDec tries to give more information
962		than just "<sig>" in the case of functors.  However, this code is
963		broken in some mysterious way if the functor's body's signature
964		has not been declared by ascription but gets inferred from the
965		implementation.  This needs fixing...
966
967		----------------------------------------------------------------------
968		Name: Matthias Blume
969		Date: 2002/05/31
970		Tag: blume-20020531-btrace-mode
971		Description:
972
973		Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
974		the wayside when I switched over to using Controls everywhere.)
975
976		----------------------------------------------------------------------
977		Name: Lal George
978		Date: 2002/05/23 12:21:40 EDT
979		Tag: george-20020523-visual-labels
980		Description:
981
982		Labels are now displayed in the graphical output to make
983		the fall-through and target blocks obvious.
984
985		----------------------------------------------------------------------
986		Name: Matthias Blume
987		Date: 2002/05/22 11:03:00 EDT
988		Tag: blume-20020522-shrink
989		Description:
990
991		John tweaked yesterday's fix for 1131 to handle an out-of-memory
992		situation that comes up when allocating huge arrays.
993
994		----------------------------------------------------------------------
995		Name: Matthias Blume
996		Date: 2002/05/21 16:00:00 EDT
997		Tag: Release_110_40
998		Description:
999
1000		New working release (110.40).  New bootfiles.
1001
1002		[Also: John Reppy fixed GC bug 1131.]
1003
1004		----------------------------------------------------------------------
1005		Name: Matthias Blume
1006		Date: 2002/05/21 12:35:00 EDT
1007		Tag: blume-20020521-cmdoc
1008		Description:
1009
1010		CM documentation update.
1011
1012		----------------------------------------------------------------------
1013		Name: Matthias Blume
1014		Date: 2002/05/21 10:55:00 EDT
1015		Tag: blume-20020521-misc
1016		Description:
1017
1018		- John tweaked runtime to be silent on heap export (except when
1019		GC messages are on).
1020		- I added a few more things (cross-compiling versions of CMB) to
1021		config/preloads (as suggestions).
1022
1023		----------------------------------------------------------------------
1024		Name: Matthias Blume
1025		Date: 2002/05/20 22:25:00 EDT
1026		Tag: blume-20020520-controls
1027		Description:
1028
1029		- Added ControlUtil structure to control-lib.cm.
1030		- Use it throughout.
1031		- Used Controls facility to define MLRISC controls (as opposed to
1032		registering MLRISC control ref cells with Controls after the fact)
1033		- Fixed messed-up controls priorities.
1034
1035		* Removed again all the stuff from config/preloads that one wouldn't
1036		be able to preload at the time the initial heap image is built.
1037		(Many libraries, e.g., CML, do not exist yet at this time.  The
1038		only libraries that can be preloaded via config/preloads are those
1039		that come bundled with the bootfiles.)
1040
1041		----------------------------------------------------------------------
1042		Name: Matthias Blume
1043		Date: 2002/05/20 10:59:00 EDT
1044		Tag: blume-20020520-preloads
1045		Description:
1046
1047		Added a lot of commented-out suggestions for things to be included
1048		in config/preloads.
1049
1050		----------------------------------------------------------------------
1051		Name: Allen Leung
1052		Date: 2002/05/18 14:20:00 EDT
1053		Tag: leunga-20020518-mdl
1054		Description:
1055
1056		o Made the mdl tool stuff compile and run again.
1057		o I've disabled all the stuff that depends on RTL specifications; they
1058		are all badly broken anyway.
1059
1060		----------------------------------------------------------------------
1061		Name: Matthias Blume
1062		Date: 2002/05/17 16:49:00 EDT
1063		Tag: blume-20020517-controls
1064		Description:
1065
1066		0. John Reppy made several modifications to the SML/NJ library.
1067		In particular, there is a shiny new controls-lib.cm.
1068
1069		1. Pushed new controls interface through compiler so that everything
1070		compiles again.
1071
1072		2. Added FormatComb and FORMAT_COMB to the CML version of the
1073		SML/NJ library (so that CML compiles again).
1074
1075		3. Modified init scripts because XXX_DEFAULT environment variables
1076		are no longer with us.  (Boot-time initialization is now done
1077		using the same environment variables that are also used for
1078		startup-time initialization of controls.)
