Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 628, Tue Apr 25 21:03:34 2000 UTC	revision 756, Thu Dec 14 16:01:33 2000 UTC
#	Line 12	Line 12
12	Tag: <post-commit CVS tag>	Tag: <post-commit CVS tag>
13	Description:	Description:
14	----------------------------------------------------------------------	----------------------------------------------------------------------
15		Name: Matthias Blume
16		Date: 2000/12/15 00:01:05 JST
17		Tag: blume-20001215-dirtool
18		Description:
19
20		- "dir" tool added.
21		- improvements and cleanup to Tools structure
22		- documentation updates
23
24		----------------------------------------------------------------------
25		Name: Allen Leung
26		Date: Thu Dec 14 03:45:24 EST 2000
27		Description:
28		Tag:  leunga-20001214-int-inf
29		Description:
30
31		In IntInf, added these standard functions, which are missing from our
32		implementation:
33
34		andb : int * int -> int
35		xorb : int * int -> int
36		orb  : int * int -> int
37		notb : int -> int
38		<<   : int * word -> int
39		~>>  : int * word -> int
40
41		Not tested, I hope they are correct.
42
43		----------------------------------------------------------------------
44		Name: Allen Leung
45		Date: Fri Dec  8 19:23:26 EST 2000
46		Description:
47		Tag:  leunga-20001208-nowhere
48		Description:
49
50		Slight improvements to the 'nowhere' tool to handle OR-patterns,
51		to generate better error messages etc.  Plus a brief manual.
52
53		----------------------------------------------------------------------
54		Name: Lal George
55		Date: 2000/12/08 09:54:02 EST 2000
56		Tag: Release_110_31
57		Description:
58
59		- Version 110.31
60		----------------------------------------------------------------------
61		Name: Allen Leung
62		Date: Thu Dec  7 22:01:04 EST 2000
63		Tag:  leunga-20001207-cell-monster-hack
64		Description:
65
66		Major MLRISC internal changes.  Affect all clients.
67		Summary:
68
69		1.  Type CELLS.cell = int is now replaced by a datatype.
70		As a result, the old regmap is now gone.  Almost all interfaces
71		in MLRISC change as a consequence.
72
73		2.  A new brand version of machine description tool (v3.0) that generates
74		modules expecting the new interface.  The old version is removed.
75
76		3.  The RA interface has been further abstracted into two new functors.
77		RISC_RA and X86RA.  These functors have much simpler interfaces.
78		[See also directory MLRISC/demo.]
79
80		4.  Some other new source->source code generation tools are available:
81
82		a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
83		b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
84		I use this tool to generate the peephole optimizers---with the new
85		cell type changes, peephole rules are becoming difficult to write
86		without conditional pattern matching.
87
88		5.  More Intmap -> IntHashTable change.  Previous changes by Matthias didn't
89		cover the entire MLRISC source tree so many things broke.
90
91		6.  CM files have been moved to the subdirectory MLRISC/cm.
92		They are moved because there are a lot of them and they clutter up the
93		root dir.
94
95		7.  More detailed documentation to come...
96
97		NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
98		a makeml -rebuild first.  This is because of other other
99		changes that Matthias has made (see below).
100
101
102		----------------------------------------------------------------------
103		Name: Matthias Blume
104		Date: 2000/11/30 23:12:00 JST
105		Tag: blume-20001130-filereorg
106		Description:
107
108		Some manual updates and some file reorganizations in CM.
109
110		----------------------------------------------------------------------
111		Name: Matthias Blume
112		Date: 2000/11/24 17:45:00 JST
113		Tag: blume-20001124-link
114		Description:
115
116		Drastically improved link traversal code for the case that the dynamic
117		value was already loaded at bootstrap time.  As a result, CM and CMB
118		now both load blazingly fast -- even on a very slow machine.  Also,
119		memory consumption has been further reduced by this.
120
121		Warning: The format of the PIDMAP file has changed.  THerefore, to
122		bootstrap you have to do this:
123
124		1. Run CMB.make
125		2. Make a symbolic link for the boot directory:
126		ln -s sml.boot.ARCH-OS xxx
127		3. "Rebuild" the boot directory:
128		./makeml -boot xxx -rebuild sml ; rm xxx
129		4. Boot normally:
130		./makeml
131
132		----------------------------------------------------------------------
133		Name: Matthias Blume
134		Date: 2000/11/21 21:20:00 JST
135		Tag: blume-20001121-tools
136		Description:
137
138		Continued hacking on autoloading problem -- with success this time.
139		Also changed tool-plugin mechanism.  See new CM manual.
140
141		----------------------------------------------------------------------
142		Name: Matthias Blume
143		Date: 2000/11/19 14:30:00 JST
144		Tag:  blume-20001119-autoload
145		Description:
146
147		Some hacking to make autoloading faster.  Success for CMB, no success
148		so far for CM.  There is a reduced structure CM' that autoloads faster.
149		(This is a temporary, non-documented hack to be eliminated again when
150		the general problem is solved.)
151
152		----------------------------------------------------------------------
153		Name: Matthias Blume
154		Date: 2000/11/17 14:10:00 JST
155		Tag: blume-20001117-pickle-lib
156		Description:
157
158		1. Eliminated comp-lib.cm
159		2. Made pickle-lib.cm
160		3. Eliminated all uses of intset.sml (from comp-lib.cm)
161		4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
162		equivalent constructs from smlnj-lib.cm (INtHashTable).
163		5. Point 4. also goes for those uses of intmap.* in MLRISC.
164		Duplicated intmap modules thrown out.
165		6. Hunted down all duplicated SCC code and replaced it with
166		equivalent stuff (GraphSCCFn from smlnj-lib.cm).
167		7. Rewrote Feedback module.
168		8. Moved sortedlist.sml into viscomp-lib.cm.  Eventually it
169		should be thrown out and equivalent modules from smlnj-lib.cm
170		should be used (IntRedBlackSet, IntListSet, ...).
171
172		Confirmed that compiler compiles to fixpoint.
173
174		----------------------------------------------------------------------
175		Name: Allen Leung
176		Date: 2000/11/10 18:00:00
177		Tag: leunga-20001110-new-x86-fp
178
179		A new x86 floating point code generator has been added.
180		By default this is turned off.  To turn this on, do:
181
182		CM.autoload "$smlnj/compiler.cm";
183		Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
184
185		Changes:
186
187		1.  Changed FTAN to FPTAN so that the assembly output is correct.
188		2.  Changed the extension callback for FTANGENT to generate:
189
190		fptan
191		fstp  %st(0)
192		instead of
193		fptan
194		fstpl ftempmem
195
196		3.  Numerous assembly fixes for x86.
