Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 589, Fri Mar 31 16:14:36 2000 UTC	revision 775, Fri Jan 12 01:17:51 2001 UTC
#	Line 11	Line 11
11	Date:	Date:
12	Tag: <post-commit CVS tag>	Tag: <post-commit CVS tag>
13	Description:	Description:
14		----------------------------------------------------------------------
15		Name: Allen Leung
16		Date: Thu Jan 11 21:03:00 EST 2001
17		Tag: leunga-20010111-labexp=mltree
18		Description:
19
20		1.  Removed the type LabelExp and replace it by MLTree.
21		2.  Rewritten mltree-simplify with the pattern matcher tool.
22		3.  There were some bugs in alpha code generator which would break
23		64-bit code generation.
24		4.  Redo the tools to generate code with the
25		5.  The CM files in MLRISC (and in src/system/smlnj/MLRISC)
26		are now generated by perl scripts.
27
28		----------------------------------------------------------------------
29		Name: Matthias Blume
30		Date: 2001/01/10 21:55:00 JST
31		Tag: blume-20010110-rcc
32		Description:
33
34		The RCC stuff now seems to work (but only on the x86).
35		This required hacking of the c-calls interface (and -implementation) in
36		MLRISC.
37
38		Normal compiler users should be unaffected.
39
40		----------------------------------------------------------------------
41		Name: Matthias Blume
42		Date: 2001/01/09 01:20:00 JST
43		Tag: blume-20010109-rcc
44		Description:
45
46		This is a fairly big patch, flushing out a large number of pending
47		changes that I made to my development copy over the last couple of days.
48
49		Of practical relevance at this moment is a workaround for a pickling
50		bug that Allen ran into the other day.  The cause of the bug itself is
51		still unknown and it might be hard to fix it properly, but the
52		workaround has some merits of its own (namely somewhat reducing pickling
53		overhead for certain libraries).  Therefore, I think this solution should
54		be satisfactory at this time.
55
56		The rest of the changes (i.e., the vast majority) has to do with my
57		ongoing efforts of providing direct support for C function calls from
58		ML.  At the moment there is a new primop "RAW_CCALL", typing magic
59		in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
60		case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
61		translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
62		handle RCC.
63
64		The last part (the changes to mlriscGen.sml) are still known to be
65		wrong on the x86 and not implemented on all other architectures.  But
66		the infrastructure is in place. I had to change a few functor
67		signatures in the backend to be able to route the CCalls interface
68		from MLRISC there, and I had to specialize the mltree type (on the
69		x86) to include the necessary extensions. (The extensions themselves
70		were already there and redy to go in MLRISC/x86).
71
72		Everything should be very happy as soon as someone helps me with
73		mlriscGen.sml...
74
75		In any case, nothing of this should matter to anyone as long as the
76		new primop is not being used (which is going to be the case unless you
77		find it where I hid it :). The rest of the compiler is completely
78		unaffected.
79
80		----------------------------------------------------------------------
81		Name: Matthias Blume
82		Date: 2001/01/05 00:30:00 JST
83		Tag: blume-20010105-primops
84		Description:
85
86		Added some experimental support for work that I am doing right now.
87		These changes mostly concern added primops, but there is also a new
88		experimental C library in the runtime system (but currently not enabled
89		anywhere except on Linux/X86).
90
91		In the course of adding primops (and playing with them), I discovered that
92		Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
93		broken.  (Zhong was very right he labeled this stuff as "major gross hack".)
94		To recover, I made type information in INL_PRIM mandatory and changed
95		prim.sml as well as built-in.sml accordingly.  The InLine structure now
96		has complete, correct type information (i.e., no bottom types).
97
98		Since all these changes mean that we need new binfiles, I also bumped the
99		version number to 110.32.1.
100
101		----------------------------------------------------------------------
102		Name: Matthias Blume
103		Date: 2000/12/30 22:10:00 JST
104		Tag: blume-20001230-various
105		Description:
106
107		Added proxy libraries for MLRISC and let MLRISC libraries refer
108		to each other using path anchors.  (See CM manual for explanation.)