1079
1080		----------------------------------------------------------------------
1081		Name: Lal George
1082		Date: 2002/05/15 09:20:10 EDT
1083		Tag: george-20020515-pseudo-op-decls
1084		Description:
1085
1086		All pseudo-ops emitted before the first segment declaration
1087		such as TEXT, DATA, and BSS directives are assumed to be global
1088		declarations and are emitted first in the assembly file. This is
1089		useful in a number of situations where one has pseudo-ops that are not
1090		specific to any segment, and also works around the constraint that one
1091		cannot have client pseudo-ops in the TEXT segment.
1092
1093		Because no segment is associated with these declarations it is
1094		an error to allocate any space or objects before the first segment
1095		directive and an exception will be raised. However, we cannot make
1096		this check for client pseudo-ops.
1097
1098		These top level declarations are a field in the CFG graph_info.
1099		In theory you can continue to add to this field after the CFG has been
1100		built -- provided you know what you are doing;-)
1101
1102		----------------------------------------------------------------------
1103		Name: Matthias Blume
1104		Date: 2002/05/13 16:40:00 EDT
1105		Tag: blume-20020513-pp-etc
1106		Description:
1107
1108		A few minor bugfixes:
1109
1110		- Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1111		(Bogus printouts for redefined bindings still occur.  Compiler
1112		bug should no longer occur now.  We need to redo the prettyprinter
1113		from scratch.)
1114
1115		- CM pathname printer now also adds escape sequences for ( and )
1116
1117		- commend and docu fixes for ml-nlffi
1118
1119		----------------------------------------------------------------------
1120		Name: Matthias Blume
1121		Date: 2002/05/10 16:40:00 EDT
1122		Tag: blume-20020510-erg-textio
1123		Description:
1124
1125		Applied the following bugfix provided by Emden Gansner:
1126
1127		Output is corrupted when outputSubstr is used rather than output.
1128
1129		The problem occurs when a substring
1130
1131		ss = (s, dataStart, dataLen)
1132
1133		where dataStart > 0, fills a stream buffer with avail bytes left.
1134		avail bytes of s, starting at index dataStart, are copied into the
1135		buffer, the buffer is flushed, and then the remaining dataLen-avail
1136		bytes of ss are copied into the beginning of the buffer. Instead of
1137		starting this copy at index dataStart+avail in s, the current code
1138		starts the copy at index avail.
1139
1140		Fix:
1141		In text-io-fn.sml, change line 695 from
1142		val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1143		to
1144		val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1145
1146		----------------------------------------------------------------------
1147		Name: Matthias Blume
1148		Date: 2002/04/12 13:55:00 EDT
1149		Tag: blume-20020412-assyntax
1150		Description:
1151
1152		1. Grabbed newer assyntax.h from the XFree86 project.
1153		2. Fiddled with how to compile X86.prim.asm without warnings.
1154		3. (Very) Minor cleanup in CM.
1155
1156		----------------------------------------------------------------------
1157		Name: Matthias Blume
1158		Date: 2002/04/01 (no joke!) 17:07:00 EST
1159		Tag: blume-20020401-x86div
1160		Description:
1161
1162		Added full support for div/mod/rem/quot on the x86, using the machine
1163		instruction's two results (without clumsily recomputing the remainder)
1164		directly where appropriate.
1165
1166		Some more extensive power-of-two support was added to the x86 instruction
1167		selector (avoiding expensive divs, mods, and muls where they can be
1168		replaced with cheaper shifts and masks).  However, this sort of thing
1169		ought to be done earlier, e.g., within the CPS optimizer so that
1170		all architectures benefit from it.
1171
1172		The compiler compiles to a fixed point, but changes might be somewhat
1173		fragile nevertheless.  Please, report any strange things that you might
1174		see wrt. div/mod/quot/rem...
1175
1176		----------------------------------------------------------------------
1177		Name: Matthias Blume
1178		Date: 2002/03/29 17:22:00
1179		Tag: blume-20020329-div
1180		Description:
1181
1182		Fixed my broken div/mod logic.  Unfortunately, this means that the
1183		inline code for div/mod now has one more comparison than before.