197
198		5.  Cleaned up the machine code output module x86/x86MC.sml and added
199		support for a whole bunch of instructions and addressing modes:
200
201		fadd/fsub/fsubr/fmul/fdiv/fdivr  %st, %st(n)
202		faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp  %st, %st(n)
203		fadd/fsub/fsubr/fmul/fdiv/fdivr  %st(n), %st
204		fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
205		fxch %st(n)
206		fld %st(n)
207		fst %st(n)
208		fst mem
209		fstp %st(n)
210		fucom %st(n)
211		fucomp %st(n)
212
213		All these are now generated when the fast fp mode is turned on.
214
215		6.  Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
216
217		----------------------------------------------------------------------
218		Name: Matthias Blume
219		Date: 2000/11/09 11:20:00 JST
220		Tag: blume-20001109-scc
221		Description:
222
223		Eliminated some code duplication:
224
225		1. Added "where" clause to GraphSCCFn in SML/NJ Library.
226		(Otherwise the functor is useless.)
227		2. Used GraphSCCFn where SCCUtilFun was used previously.
228		3. Got rid of SCCUtilFun (in comp-lib.cm).
229
230		----------------------------------------------------------------------
231		Name: Lal George
232		Date: 2000/11/06 09:02:21 EST 2000
233		Tag: Release_110_30
234		Description:
235
236		- Version 110.30
237		----------------------------------------------------------------------
238		Name: Matthias Blume
239		Date: 2000/11/04 14:45:00
240		Tag: blume-20001104-mlbuild
241		Description:
242
243		- Made ml-build faster on startup.
244		- Documentation fixes.
245
246		----------------------------------------------------------------------
247		Name: Matthias Blume
248		Date: 2000/11/02 17:00:00 JST
249		Tag: blume-20001102-condcomp
250		Description:
251
252		- Small tweaks to pickler -- new BOOTFILES!
253		- Version bumped to 110.29.2.
254		- Added conditional compilation facility to init.cmi (see comment there).
255		----------------------------------------------------------------------
256		Name: Allen Leung
257		Date: 2000/10/23 19:31:00
258		Tag: leunga-20001023-demo-ra
259
260		1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
261		2. Test programs for the graph library updated
262		3. Some new MLRISC demo programs added
263
264		----------------------------------------------------------------------
265		Name: Matthias Blume
266		Date: 2000/08/31 22:15:00 JST
267		Tag: blume-20001017-errmsg
268		Description:
269
270		More error message grief: Where there used to be no messages, there
271		now were some that had bogus error regions.  Fixed.
272
273		----------------------------------------------------------------------
274		Name: Matthias Blume
275		Date: 2000/08/31 17:30:00 JST
276		Tag: blume-20001017-v110p29p1
277		Description:
278
279		I made a version 110.29.1 with new bootfiles.
280
281		Changes:  Modified pickler/unpickler for faster and leaner unpickling.
282		CM documentation changes and a small bugfix in CM's error reporting.
283
284		----------------------------------------------------------------------
285		Name: Lal George
286		Date: 2000/09/27 14:42:35 EDT
287		Tag: george-20000927-nodestatus
288		Description:
289
290		Changed the type of the nodestatus, so that:
291
292		SPILLED(~1)             is now SPILLED
293		SPILLED(m) where m>=0   is now MEMREG(m)
294		SPILLED(s) where s<~1   is now SPILL_LOC(~s)
295
296		----------------------------------------------------------------------
297		Name: Matthias Blume
298		Date: 2000/09/07 14:45:00 JST
299		Tag: blume-20000907-cmerrmsg
300		Description:
301
302		Small tweak to CM to avoid getting ML syntax error messages twice.
303
304		----------------------------------------------------------------------
305		Name: Matthias Blume
306		Date: 2000/08/31 18:00:00 JST
307		Tag: blume-20000831-cvsbootfiles
308		Description:
309
310		New URL for boot files (because the 110.29 files on the BL server do
311		now work correctly with my updated install scripts for yacc and lex).
312
313		----------------------------------------------------------------------
314		Name: Matthias Blume
315		Date: 2000/08/08 12:33:00 JST
316		Tag: blume-20000808-manual
317		Description:
318
319		Tiny update to CM manual.
320
321		----------------------------------------------------------------------
322		Name: Allen Leung
323		Date: 2000/08/7 19:31:00
324		Tag: leunga-20000807-a-whole-bunch-of-stuff
325
326		Moby, C--, SSA, x86, machine descriptions etc.  Should only affect C--
327		and Mobdy.
328
329		1.  x86
330
331		a.  Fixes to peephole module by John and Dan.
332		b.  Assembly fix to SETcc by Allen.
333		c.  Fix to c-call by John.
334		d.  Fix to spilling by John.  (This one deals with the missing FSTPT case)
335		e.  Instruction selection optimization to SETcc as suggested by John.
336
337		For example,
338
339		MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
340
341		should generate:
342
343		MOVL a, x
344		SUBL b, x
345		SHRL 31, x
346
347		2.  IR stuff
348
349		A bunch of new DJ-graph related algorithms added.  These
350		speed up SSA construction.
351
352		3.  SSA + Scheduling
353
354		Added code for SSA and scheduling to the repository
355
356		----------------------------------------------------------------------
357		Name: Lal George
358		Date: 2000/07/27 11:53:14 EDT
359
360		Tag: lal-20000727-linux-ppc
361		Description:
362
363		Made changes to support Linux PPC.
364		p.s. I have confirmation that the 110.29 boot files work fine.
365
366		----------------------------------------------------------------------
367		Name: Matthias Blume
368		Date: 2000/07/27 17:40:00 JST
369		Tag: blume-20000727-scripts
370		Description:
371
372		!!!! WARNING !!!!
373		You must recompile the runtime system!
374		!!!! WARNING !!!!
375
376		This is basically another round of script-enhancements:
377
378		1. sml, ml-build, and ml-makedepend accept options -D and -U to define
379		and undefine CM preprocessor symbols.
380
381		2. ml-build avoids generating a new heap image if it finds that the
382		existing one is still ok.  (The condition is that no ML file had to
383		be recompiled and all ML files are found to be older that the heap
384		file.)
385
386		To make this work smoothly, I also hacked the runtime system as
387		well as SMLofNJ.SysInfo to get access to the heap image suffix
388		(.sparc-solaris, ...) that is currently being used.
389
390		Moreover, the signature of CM.mk_standalone has changed.  See the
391		CM manual.
392
393		3. ml-makedepend accepts additional options -n, -a, and -o.  (See the
394		CM manual for details.)
395
396		4. More CM manual updates:
397		- all of the above has been documented.
398		- there is now a section describing the (CM-related) command line
399		arguments that are accepted by the "sml" command
400
401		----------------------------------------------------------------------
402		Name: Matthias Blume
403		Date: 2000/07/25 16:20:00 JST
404		Tag: blume-20000725-makedepend
405		Description:
406
407		Added a script called ml-makedepend.  This can be used in makefiles
408		for Unix' make in a way very similar to the "makedepend" command for
409		C.
410
411		The script internally uses function CM.sources.
412
413		Synopsis:
414
415		ml-makedepend [-f makefile] cmfile targetname
416
417		The default for the makefile is "makefile" (or "Makefile" should
418		"makefile" not exist).