109
110		Updated CM documentation.
111
112		Fixed some bugs in CM.
113
114		Implemented "proxy" libraries (= syntactic sugar for CM).
115
116		Added "-quiet" option to makeml and changed runtime system accordingly.
117
118		Added cleanup handler for exportML to reset timers and compiler stats.
119
120		----------------------------------------------------------------------
121		Name: Lal George
122		Date: 2000/12/22 22:22:58 EST 2000
123		Tag: Release_110_32
124		Description:
125
126		Infinite precision used throughout MLRISC.
127		see MLRISC/mltree/machine-int.sig
128
129		----------------------------------------------------------------------
130		Name: Matthias Blume
131		Date: 2000/12/22 23:16:00 JST
132		Tag: blume-20001222-warn
133		Description:
134
135		Corrected wording and formatting of some CM warning message which I
136		broke in my previous patch.
137
138		----------------------------------------------------------------------
139		Name: Matthias Blume
140		Date: 2000/12/22 21:20:00 JST
141		Tag: blume-20001222-anchorenv
142		Description:
143
144		Fixed CM's handling of anchor environments in connection with CMB.make.
145
146		----------------------------------------------------------------------
147		Name: Matthias Blume
148		Date: 2000/12/22 13:15:00 JST
149		Tag: blume-20001222-cleanup
150		Description:
151
152		Removed src/cm/ffi which does not (and did not) belong here.
153
154		----------------------------------------------------------------------
155		Name: Matthias Blume
156		Date: 2000/12/21 23:55:00 JST
157		Tag: blume-20001221-exn
158		Description:
159
160		Probably most important: CM no longer silently swallows all exceptions
161		in the compiler.
162		Plus: some other minor CM changes.  For example, CM now reports some
163		sizes for generated binfiles (code, data, envpickle, lambdapickle).
164
165		----------------------------------------------------------------------
166		Name: Matthias Blume
167		Date: 2000/12/15 00:01:05 JST
168		Tag: blume-20001215-dirtool
169		Description:
170
171		- "dir" tool added.
172		- improvements and cleanup to Tools structure
173		- documentation updates
174
175		----------------------------------------------------------------------
176		Name: Allen Leung
177		Date: Thu Dec 14 03:45:24 EST 2000
178		Description:
179		Tag:  leunga-20001214-int-inf
180		Description:
181
182		In IntInf, added these standard functions, which are missing from our
183		implementation:
184
185		andb : int * int -> int
186		xorb : int * int -> int
187		orb  : int * int -> int
188		notb : int -> int
189		<<   : int * word -> int
190		~>>  : int * word -> int
191
192		Not tested, I hope they are correct.
193
194		----------------------------------------------------------------------
195		Name: Allen Leung
196		Date: Fri Dec  8 19:23:26 EST 2000
197		Description:
198		Tag:  leunga-20001208-nowhere
199		Description:
200
201		Slight improvements to the 'nowhere' tool to handle OR-patterns,
202		to generate better error messages etc.  Plus a brief manual.
203
204		----------------------------------------------------------------------
205		Name: Lal George
206		Date: 2000/12/08 09:54:02 EST 2000
207		Tag: Release_110_31
208		Description:
209
210		- Version 110.31
211		----------------------------------------------------------------------
212		Name: Allen Leung
213		Date: Thu Dec  7 22:01:04 EST 2000
214		Tag:  leunga-20001207-cell-monster-hack
215		Description:
216
217		Major MLRISC internal changes.  Affect all clients.
218		Summary:
219
220		1.  Type CELLS.cell = int is now replaced by a datatype.
221		As a result, the old regmap is now gone.  Almost all interfaces
222		in MLRISC change as a consequence.
223
224		2.  A new brand version of machine description tool (v3.0) that generates
225		modules expecting the new interface.  The old version is removed.
226
227		3.  The RA interface has been further abstracted into two new functors.
228		RISC_RA and X86RA.  These functors have much simpler interfaces.
229		[See also directory MLRISC/demo.]
230
231		4.  Some other new source->source code generation tools are available:
232
233		a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
234		b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
235		I use this tool to generate the peephole optimizers---with the new
236		cell type changes, peephole rules are becoming difficult to write
237		without conditional pattern matching.
238
239		5.  More Intmap -> IntHashTable change.  Previous changes by Matthias didn't
240		cover the entire MLRISC source tree so many things broke.