1184		Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1185		The problem was with quotient = 0, because that alone does not tell
1186		us which way the rounding went.  One then has to look at whether
1187		remainder and divisor have the same sign...  :(
1188
1189		Anyway, I replaced the bootfiles with fresh ones...
1190
1191		----------------------------------------------------------------------
1192		Name: Matthias Blume
1193		Date: 2002/03/29 14:10:00 EST
1194		Tag: blume-20020329-inlprims
1195		Description:
1196
1197		NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1198
1199		Primops have changed. This means that the bin/boot-file formats have
1200		changed as well.
1201
1202		To make sure that there is no confusion, I made a new version.
1203
1204
1205		CHANGES:
1206
1207		* removed REMT from mltree (remainder should never overflow).
1208
1209		* added primops to deal with divisions of all flavors to the frontend
1210
1211		* handled these primops all the way through so they map to their respective
1212		MLRISC support
1213
1214		* used these primops in the implementation of Int, Int32, Word, Word32
1215
1216		* removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1217
1218		* parameterized INLMIN, INLMAX, and INLABS by a numkind
1219
1220		* translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1221		floating point
1222
1223		* used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1224		and Real (but Real.abs maps to a separate floating-point-only primop)
1225
1226
1227		TODO items:
1228
1229		* Hacked Alpha32 instruction selection, disabling the selection of REMx
1230		instructions because the machine instruction encoder cannot handle
1231		them.  (Hppa, PPC, and Sparc instruction selection did not handle
1232		REM in the first place, and REM is supported by the x86 machine coder.)
1233
1234		* Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1235		selection phase.  (The two can be streamlined because the hardware
1236		delivers both quotient and remainder at the same time anyway.)
1237
1238		* Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1239		(Currently the behavior is inconsistent both across architectures and
1240		wrt. the draft Basis spec.)
1241
1242		* Word8 should eventually be handled natively, too.
1243
1244		* There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1245		as if there currently is no execution path that could trigger it in
1246		SML/NJ.  (The assumptions underlying functions arith and promotable do not
1247		hold for things like multiplication and division.)
1248
1249		----------------------------------------------------------------------
1250		Name: Matthias Blume
1251		Date: 2002/03/27 16:27:00 EST
1252		Tag: blume-20020327-mlrisc-divisions
1253		Description:
1254
1255		Added support for all four division operations (ML's div, mod, quot,
1256		and rem) to MLRISC.  In the course of doing so, I also rationalized
1257		the naming (no more annoying switch-around of DIV and QUOT), by
1258		parameterizing the operation by div_rounding_mode (which can be either
1259		DIV_TO_ZERO or DIV_TO_NEGINF).
1260
1261		The generic MLTreeGen functor takes care of compiling all four
1262		operations down to only round-to-zero div.
1263
1264		Missing pieces:
1265
1266		* Doing something smarter than relying on MLTreeGen on architectures
1267		like, e.g., the x86 where hardware division delivers both quotient and
1268		remainder at the same time.  With this, the implementation of the
1269		round-to-neginf operations could be further streamlined.
1270
1271		* Remove inlining support for div/mod/rem from the frontend and replace it
1272		with primops that get carried through to the backend.  Do this for all
1273		int and word types.
1274
1275		----------------------------------------------------------------------
1276		Name: Matthias Blume
1277		Date: 2002/03/25 17:25:00 EST
1278		Tag: blume-20020325-divmod
1279		Description:
1280
1281		I improved (hopefully without breaking them) the implementation of Int.div,
1282		Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1283		advantage of the following observations:
1284
1285		Let  q = x quot y      r = x rem y
1286		d = x div  y      m = x mod y
1287
1288		where "quot" is the round-to-zero version of integer division that
1289		hardware usually provides.  Then we have:
1290
1291		r = x - q * y        where neither the * nor the - will overflow
1292		d = if q >= 0 orelse x = q * y then q else q - 1
1293		where neither the * nor the - will overflow
1294		m = if q >= 0 orelse r = 0 then r else r + y
1295		where the + will not overflow
1296
1297		This results in substantial simplification of the generated code.
1298		The following table shows the number of CFG nodes and edges generated
1299		for
1300		fun f (x, y) = x OPER y
1301		(* with OPER \in div, mod, quot, rem *)
1302
1303
1304		OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1305		--------------------------------------------------------
1306		div |         24 |         39 |         12 |         16
1307		mod |         41 |         71 |         12 |         16
1308		quot |          8 |         10 |          8 |         10
1309		rem |         10 |         14 |          8 |         10
1310
1311
1312		----------------------------------------------------------------------
1313		Name: Matthias Blume
1314		Date: 2002/03/25 22:06:00 EST
1315		Tag: blume-20020325-cprotobug
1316		Description:
1317
1318		Fixed a bug in cproto (c prototype decoder).
1319
1320		----------------------------------------------------------------------
1321		Name: Matthias Blume
1322		Date: 2002/03/25 16:00:00 EST
1323		Tag: blume-20020325-raw-primops
1324		Description:
1325
1326		I did some cleanup to Allen's new primop code and
1327		replaced yesterday's bootfiles with new ones.