419
420		ml-makedepend adds a cmfile/targetname-specific section to this
421		makefile (after removing the previous version of this section).  The
422		section contains a single dependency specification with targetname on
423		the LHS (targetname is an arbitrary name), and a list of files derived
424		from the cmfile on the RHS.  Some of the files on the RHS are
425		ARCH/OPSYS-specific.  Therefore, ml-makedepend inserts references to
426		"make" variables $(ARCH) and $(OPSYS) in place of the corresponding
427		path names.  The makefile writer is responsible for making sure that
428		these variables have correct at the time "make" is invoked.
429
430		----------------------------------------------------------------------
431		Name: Matthias Blume
432		Date: 2000/07/22 23:30:00 JST
433		Tag: blume-20000722-urlupdate
434		Description:
435
436		Changed BOOT and config/srcarchiveurl to point to BL server:
437
438		ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
439
440		----------------------------------------------------------------------
441		Name: Matthias Blume
442		Date: 2000/07/18 18:00:00 JST
443		Tag: blume-20000718-Version_110_29
444		Description:
445
446		1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
447
448		2. Updated config/version to 110.29
449
450		3. Updated config/srcarchiveurl
451
452		3. New boot files!
453		ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
454
455		----------------------------------------------------------------------
456		Name: Matthias Blume
457		Date: 2000/07/11 13:58:00 JST
458		Tag: blume-20000711-doctypo
459		Description:
460
461		Fixed a few typos in CM manual.
462
463		----------------------------------------------------------------------
464		Name: Allen Leung
465		Date: 2000/06/15 00:38:00
466		Tag: leunga-20000704-sparc-x86
467
468		1. x86 peephole improvement sp += k; sp -= k => nop  [from John]
469		2. fix to x86 RET bug [found by Dan Grossman]
470		3. sparc assembly bug fix for ticc instructions [found by Fermin]
471
472		Affects c-- and moby only
473
474		----------------------------------------------------------------------
475		Name: Matthias Blume
476		Date: 2000/07/04 15:26:00
477		Tag: blume-20000704-trigger
478		Description:
479
480		1. Improvements to CM manual.
481		2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
482		of getting a back-trace.  The function, when called, raises an
483		internal exception which explicitly carries the full back-trace history,
484		so it is unaffected by any intervening handle-raise pairs ("trivial"
485		or not).  The interactive loop will print that history once it arrives
486		at top level.
487		Short of having all exceptions implicitly carry the full history, the
488		recommended way of using this facility is:
489		- compile your program with instrumentation "on"
490		- run it, when it raises an exception, look at the history
491		- if the history is "cut off" because of some handler, go and modify
492		your program so that it explicitly calls BTrace.trigger
493		- recompile (still instrumented), and rerun; look at the full history
494
495		----------------------------------------------------------------------
496		Name: Matthias Blume
497		Date: 2000/07/03 15:36:00 JST
498		Tag: blume-20000702-manual
499		Description:
500
501		Small corrections and updates to CM manual.
502
503		----------------------------------------------------------------------
504		Name: Matthias Blume
505		Date: 2000/06/29 16:04:00 JST
506		Tag: blume-20000629-yacctool
507		Description:
508
509		Changes:
510
511		1. Class "mlyacc" now takes separate arguments to pass options to
512		generated .sml- and .sig-files independently.
513		2. Corresponding CM manual updates.
514		3. BTrace module now also reports call sites.  (However, for loop clusters
515		it only shows from where the cluster was entered.)  There are associated
516		modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
517
518		----------------------------------------------------------------------
519		Name: Matthias Blume
520		Date: 2000/06/27 16:51:00 JST
521		Tag: blume-20000627-noweb
522		Description:
523
524		Changes:
525
526		1. Implemented "subdir" and "witness" options for noweb tool.
527		This caused some slight internal changes in CM's tool implementation.
528		2. Fixed bug in "tool plugin" mechanism.  This is essentially cleaning
529		some remaining issues from earlier path anchor changes.
530		3. Updated CM manual accordingly.
531
532		4. Changed implementation of back-tracing so that I now consider it
533		ready for prime-time.
534
535		In particular, you don't have to explicitly trigger the back-trace
536		anymore.  Instead, if you are running BTrace-instrumented code and
537		there is an uncaught exception (regardless of whether or not it was
538		raised in instrumented code), the top-level evalloop will print
539		the back-trace.
540
541		Features:
542
543		- Instrumented and uninstrumented code work together seemlessly.
544		(Of course, uninstrumented code is never mentioned in actual
545		back-traces.)
546
547		- Asymptotic time- and space-complexity of instrumented code is
548		equal to that of uninstrumented code.  (This means that
549		tail-recursion is preserved by the instrumentation phase.)
550
551		- Modules whose code has been instrumented in different sessions
552		work together without problem.
553
554		- There is no penalty whatsoever on uninstrumented code.
555
556		- There is no penalty on "raise" expressions, even in
557		instrumented code.
558
559		A potential bug (or perhaps it is a feature, too):
560
561		A back-trace reaches no further than the outermost instrumented
562		non-trivial "raise".  Here, a "trivial" raise is one that is the
563		sole RHS of a "handle" rule.  Thus, back-traces reach trough
564
565		<exp> handle e => raise e
566
567		and even
568
569		<exp> handle Foo => raise Bar
570
571		and, of course, through
572
573		<exp> handle Foo => ...
574
575		if the exception was not Foo.
576
577		Back-traces always reach right through any un-instrumented code
578		including any of its "handle" expressions, trivial or not.
579
580		To try this out, do the following:
581
582		- Erase all existing binfiles for your program.
583		(You may keep binfiles for those modules where you think you
584		definitely don't need back-tracing.)
585		- Turn on back-trace instrumentation:
586		SMLofNJ.Internals.BTrace.mode (SOME true);
587		- Recompile your program.  (I.e., run "CM.make" or "use".)
588		- You may now turn instrumentation off again (if you want):
589		SMLofNJ.Internals.BTrace.mode (SOME false);
590		- Run your program as usual.  If it raises an exception that
591		reaches the interactive toplevel, then a back-trace will
592		automatically be printed.  After that, the toplevel loop
593		will print the exception history as usual.
594
595		----------------------------------------------------------------------
596		Name: Matthias Blume
597		Date: 2000/06/26 09:56:46 JST
598		Tag: blume-20000626-setup
599		Description:
600
601		CM: - setup-parameter to "sml" added; this can be used to run arbitrary
602		ML code before and after compiling a file (e.g., to set compiler
603		flags)
604
605		Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
606		- associated changes to btrace.sml (BTrace instrumentation pass)
607		- cleaner implementation of btimp.sml (BTrace tracing and report
608		module)
609
610		CM manual: * new path encoding documented
611		* description of setup-parameter to "sml" added
612
613		The biggest user-visible change to back-tracing is that it is no
614		longer necessary to compile all traced modules within the same
615		session.  (This was a real limitation.)
616
617		----------------------------------------------------------------------
618		Name: Matthias Blume
619		Date: 2000/06/24 12:40:00 JST
620		Tag: blume-20000624-startup
621		Description:
622
623		Fixes startup slowdown problem.  (I was calling SrcPath.sync a _tad_
624		bit too often -- to put it mildly. :)
625
626		----------------------------------------------------------------------
627		Name: Matthias Blume
628		Date: 2000/06/23 18:20:00 JST
629		Tag: blume-20000623-btrace
630		Description:
631
632		This updates adds a backtrace facility to aid programmers in debugging
633		their programs.  This involves the following changes:
634
635		1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
636		keeping track of the current call stack.  When programs are compiled
637		in a special mode, the compiler will insert calls to these hooks
638		into the user program.