1328		(But they are stored in the same place.)
1329
1330		----------------------------------------------------------------------
1331		Name: Matthias Blume
1332		Date: 2002/03/24 22:40:00 EST
1333		Tag: blume-20020324-bootfiles
1334		Description:
1335
1336		Made the bootfiles that Allen asked for.
1337
1338		----------------------------------------------------------------------
1339		Name: Allen Leung
1340		Date: 2002/03/23 15:50:00 EST
1341		Tag: leunga-20020323-flint-cps-rcc-primops
1342		Description:
1343
1344		1. Changes to FLINT primops:
1345
1346		(* make a call to a C-function;
1347		* The primop carries C function prototype information and specifies
1348		* which of its (ML-) arguments are floating point. C prototype
1349		* information is for use by the backend, ML information is for
1350		* use by the CPS converter. *)
1351		| RAW_CCALL of { c_proto: CTypes.c_proto,
1352		ml_args: ccall_type list,
1353		ml_res_opt: ccall_type option,
1354		reentrant : bool
1355		} option
1356		(* Allocate uninitialized storage on the heap.
1357		* The record is meant to hold short-lived C objects, i.e., they
1358		* are not ML pointers.  With the tag, the representation is
1359		* the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1360		*)
1361		| RAW_RECORD of {tag:bool,sz:int}
1362		and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1363
1364		2.  These CPS primops are now overloaded:
1365
1366		rawload of {kind:numkind}
1367		rawstore of {kind:numkind}
1368
1369		The one argument form is:
1370
1371		rawload {kind} address
1372
1373		The two argument form is:
1374
1375		rawload {kind} [ml object, byte-offset]
1376
1377		3. RAW_CCALL/RCC now takes two extra arguments:
1378
1379		a. The first is whether the C call is reentrant, i.e., whether
1380		ML state should be saved and restored.
1381		b. The second argument is a string argument specifying the name of
1382		library and the C function.
1383
1384		These things are currently not handled in the code generator, yet.
1385
1386		4. In CProto,
1387
1388		An encoding type of "bool" means "ml object" and is mapped into
1389		C prototype of PTR.  Note that "bool" is different than "string",
1390		even though "string" is also mapped into PTR, because "bool"
1391		is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1392
1393		5. Pickler/unpicker
1394
1395		Changed to handle RAW_RECORD and newest RAW_CCALL
1396
1397		6. MLRiscGen,
1398
1399		1. Changed to handle the new rawload/rawstore/rawrecord operators.
1400		2. Code for handling C Calls has been moved to a new module CPSCCalls,
1401		in the file CodeGen/cpscompile/cps-c-calls.sml
1402
1403		7. Added the conditional move operator
1404
1405		condmove of branch
1406
1407		to cps.  Generation of this is still buggy so it is currently
1408		disabled.
1409
1410		----------------------------------------------------------------------
1411		Name: Lal George
1412		Date: 2002/03/22 14:18:25 EST
1413		Tag: george-20020322-cps-branch-prob
1414		Description:
1415
1416		Implemented the Ball-Larus branch prediction-heuristics, and
1417		incorporated graphical viewers for control flow graphs.
1418
1419		Ball-Larus Heuristics:
1420		---------------------
1421		See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1422
1423		By design it uses the Dempster-Shafer theory for combining
1424		probabilities.  For example, in the function:
1425
1426		fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1427
1428		the ball-larus heuristics predicts that the n=0 is unlikely
1429		(OH-heuristic), and the 'then' branch is unlikely because of the
1430		RH-heuristic -- giving the 'then' branch an even lower combined
1431		probability using the Dempster-Shafer theory.