639		"Hook" means that it is possible for different implementations of
640		back-tracing to register themselves (at different times).
641
642		2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
643		as an Absyn.dec->Absyn.dec rewrite.  Normally this phase is turned off.
644		It can be turned on using this call:
645		SMLofNJ.Internals.BTrace.mode (SOME true);
646		Turning it off again:
647		SMLofNJ.Internals.BTrace.mode (SOME false);
648		Querying the current status:
649		SMLofNJ.Internals.BTrace.mode NONE;
650		Annotated programs are about twice as big as normal ones, and they
651		run a factor of 2 to 4 slower with a dummy back-trace plugin (one
652		where all hooks do nothing).  The slowdown with a plugin that is
653		actually useful (such as the one supplied by default) is even greater,
654		but in the case of the default plugin it is still only an constant
655		factor (amortized).
656
657		3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
658		with a sub-structure BTrace for controlling back-tracing.  In particular,
659		the above-mentioned function "mode" controls whether the annotation
660		phase is invoked by the compiler.  Another important function is
661		"trigger": when called it aborts the current execution and causes
662		the top-level loop to print a full back-trace.
663
664		4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
665		for back-tracing.  It keeps track of the dynamic call stack and in
666		addition to that it keeps a partial history at each "level" of that
667		stack.  For example, if a tail-calls b, b tail-calls c, and c tail-calls
668		d and b (at separate times, dynamically), then the report will show:
669
670		GOTO   d
671		/c
672		GOTO  \b
673		CALL   a
674
675		This shows that there was an initial non-tail call of a, then a
676		tail-call to b or c, looping behavior in a cluster of functions that
677		consist of b and c, and then a goto from that cluster (i.e., either from
678		b or from c) to d.
679
680		Note that (depending on the user program) the amount of information
681		that the back-trace module has to keep track of at each level is bounded
682		by a constant.  Thus, the whole implementation has the same asymptotical
683		complexity as the original program (both in space and in time).
684
685		5. compiler/TopLevel/interact/evalloop.sml has been modified to
686		handle the special exception SMLofNJ.Internals.BTrace.BTrace
687		which is raised by the "trigger" function mentioned above.
688
689		Notes on usage:
690
691		- Annotated code works well together with unannotated code:
692		Unannotated calls simply do not show up at all in the backtrace.
693
694		- It is not a good idea to let modules that were annotated during
695		different sessions run at the same time.  This is because the compiler
696		chooses small integers to identify individual functions, and there
697		will be clashes if different modules were compiled in separate sessions.
698		(Nothing will crash, and you will even be told about the clashes, but
699		back-trace information will in general not be useful.)
700
701		- Back-tracing can be confused by callcc and capture.
702
703		- The only way of getting a back-trace right now is to explicitly
704		invoke the "trigger" function from your user program.  Eventually, we
705		should make every exception carry back-trace information (if
706		available).  But since this creates more overhead at "raise"-time
707		(similar to the current exnHistory overhead), I have not yet
708		implemented this.  (The implementation will be rather easy.)  With
709		exceptions carrying back-trace information, this facility will be even
710		more useful because users don't need to modify their programs...
711
712		- While it is possible to compile the compiler with back-trace
713		annotations turned on (I did it to get some confidence in
714		correctness), you must make absolutely sure that core.sml and
715		btimp.sml are compiled WITHOUT annotation!  (core.sml cannot actually
716		be compiled with annotation because there is no core access yet, but
717		if you compile btimp.sml with annotation, then the system will go into
718		an infinite recursion and crash.)
719		Since CM currently does not know about BTrace, the only way to turn
720		annotations on and off for different modules of the compiler is to
721		interrupt CMB.make, change the settings, and re-invoke it.  Of course,
722		this is awkward and clumsy.
723
724		Sample sessions:
725
726		Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
727		- SMLofNJ.Internals.BTrace.mode (SOME true);
728		[autoloading]
729		[autoloading done]
730		val it = false : bool
731		- structure X = struct
732		-     fun main n = let
733		-         fun a (x, 0) = d x
734		-           | a (x, n) = b (x, n - 1)
735		-         and b (x, n) = c (x, n)
736		-         and c (x, n) = a (x, n)
737		-         and d x = e (x, 3)
738		-         and e (x, 0) = f x
739		-           | e (x, n) = e (x, n - 1)
740		-         and f 0 = SMLofNJ.Internals.BTrace.trigger ()
741		-           | f n = n * g (n - 1)
742		-         and g n = a (n, 3)
743		-     in
744		-         f n
745		-     end
746		- end;
747		structure X : sig val main : int -> int end
748		- X.main 3;
749		*** BACK-TRACE ***
750		GOTO   stdIn:4.2-13.20: X.main[2].f
751		GOTO-( stdIn:4.2-13.20: X.main[2].e
752		GOTO   stdIn:4.2-13.20: X.main[2].d
753		/ stdIn:4.2-13.20: X.main[2].a
754		| stdIn:4.2-13.20: X.main[2].b
755		GOTO-\ stdIn:4.2-13.20: X.main[2].c
756		CALL   stdIn:4.2-13.20: X.main[2].g
757		GOTO   stdIn:4.2-13.20: X.main[2].f
758		GOTO-( stdIn:4.2-13.20: X.main[2].e
759		GOTO   stdIn:4.2-13.20: X.main[2].d
760		/ stdIn:4.2-13.20: X.main[2].a
761		| stdIn:4.2-13.20: X.main[2].b
762		GOTO-\ stdIn:4.2-13.20: X.main[2].c
763		CALL   stdIn:4.2-13.20: X.main[2].g
764		GOTO   stdIn:4.2-13.20: X.main[2].f
765		GOTO-( stdIn:4.2-13.20: X.main[2].e
766		GOTO   stdIn:4.2-13.20: X.main[2].d
767		/ stdIn:4.2-13.20: X.main[2].a
768		| stdIn:4.2-13.20: X.main[2].b
769		GOTO-\ stdIn:4.2-13.20: X.main[2].c
770		CALL   stdIn:4.2-13.20: X.main[2].g
771		GOTO   stdIn:4.2-13.20: X.main[2].f
772		CALL   stdIn:2.15-17.4: X.main[2]
773		-
774
775		(Note that because of a FLINt bug the above code currently does not
776		compile without BTrace turned on.)
777
778		Here is another example, using my modified Tiger compiler:
779
780		Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
781		- SMLofNJ.Internals.BTrace.mode (SOME true);
782		[autoloading]
783		[autoloading done]
784		val it = false : bool
785		- CM.make "sources.cm";
786		[autoloading]
787		...
788		[autoloading done]
789		[scanning sources.cm]
790		[parsing (sources.cm):parse.sml]
791		[creating directory CM/SKEL ...]
792		[parsing (sources.cm):tiger.lex.sml]
793		...
794		[wrote CM/sparc-unix/semant.sml]
795		[compiling (sources.cm):main.sml]
796		[wrote CM/sparc-unix/main.sml]
797		[New bindings added.]
798		val it = true : bool
799		- Main.compile ("../testcases/merge.tig", "foo.out");
800		*** BACK-TRACE ***
801		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
802		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
803		CALL   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
804		GOTO   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
805		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
806		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
807		CALL   lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
808		/ lib/semant.sml:411.65-543.8: SemantFun[2].transDec
809		CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
810		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
811		CALL   lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
812		CALL   main.sml:1.18-118.4: Main.compile[2]
813		-
814
815		----------------------------------------------------------------------
816		Name: Matthias Blumen
817		Date: 2000/06/21 18:00:00 JST
818		Tag: blume-20000621-manual
819		Description:
820
821		CM manual update: Path environments documented.
822
823		----------------------------------------------------------------------
824		Name: Matthias Blume
825		Date: 2000/06/19 13:40:00
826		Tag: blume-20000619-manual
827		Description:
828
829		CM manual and system/README update.  This only covers the fact that
830		there are no more implicit anchors.  (Path environments and the "bind"
831		option to "cm" have yet to be documented.)
832
833		----------------------------------------------------------------------
834		Name: Matthias Blume
835		Date: 2000/06/19 11:05:00 JST
836		Tag: blume-20000619-chdir-bugfix
837		Description:
838
839		Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
840
841		----------------------------------------------------------------------
842		Name: Matthias Blume
843		Date: 2000/06/18 22:00:10 JST
844		Tag: blume-20000618-implicit-anchors-really-gone
845		Description:
846
847		I updates the previous HISTORY entry where I forgot to mention that
848		implicit anchors are no longer with us.
849
850		The current update also gets rid of the (now useless) controller
851		CM.Control.implicit_anchors.
852
853		----------------------------------------------------------------------
854		Name: Matthias Blume
855		Date: 2000/06/16 17:30:00 JST
856		Tag: blume-20000616-anchorenv
857		Description:
858
859		This patch implements the long anticipated (just kidding :) "anchor
860		environment" mechanism.  In the course of doing this, I also
861		re-implemented CM's internal "SrcPath" module from scratch.  The new
862		one should be more robust in certain boundary cases.  In any case, it
863		is a lot cleaner than its predecessor (IMHO).
864
865		This time, although there is yet another boot file format change, I
866		kept the unpickler backward-compatible.  As a result, no new bootfiles
867		are necessary and bootstrapping is straightforward.  (You cannot read
868		new bootfiles into an old system, but the other way around is no
869		problem.)
870
871		Visible changes:
872
873		** 0. Implicit path anchors (without the leading $-symbol) are no
874		longer recognized at all. This means that such path names are not
875		illegal either.  For example, the name basis.cm simply refers to a
876		local file called "basis.cm" (i.e, the name is an ordinary path
877		relative to .cm-files directory).  Or, to put it differently, only
878		names that start with $ are anchored paths.
879
880		** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
881		vanished.
882
883		John (Reppy) had critizised this as soon as I originally proposed and
884		implemented it, but at that time I did not really deeply believe
885		him. :) Now I came full-circle because I need the $<singlearc> syntax
886		in another place where it cannot be seen as an abbreviation for
887		$/<singlearc>.  To avoid the confusion, $<singlearc> now means what it
888		seems to mean (i.e., it "expands" into the corresponding anchor
889		value).
890
891		However, when paths are used as members in CM description files, it
892		continues to be true that there must be at least another arc after the
893		anchor.  This is now enforced separately during semantic analysis
894		(i.e., from a lexical/syntactical point of view, the notation is ok.)
895
896		** 2. The "cm" class now accepts an option "bind".  The option's value
897		is a sub-option list of precisely two items -- one labeled "anchor"
898		and the other one labeled "value".  As you might expect, "anchor" is
899		used to specify an anchor name to be bound, and "value" specifies what
900		the anchor is being bound to.
901
902		The value must be a directory name and can be given in either standard
903		syntax (including the possibility that it is itself an anchored path)
904		or native syntax.
905
906		Examples:
907
908		foo.cm (bind:(anchor:bar value:$mystuff/bar))
909		lib.cm (bind:(anchor:a value:"H:\\x\\y\\z"))  (* only works under windows *)
910
911		and so on.
912
913		The meaning of this is that the .cm-file will be processed with an
914		augmented anchor environment where the given anchor(s) is/are bound to
915		the given values(s).
916
917		The rationale for having this feature is this: Suppose you are trying
918		to use two different (already stable) libraries a.cm and b.cm (that
919		you perhaps didn't write yourself).  Further, suppose each of these
920		two libraries internally uses its own auxiliary library $aux/lib.cm.
921		Normally you would now have a problem because the anchor "lib" can not
922		be bound to more than one value globally.  Therefore, the project that
923		uses both a.cm and b.cm must locally redirect the anchor to some other
924		place:
925
926		a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
927		b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
928
929		This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
930		/usr/lib/smlnj/b-stuff/aux.cm, respectively.
931
932		Hard-wiring path names is a bit inflexible (and CM will verbosely warn
933		you when you do so at the time of CM.stabilize).  Therefore, you can
934		also use an anchored path as the value:
935
936		a.cm (bind:(anchor:lib value:$a-lib))
937		b.cm (bind:(anchor:lib value:$b-lib))
938
939		Now you can globally configure (using the usual CM.Anchor.anchor or
940		pathconfig machinery) bindings for "a-lib" and "b-lib".  Since "lib"
941		itself is always locally bound, setting it globally is no longer
942		meaningful or necessary (but it does not hurt either).  In fact, "lib"
943		can still be used as a global anchor for separate purposes.  As a
944		matter of fact, one can locally define "lib" in terms of a global
945		"lib":
946
947		a.cm (bind:(anchor:lib value:$lib/a))
948		b.cm (bind:(anchor:lib value:$lib/b))
949
950		** 3: The encoding of path names has changed.  This affects the way
951		path names are shown in CM's progress report and also the internal
952		protocol encoding used for parallel make.
953
954		The encoding now uses one or more ':'-separated segments.  Each
955		segments corresponds to a file that has been specified relative to the
956		file given by its preceding segment.  The first segment is either
957		relative to the CWD, absolute, or anchored.  Each segment itself is
958		basically a Unix pathname; all segments but the first are relative.
959
960		Example:
961
962		$foo/bar/baz.cm:a/b/c.sml
963
964		This path denotes the file bar/a/b/c.sml relative to the directory
965		denoted by anchor "foo".  Notice that the encoding also includes
966		baz.cm which is the .cm-file that listed a/b/c.sml.  As usual, such
967		paths are resolved relative to the .cm-files directory, so baz.cm must
968		be ignored to get the "real" pathname.
969
970		To make this fact more obvious, CM puts the names of such "virtual
971		arcs" into parentheses when they appear in progress reports. (No
972		parentheses will appear in the internal protocol encoding.)  Thus,
973		what you really see is:
974
975		$foo/bar/(baz.cm):a/b/c.sml
976
977		I find this notation to be much more informative than before.
978
979		Another new feature of the encoding is that special characters
980		including parentheses, colons, (back)slashes, and white space are
981		written as \ddd (where ddd is the decimal encoding of the character).