1432
1433		Finally, John Reppy's loop analysis in MLRISC, further lowers the
1434		probability of the 'then' branch because of the loop in the else
1435		branch.
1436
1437
1438		Graphical Viewing:
1439		------------------
1440		I merely plugged in Allen's graphical viewers into the compiler. The
1441		additional code is not much. At the top level, saying:
1442
1443		Control.MLRISC.getFlag "cfg-graphical-view" := true;
1444
1445		will display the graphical view of the control flow graph just before
1446		back-patching.  daVinci must be in your path for this to work. If
1447		daVinci is not available, then the default viewer can be changed
1448		using:
1449
1450		Control.MLRISC.getString "viewer"
1451
1452		which can be set to "dot" or "vcg" for the corresponding viewers. Of
1453		course, these viewers must be in your path.
1454
1455		The above will display the compilation unit at the level of clusters,
1456		many of which are small, boring, and un-interesting. Also setting:
1457
1458		Control.MLRISC.getInt "cfg-graphical-view_size"
1459
1460		will display clusters that are larger than the value set by the above.
1461
1462
1463		----------------------------------------------------------------------
1464		Name: Matthias Blume
1465		Date: 2002/03/21 22:20:00 EST
1466		Tag: blume-20020321-kmp-bugfix
1467		Description:
1468
1469		Changed the interface to the KMP routine in PreString and fixed
1470		a minor bug in one place where it was used.
1471
1472		----------------------------------------------------------------------
1473		Name: Allen Leung
1474		Date: 2002/03/21 20:30:00 EST
1475		Tag: leunga-20020321-cfg
1476		Description:
1477
1478		Fixed a potential problem in cfg edge splitting.
1479
1480		----------------------------------------------------------------------
1481		Name: Allen Leung
1482		Date: 2002/03/21 17:15:00 EST
1483		Tag: leunga-20020321-x86-fp-cfg
1484		Description:
1485
1486		1. Recoded the buggy parts of x86-fp.
1487
1488		a. All the block reordering code has been removed.
1489		We now depend on the block placement phases to do this work.
1490
1491		b. Critical edge splitting code has been simplified and moved into the
1492		CFG modules, as where they belong.
1493
1494		Both of these were quite buggy and complex.  The code is now much, much
1495		simpler.
1496
1497		2. X86 backend.
1498
1499		a. Added instructions for 64-bit support.  Instruction selection for
1500		64-bit has not been committed, however, since that
1501		requires changes to MLTREE which haven't been approved by
1502		Lal and John.
1503
1504		b. Added support for FUCOMI and FUCOMIP when generating code for
1505		PentiumPro and above.  We only generate these instructions in
1506		the fast-fp mode.
1507
1508		c. Added cases for JP and JNP in X86FreqProps.
1509
1510		3. CFG
1511
1512		CFG now has a bunch of methods for edge splitting and merging.
1513
1514		4. Machine description.
1515
1516		John's simplification of MLTREE_BASIS.fcond broke a few machine
1517		description things:
1518
1519		rtl-build.{sig,sml} and hppa.mdl fixed.
1520
1521		NOTE: the machine description stuff in the repository is still broken.
1522		Again, I can't put my fixes in because that involves
1523		changes to MLTREE.
1524
1525		----------------------------------------------------------------------
1526		Name: Matthias Blume
1527		Date: 2002/03/20 15:55:00 EST
1528		Tag: blume-20020320-kmp
1529		Description:
1530
1531		Implemented Knuth-Morris-Pratt string matching in PreString and used
1532		it for String.isSubstring, Substring.isSubstring, and
1533		Substring.position.
1534
1535		(Might need some stress-testing.  Simple examples worked fine.)
1536
1537		----------------------------------------------------------------------
1538		Name: Matthias Blume
1539		Date: 2002/03/19 16:37:00 EST
1540		Tag: blume-20020319-witnesses
1541		Description:
1542
1543		Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1544
1545		This implements a generic mechanism for changing constness qualifiers
1546		anywhere within big C types without resorting to outright "casts".
1547		(So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1548		modify the constness at the outermost level.)
1549		The implementation of "convert" is based on the idea of "witness"
1550		values -- values that are not used by the operation but whose types
1551		"testify" to their applicability.  On the implementation side, "convert"
1552		is simply a projection (returning its second curried argument).  With
1553		cross-module inlining, it should not result in any machine code being
1554		generated.
1555
1556		----------------------------------------------------------------------
1557		Name: Matthias Blume
1558		Date: 2002/03/15 16:40:00 EST
1559		Tag: blume-20020315-basis
1560		Description:
1561
1562		Provided (preliminary?) implementations for
1563
1564		{String,Substring}.{concatWith,isSuffix,isSubstring}
1565
1566		and
1567
1568		Substring.full
1569
1570		Those are in the Basis spec but they were missing in SML/NJ.
1571
1572		----------------------------------------------------------------------
1573		Name: Matthias Blume
1574		Date: 2002/03/14 21:30:00 EST
1575		Tag: blume-20020314-controls
1576		Description:
1577
1578		Controls:
1579		---------
1580
1581		1. Factored out the recently-added Controls : CONTROLS stuff and put
1582		it into its own library $/controls-lib.cm.  The source tree for
1583		this is under src/smlnj-lib/Controls.
1584
1585		2. Changed the names of types and functions in this interface, so they
1586		make a bit more "sense":
1587
1588		module -> registry
1589		'a registry -> 'a group
1590
1591		3. The interface now deals in ref cells only.  The getter/setter interface
1592		is (mostly) gone.
1593
1594		4. Added a function that lets one register an already-existing ref cell.
1595
1596		5. Made the corresponding modifications to the rest of the code so that
1597		everything compiles again.
1598
1599		6. Changed the implementation of Controls.MLRISC back to something closer
1600		to the original.  In particular, this module (and therefore MLRISC)
1601		does not depend on Controls.  There now is some link-time code in
1602		int-sys.sml that registers the MLRISC controls with the Controls
1603		module.
1604
1605		CM:
1606		---
1607
1608		* One can now specify the lambda-split aggressiveness in init.cmi.
1609
1610		----------------------------------------------------------------------
1611		Name: Allen Leung
1612		Date: 2002/03/13 17:30:00 EST
1613		Tag: leunga-20020313-x86-fp-unary
1614		Description:
1615
1616		Bug fix for:
1617
1618		> leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1619		> Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1620		> - fun f(x,(y,z)) = Real.~ y;
1621		> [autoloading]
1622		> [autoloading done]
1623		>       fchsl   (%eax), 184(%esp)
1624		> Error: MLRisc bug: X86MCEmitter.emitInstr
1625		>
1626		> uncaught exception Error
1627		>   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1628
1629		The problem was that the code generator did not generate any fp registers
1630		in this case, and the ra didn't know that it needed to run the X86FP phase to
1631		translate the pseudo fp instruction.   This only happened with unary fp
1632		operators in certain situations.
1633
1634		----------------------------------------------------------------------
1635		Name: Matthias Blume
1636		Date: 2002/03/13 14:00:00 EST
1637		Tag: blume-20020313-overload-etc
1638		Description:
1639
1640		1. Added _overload as a synonym for overload for backward compatibility.
1641		(Control.overloadKW must be true for either version to be accepted.)
1642
1643		2. Fixed bug in install script that caused more things to be installed
1644		than what was requested in config/targets.
1645
1646		3. Made CM aware of the (_)overload construct so that autoloading
1647		works.
1648
1649		----------------------------------------------------------------------
1650		Name: Matthias Blume
1651		Date: 2002/03/12 22:03:00 EST
1652		Tag: blume-20020312-url
1653		Description:
1654
1655		Forgot to update BOOT and srcarchiveurl.
1656
1657		----------------------------------------------------------------------
1658		Name: Matthias Blume
1659		Date: 2002/03/12 17:30:00 EST
1660		Tag: blume-20020312-version110392
1661		Description:
1662
1663		Yet another version number bump (because of small changes to the
1664		binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1665
1666		Changes:
1667
1668		The new pid generation scheme described a few weeks ago was overly
1669		complicated.  I implemented a new mechanism that is simpler and
1670		provides a bit more "stability":  Once CM has seen a compilation
1671		unit, it keeps its identity constant (as long as you do not delete
1672		those crucial CM/GUID/* files).  This means that when you change
1673		an interface, compile, then go back to the old interface, and
1674		compile again, you arrive at the original pid.
1675
1676		There now also is a mechanism that instructs CM to use the plain
1677		environment hash as a module's pid (effectively making its GUID
1678		the empty string).  For this, "noguid" must be specified as an
1679		option to the .sml file in question within its .cm file.