982
983		*** The CM manual still needs to be updated.
984
985		----------------------------------------------------------------------
986		Name: Allen Leung
987		Date: 2000/06/15 00:38:00
988		Tag: leunga-20000615-x86-peephole
989
990		x86 Peephole fix by Fermin.  Affects c-- and moby only.
991
992		----------------------------------------------------------------------
993		Name: Matthias Blume
994		Date: 2000/06/12 11:40:00
995		Tag: blume-20000612-parmakefix
996		Description:
997
998		More cleanup after changing the file naming scheme: This time I
999		repaired the parallel make mechanism for CMB.make which I broke earlier.
1000
1001		----------------------------------------------------------------------
1002		Name: Allen Leung
1003		Date: 2000/06/09 01:25:00
1004		Tag: leunga-20000609-various
1005
1006		None of these things should affect normal SML/NJ operations
1007
1008		1. Peephole improvements provided by Fermin (c--)
1009		2. New annotation DEFUSE for adding extra dependence (moby)
1010		3. New X86 LOCK instructions (moby)
1011		4. New machine description language for reservation tables (scheduling)
1012		5. Fixes to various optimization/analysis modules (branch chaining, dominator
1013		trees etc.)
1014		6. I've changed the CM files so that they can work with versions
1015		110.0.6, 110.25 and 110.28
1016
1017		----------------------------------------------------------------------
1018		Name: Matthias Blume
1019		Date: 2000/06/09 12:40:00
1020		Tag: blume-20000609-log
1021		Description:
1022
1023		- Removed all(?) remaining RCS Log entries from sources.
1024
1025		- Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
1026		anchored paths).
1027
1028		----------------------------------------------------------------------
1029		Name: Matthias Blume
1030		Date: 2000/06/07 17:00:00 JST
1031		Tag: blume-20000607-no-implicit-anchors
1032		Description:
1033
1034		1. This update changes the default setting for
1035		CM.Control.implicit_anchors from true to false.  This means that
1036		implicit anchors are no longer permitted by default.  I also tried to
1037		make sure that nothing else still relies on implicit anchors.
1038		(This is the next step on the schedule towards a CM that does not even
1039		have the notion of implicit anchors anymore.)
1040
1041		2. More CM manual updates.
1042
1043		3. I managed to track down and fix the pickling bug I mentioned last
1044		time.  Because of the previously existing workaround, this entails no
1045		immediate practical changes.
1046
1047		----------------------------------------------------------------------
1048		Name: Matthias Blume
1049		Date: 2000/06/06 11:15:00 JST
1050		Tag: blume-20000606-lazierpickle
1051		Description:
1052
1053		!!!! NEW BOOT FILES !!!!
1054
1055		* The main purpose of this update is to make library pickles lazier in
1056		order to reduce the initial space penalty for autoloading a library.
1057		As a result, it is now possible to have $smlnj/compiler.cm
1058		pre-registered.  This should take care of the many complaints or
1059		inquiries about missing structure Compiler.  This required changes to
1060		CM's internal data structures and small tweaks to some algorithms.
1061
1062		As a neat additional effect, it is no longer necessary (for the sake
1063		of lean heap image files) to distinguish between a "minimal" CM and a
1064		"full" CM.  Now, there is only one CM (i.e., the "full" version:
1065		$smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
1066		interactive top level. ($smlnj/cm/minimal.cm is gone.)
1067
1068		To make the life of compiler-hackers easier, "makeml" now also
1069		pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm).  In other
1070		words, after you bootstrap a new sml for the first time, you will not
1071		have to autoload $smlnj/cmb.cm again afterwards.  (The first time
1072		around you will still have to do it, though.)
1073
1074		* A second change consists of major updates to the CM manual.  There
1075		are now several appendices with summary information and also a full
1076		specification of the CM description file syntax.
1077
1078		* In directory src/system I added the script "allcross".  This script
1079		invokes sml and cross-compiles the compiler for all supported
1080		architectures.  (Useful when providing a new set of boot files.)
1081
1082		* There seems to be a latent bug in my "lazy pickles" mechanism.  I
1083		added a small tweak to pickle-util.sml to work around this problem,
1084		but it is not a proper fix yet.  I will investigate further.  (The
1085		effect of the bug was an inflation of library pickle size.)
1086
1087		* Version number increased to 110.28.1 (to avoid compatibility problems).
1088
1089		----------------------------------------------------------------------
1090		Name: Allen Leung
1091		Date: 2000/05/25 17:28 EDT
1092		Tag: leunga-20000525-ra
1093		Description:
1094
1095		Fixed a bug in freezing phase of the register allocator.
1096
1097		----------------------------------------------------------------------
1098		Name: Allen Leung
1099		Date: 2000/05/15 22:53 EDT
1100		Tag: leunga-20000515-alpha-x86-ra
1101		Description:
1102
1103		1. Alpha
1104
1105		Slight cleanup.  Removed the instruction SGNXL
1106
1107		2. X86
1108
1109		Added the following instructions to the instruction set:
1110
1111		ROLx, RORx,
1112		BTx, BTSx, BTLx, BTRx,
1113		XCHGx, and variants with the LOCK prefix
1114
1115		3. Register Allocation
1116
1117		The module ra-rewrite-with-renaming has been improved.
1118
1119		These have no effect on SML/NJ.
1120
1121		----------------------------------------------------------------------
1122		Name: Matthias Blume
1123		Date: 2000/05/15 16:20:00 JST
1124		Tag: blume-20000515-lightrebuild
1125		Description:
1126
1127		1. I added an alternative to "-rebuild" to "makeml".  The difference is
1128		that prior to calling CMB.make' the CM-variable "LIGHT" will be
1129		defined.  In effect, the command will not build any cross-compiler
1130		backends and therefore finish more quickly.
1131
1132		The "fixpt" script also takes a "-light" switch to be able to use
1133		this new facility while compiling for a fixpoint.
1134
1135		2. I replaced all mentions of anchored paths in group owner specifications
1136		with simple relative paths (usually starting with "..").
1137		The rationale is that a library's internal workings should not be
1138		compromised by the lack of some anchor.  (An anchor is necessary
1139		for someone who wants to refer to the library by an anchored path,
1140		but it should not be necessary to build the same library in the first
1141		place.)
1142
1143		3. I changed the way CM's tool mechanism determines the shell command
1144		string used for things like ml-yacc etc. so that it does not break
1145		when CM.Control.implicit_anchors is turned off.
1146
1147		----------------------------------------------------------------------
1148		Name: Matthias Blume
1149		Date: 2000/05/12 18:20:00 JST
1150		Tag: blume-20000512-ml-build
1151		Description:
1152
1153		Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
1154		getting installed properly (by config/install.sh).
1155
1156		----------------------------------------------------------------------
1157		Name: Matthias Blume
1158		Date: 2000/05/12 17:30:00 JST
1159		Tag: blume-20000512-anchors
1160		Description:
1161
1162		!!! NEW BOOT FILES !!!