1680		This is most useful for code that is being generated by tools such
1681		as ml-nlffigen (because during development programmers tend to
1682		erase the tool's entire output directory tree including CM's cached
1683		GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1684		revert to the old, broken behavior of SML/NJ, but in specific cases
1685		where there is no danger of interface confusion, its use is ok
1686		(I think).
1687
1688		ml-nlffigen by default generates "noguid" annotations.  They can be
1689		turned off by specifying -guid in its command line.
1690
1691		----------------------------------------------------------------------
1692		Name: Lal George
1693		Date: 2002/03/12 12 14:42:36 EST
1694		Tag: george-20020312-frequency-computation
1695		Description:
1696
1697		Integrated jump chaining and static block frequency into the
1698		compiler. More details and numbers later.
1699
1700		----------------------------------------------------------------------
1701		Name: Lal George
1702		Date: 2002/03/11 11 22:38:53 EST
1703		Tag: george-20020311-jump-chain-elim
1704		Description:
1705
1706		Tested the jump chain elimination on all architectures (except the
1707		hppa).  This is on by default right now and is profitable for the
1708		alpha and x86, however, it may not be profitable for the sparc and ppc
1709		when compiling the compiler.
1710
1711		The gc test will typically jump to a label at the end of the cluster,
1712		where there is another jump to an external cluster containing the actual
1713		code to invoke gc. This is to allow factoring of common gc invocation
1714		sequences. That is to say, we generate:
1715
1716		f:
1717		testgc
1718		ja   L1      % jump if above to L1
1719
1720		L1:
1721		jmp L2
1722
1723
1724		After jump chain elimination the 'ja L1' instructions is converted to
1725		'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1726		up being implemented in their long form (if L2 is far away) using:
1727
1728		jbe     L3      % jump if below or equal to L3
1729		jmp     L2
1730		L3:
1731		...
1732
1733
1734		For large compilation units L2  may be far away.
1735
1736
1737		----------------------------------------------------------------------
1738		Name: Matthias Blume
1739		Date: 2002/03/11 13:30:00 EST
1740		Tag: blume-20020311-mltreeeval
1741		Description:
1742
1743		A functor parameter was missing.
1744
1745		----------------------------------------------------------------------
1746		Name: Allen Leung
1747		Date: 2002/03/11 10:30:00 EST
1748		Tag: leunga-20020311-runtime-string0
1749		Description:
1750
1751		The representation of the empty string now points to a
1752		legal null terminated C string instead of unit.  It is now possible
1753		to convert an ML string into C string with InlineT.CharVector.getData.
1754		This compiles into one single machine instruction.
1755
1756		----------------------------------------------------------------------
1757		Name: Allen Leung
1758		Date: 2002/03/10 23:55:00 EST
1759		Tag: leunga-20020310-x86-call
1760		Description:
1761
1762		Added machine generation for CALL instruction (relative displacement mode)
1763
1764		----------------------------------------------------------------------
1765		Name: Matthias Blume
1766		Date: 2002/03/08 16:05:00
1767		Tag: blume-20020308-entrypoints
1768		Description:
1769
1770		Version number bumped to 110.39.1.  NEW BOOTFILES!
1771
1772		Entrypoints: non-zero offset into a code object where execution should begin.
1773
1774		- Added the notion of an entrypoint to CodeObj.
1775		- Added reading/writing of entrypoint info to Binfile.
1776		- Made runtime system bootloader aware of entrypoints.
1777		- Use the address of the label of the first function given to mlriscGen
1778		as the entrypoint.  This address is currently always 0, but it will
1779		not be 0 once we turn on block placement.
1780		- Removed the linkage cluster code (which was The Other Way(tm) of dealing
1781		with entry points) from mlriscGen.
1782
1783		----------------------------------------------------------------------
1784	Name: Allen Leung	Name: Allen Leung
1785	Date: 2002/03/07 20:45:00 EST	Date: 2002/03/07 20:45:00 EST
1786	Tag: leunga-20020307-x86-cmov	Tag: leunga-20020307-x86-cmov

---

Colored Diff Long Colored Diff Unidiff Context Diff Side by Side

Legend: Removed from v.1127   changed lines   Added in v.1490

---

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