1163
1164		This change is in preparation of fading out support for "implicitly
1165		anchored path names".  I went through all sources and used the
1166		explicit (and relatively new) $-notation.  See system/README and the
1167		CM manual for more info on this.
1168
1169		I also modified the anchoring scheme for some things such as "smlnj",
1170		"MLRISC", "cm", etc. to take advantage of the fact that explicit
1171		anchors are more expressive: anchor name and first arc do not have to
1172		coincide.  This entails the following user-visible change:
1173
1174		You have to write $smlnj/foo/bar instead of smlnj/foo/bar.  In
1175		particular, when you fire up sml with a command-line argument, say,
1176		e.g.:
1177
1178		sml '$smlnj/cmb.cm'
1179
1180		At the ML toplevel prompt:
1181
1182		CM.autoload "$smlnj/cmb.cm";
1183
1184		There is also a new controller in CM.Control that can be used to turn
1185		off all remaining support for implicit anchors by saying:
1186
1187		CM.autoload "$smlnj/
1188		#set CM.Control.implicit_anchors false;
1189
1190		This causes CM to reject implicitly anchored paths.  This is (for the
1191		time being) less permissive than the "final" version where there will
1192		be no more such implicit anchors and relative paths will be just that:
1193		relative.
1194
1195		The next step (version after next version?) will be to make the
1196		default for CM.Control.implicit_anchors false.  After the dust has
1197		settled, I can then produce the "final" version of this...
1198
1199		Note: Since bootstrapping is a bit tricky, I provided new boot files.
1200
1201		----------------------------------------------------------------------
1202		Name: Matthias Blume
1203		Date: 2000/05/11 16:30:00 JST
1204		Tag: blume-20000511-sources
1205		Description:
1206
1207		The main change is that I added function CM.sources as a generalized
1208		version of the earlier CM.makedepend.  This entails the following
1209		additional changes:
1210
1211		- CM.makedepend has been dropped.
1212
1213		- CM manual has been updated.
1214
1215		- TOOLS signature and API have been changed.
1216
1217		----------------------------------------------------------------------
1218		Name: Allen Leung
1219		Date: 2000/05/10 21:17 EDT
1220		Tag: leunga-20000510-moby-c--ssa
1221		Description:
1222
1223		Various bug fixes and new features for C--, Moby and MLRISC optimizations.
1224		None of these affect SML/NJ.
1225
1226		1. Register Allocation
1227
1228		a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
1229		This module tries to remove local (i.e. basic block level) redundancies
1230		during spilling.
1231
1232		b. A new framework for performing region based register allocation.
1233		Not yet entirely functional.
1234
1235		2. X86
1236
1237		a. DefUse for POP was missing the stack pointer [found by Lal]
1238		b. Reload for CALL was incorrect in X86Spill [found by John]
1239		c. Various fixes in X86Spill so that it can be used correctly for
1240		the new spilling module.
1241
1242		3. SSA/IR
1243
1244		a. New module ir/dj-dataflow.sml implements elimination based
1245		data flow analysis.
1246
1247		4. MLRiscGen
1248
1249		a. Fix for gc type annotation
1250
1251		5. MDGen
1252
1253		Various fixes for machine description -> ml code translation.  For ssa
1254		only.
1255
1256		----------------------------------------------------------------------
1257		Name: Allen Leung
1258		Date: 2000/05/08 22:17 EDT
1259		Tag: leunga-20000508-labexp
1260		Description:
1261
1262		Fermin has found a few assembly problems with constant expressions
1263		generated in LabelExp.  Mostly, the problems involve extra parentheses,
1264		which choke on dumb assemblers.  This is his fix.
1265
1266		----------------------------------------------------------------------
1267		Name: Dave MacQueen
1268		Date: 2000/04/09 14:00 EDT
1269		Tag: dbm-20000502-Version_110_28
1270		Description:
1271
1272		1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
1273
1274		2. Updated config/version to 110.28
1275
1276		3. Updated config/srcarchiveurl
1277
1278		3. New boot files!
1279		ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
1280
1281		----------------------------------------------------------------------
1282		Name: Matthias Blume
1283		Date: 2000/05/01 19:05:00 JST
1284		Tag: blume-20000501-noweb
1285		Description:
1286
1287		A new noweb tool has been added.  The existing system is entirely
1288		unaffected by this, but some CM users have asked for renewed noweb
1289		support.  Everything is documented in the CM manual.
1290
1291		New (plugin) libraries:
1292
1293		noweb-tool.cm
1294		nw-ext.cm
1295
1296		----------------------------------------------------------------------
1297		Name: Dave MacQueen
1298		Date: 2000/04/30 12:40PM EDT
1299		Tag: dbm-20000430-bug_fixes
1300		Description:
1301
1302		1. Fix for bug 1498
1303		smlnj/src/system/Basis/Implementation/Unsafe/object.sig
1304		smlnj/src/system/Basis/Implementation/Unsafe/object.sml
1305		added toRealArray function
1306		smlnj/src/compiler/MiscUtil/print/ppobj.sml
1307		added check for tag Obj.RealArray to array printing case in ppObj
1308
1309		2. Fix for bug 1510
1310		smlnj/src/compiler/Semant/types/typesutil.sml
1311		fixed definition of dummyargs (used by equalTycon) so that
1312		dummy args are distinct types
1313
1314		----------------------------------------------------------------------
1315		Name: Matthias Blume
1316		Date: 2000/04/30 01:00:00 JST
1317		Tag: blume-20000430-versions
1318		Description:
1319
1320		1. CM version numbering added.  This is an implementation of Lal's
1321		proposal for adding version numbers and version checking to .cm
1322		files.  Lal said that his proposal was just that -- a proposal.
1323		For the time being I went ahead and implemented it so that people
1324		can comment on it.  Everything is completely backward-compatible
1325		(except for the stable library format, i.e., new bootfiles!).
1326
1327		As usual, see the CM manual for details.
1328
1329		2. An alternative syntax for anchored paths has been implemented.
1330		Dave has recently voiced the same concerns that I had when I did
1331		this, so there should be some support.  My take is that eventually
1332		I will let support for the current syntax (where anchors are
1333		"implicit") fade out in favor of the new, explicit syntax.
1334		In order to be backward-compatible, both old and new syntax are
1335		currently supported.
1336
1337		Again, see the CM manual for details.
1338
1339		3. Parallel make is trying to be slightly smarter:  When the master
1340		process finds a "bottleneck", i.e., when there is only one
1341		compilation unit that can be compiled and everybody else is
1342		waiting on it, then it will simply compile it directly instead
1343		of clumsily telling one of the slaves to do it.
1344
1345		4. Support for "unsharing" added.  This is necessary in order to be
1346		able to have two different versions of the same library running
1347		at the same time (e.g., for trying out a new MLRISC while still
1348		having the old MLRISC linked into the current compiler, etc.)
1349		See the CM manual.
1350
1351		5. Simple "makedepend" functionality added for generating Makefile
1352		dependency information.  (This is rather crude at the moment.
1353		Expect some changes here in the future.)
1354
1355		6. ".fun" added as a recognized suffix for ML files. Also documented
1356		explicitly in the manual that the fallback behavior (unknown suffix
1357		-> ML file) is not an official feature!
1358
1359		7. Small changes to the pickler for stable libraries.
1360
1361		8. Several internal changes to CM (for cleanup/improvement).
1362
1363
1364		!!!! NEW BINFILES !!!!
1365
1366		----------------------------------------------------------------------
1367		Name: Matthias Blume
1368		Date: 2000/04/28 17:30:00 JST
1369		Tag: blume-20000428-pathconfig
1370		Description:
1371
1372		1. I changed config/install.sh to remove duplicate entries from the
1373		lib/pathconfig file at the end.  Moreover, the final version of
1374		lib/pathconfig is sorted alphabetically.  The same (sorting) is done
1375		in src/system/installml.
1376
1377		2. The config/install.sh script now consistently uses relative
1378		pathnames in lib/pathconfig whenever the anchor is in the lib
1379		directory.  (So far this was true for the libraries that come
1380		pre-compiled and bundled as part of the bootfiles but not for
1381		libraries that are compiled by the script itself.)
1382
1383		----------------------------------------------------------------------
1384		Name: Matthias Blume
1385		Date: 2000/04/26 13:10:00 JST
1386		Tag: blume-20000426-fun_suffix
1387		Description:
1388
1389		Added ".fun" as a recognized file name suffix (for ML code).
1390
1391		----------------------------------------------------------------------
1392	Name: Allen Leung	Name: Allen Leung
1393	Date: 2000/04/25 17:00:00 EST	Date: 2000/04/25 17:00:00 EST
1394	Tag: leunga-20000425-alpha-ra	Tag: leunga-20000425-alpha-ra
#	Line 200	Line 1577
1577
1578	3.  Assembly	3.  Assembly
1579
1580	When generating assemby, resolve the value of client defined constants,	When generating assembly, resolve the value of client defined constants,
1581	instead of generating symbolic values.  This is controlled by the	instead of generating symbolic values.  This is controlled by the
1582	new flag "asm-resolve-constants", which is default to true.	new flag "asm-resolve-constants", which is default to true.
1583
#	Line 223	Line 1600
1600
1601	To this end, I arranged that instead of "structure Core" as "structure	To this end, I arranged that instead of "structure Core" as "structure
1602	_Core" is bound in the pervasive environment.  Core access is done via	_Core" is bound in the pervasive environment.  Core access is done via
1603	_Core (which can never be accidentially rebound because _Core is not a	_Core (which can never be accidentally rebound because _Core is not a
1604	legal surface-syntax symbol).	legal surface-syntax symbol).
1605
1606	The current solution is much cleaner because the core environment is	The current solution is much cleaner because the core environment is
#	Line 233	Line 1610
1610	with dynamic and symbolic parts of the core environment.	with dynamic and symbolic parts of the core environment.
1611
1612	Remaining hackery (to bind the "magic" symbol _Core) is localized in the	Remaining hackery (to bind the "magic" symbol _Core) is localized in the
1613	compilation mananger's bootstrap compiler (actually: in the "init group"	compilation manager's bootstrap compiler (actually: in the "init group"
1614	handling).  See the comments in src/system/smlnj/init/init.cmi for	handling).  See the comments in src/system/smlnj/init/init.cmi for
1615	more details.	more details.
1616
#	Line 348	Line 1725
1725	(specified in the .cm file at each instance where the tool's class is	(specified in the .cm file at each instance where the tool's class is
1726	used).	used).
1727
1728	This was done to accomodate the new "make" and "shell" tools which	This was done to accommodate the new "make" and "shell" tools which
1729	facilitate fairly seemless hookup to portions of code managed using	facilitate fairly seamless hookup to portions of code managed using
1730	Makefiles or Shell scripts.	Makefiles or Shell scripts.
1731
1732	There are no classes "shared" or "private" anymore.  Instead, the	There are no classes "shared" or "private" anymore.  Instead, the
#	Line 361	Line 1738
1738
1739	All existing tools are described in the CM manual.	All existing tools are described in the CM manual.
1740
1741	- Slightly better error handling.  (CM now surpresses many followup	- Slightly better error handling.  (CM now suppresses many followup
1742	error messages that tended to be more annoying than helpful.)	error messages that tended to be more annoying than helpful.)
1743
1744	2. Major changes to the compiler's static environment data structures.	2. Major changes to the compiler's static environment data structures.
#	Line 495	Line 1872
1872
1873	I've changed andl to testl in the floating point test sequence	I've changed andl to testl in the floating point test sequence
1874	whenever appropriate.  The Intel optimization guide states that	whenever appropriate.  The Intel optimization guide states that
1875	testl is perferable to andl.	testl is preferable to andl.
1876
1877	7. RA (x86 only)	7. RA (x86 only)
1878
#	Line 677	Line 2054
2054
2055	1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an	1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
2056	additional argument called "template" which is an optional string that	additional argument called "template" which is an optional string that
2057	specifiel the layout of the tool command line.  See the CM manual for	specifies the layout of the tool command line.  See the CM manual for
2058	explanation.	explanation.
2059
2060	2. A special-purpose tool can be "regisitered" by simply dropping the	2. A special-purpose tool can be "registered" by simply dropping the
2061	corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same	corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
2062	directory where the .cm file lives that uses this tool.  (The	directory where the .cm file lives that uses this tool.  (The
2063	behavior/misfeature until now was to look for the tool description	behavior/misfeature until now was to look for the tool description
#	Line 724	Line 2101
2101	rebuild.  Having sets of the form <base><k>.{bin,boot}.<arch>-unix for	rebuild.  Having sets of the form <base><k>.{bin,boot}.<arch>-unix for
2102	<k>=1,2,... is normally not a good idea when invoking fixpt.  However,	<k>=1,2,... is normally not a good idea when invoking fixpt.  However,
2103	they might be the result of an earlier partial run of fixpt (which	they might be the result of an earlier partial run of fixpt (which
2104	perhaps got accidentially killed).  In this case, fixpt will quickly	perhaps got accidentally killed).  In this case, fixpt will quickly
2105	move through what exists before continuing where it left off earlier,	move through what exists before continuing where it left off earlier,
2106	and, thus, saves a lot of time.	and, thus, saves a lot of time.
2107
#	Line 774	Line 2151
2151	it from that remote directory.	it from that remote directory.
2152	This should simplify installation further:  For machines that have	This should simplify installation further:  For machines that have
2153	access to the internet, just fetch <version>-config.tgz, unpack it,	access to the internet, just fetch <version>-config.tgz, unpack it,
2154	edit config/targets, and go (run config/install.sh).  The scipt will	edit config/targets, and go (run config/install.sh).  The script will
2155	fetch everything else that it might need all by itself.	fetch everything else that it might need all by itself.
2156
2157	For CVS users, this mechanism is not relevant for source archives, but	For CVS users, this mechanism is not relevant for source archives, but

---

Colored Diff Long Colored Diff Unidiff Context Diff Side by Side

Legend: Removed from v.628   changed lines   Added in v.756

---

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