241
242		6.  CM files have been moved to the subdirectory MLRISC/cm.
243		They are moved because there are a lot of them and they clutter up the
244		root dir.
245
246		7.  More detailed documentation to come...
247
248		NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
249		a makeml -rebuild first.  This is because of other other
250		changes that Matthias has made (see below).
251
252
253		----------------------------------------------------------------------
254		Name: Matthias Blume
255		Date: 2000/11/30 23:12:00 JST
256		Tag: blume-20001130-filereorg
257		Description:
258
259		Some manual updates and some file reorganizations in CM.
260
261		----------------------------------------------------------------------
262		Name: Matthias Blume
263		Date: 2000/11/24 17:45:00 JST
264		Tag: blume-20001124-link
265		Description:
266
267		Drastically improved link traversal code for the case that the dynamic
268		value was already loaded at bootstrap time.  As a result, CM and CMB
269		now both load blazingly fast -- even on a very slow machine.  Also,
270		memory consumption has been further reduced by this.
271
272		Warning: The format of the PIDMAP file has changed.  THerefore, to
273		bootstrap you have to do this:
274
275		1. Run CMB.make
276		2. Make a symbolic link for the boot directory:
277		ln -s sml.boot.ARCH-OS xxx
278		3. "Rebuild" the boot directory:
279		./makeml -boot xxx -rebuild sml ; rm xxx
280		4. Boot normally:
281		./makeml
282
283		----------------------------------------------------------------------
284		Name: Matthias Blume
285		Date: 2000/11/21 21:20:00 JST
286		Tag: blume-20001121-tools
287		Description:
288
289		Continued hacking on autoloading problem -- with success this time.
290		Also changed tool-plugin mechanism.  See new CM manual.
291
292		----------------------------------------------------------------------
293		Name: Matthias Blume
294		Date: 2000/11/19 14:30:00 JST
295		Tag:  blume-20001119-autoload
296		Description:
297
298		Some hacking to make autoloading faster.  Success for CMB, no success
299		so far for CM.  There is a reduced structure CM' that autoloads faster.
300		(This is a temporary, non-documented hack to be eliminated again when
301		the general problem is solved.)
302
303		----------------------------------------------------------------------
304		Name: Matthias Blume
305		Date: 2000/11/17 14:10:00 JST
306		Tag: blume-20001117-pickle-lib
307		Description:
308
309		1. Eliminated comp-lib.cm
310		2. Made pickle-lib.cm
311		3. Eliminated all uses of intset.sml (from comp-lib.cm)
312		4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
313		equivalent constructs from smlnj-lib.cm (INtHashTable).
314		5. Point 4. also goes for those uses of intmap.* in MLRISC.
315		Duplicated intmap modules thrown out.
316		6. Hunted down all duplicated SCC code and replaced it with
317		equivalent stuff (GraphSCCFn from smlnj-lib.cm).
318		7. Rewrote Feedback module.
319		8. Moved sortedlist.sml into viscomp-lib.cm.  Eventually it
320		should be thrown out and equivalent modules from smlnj-lib.cm
321		should be used (IntRedBlackSet, IntListSet, ...).
322
323		Confirmed that compiler compiles to fixpoint.
324
325		----------------------------------------------------------------------
326		Name: Allen Leung
327		Date: 2000/11/10 18:00:00
328		Tag: leunga-20001110-new-x86-fp
329
330		A new x86 floating point code generator has been added.
331		By default this is turned off.  To turn this on, do:
332
333		CM.autoload "$smlnj/compiler.cm";
334		Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
335
336		Changes:
337
338		1.  Changed FTAN to FPTAN so that the assembly output is correct.
339		2.  Changed the extension callback for FTANGENT to generate:
340
341		fptan
342		fstp  %st(0)
343		instead of
344		fptan
345		fstpl ftempmem
346
347		3.  Numerous assembly fixes for x86.
348
349		5.  Cleaned up the machine code output module x86/x86MC.sml and added
350		support for a whole bunch of instructions and addressing modes:
351
352		fadd/fsub/fsubr/fmul/fdiv/fdivr  %st, %st(n)
353		faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp  %st, %st(n)
354		fadd/fsub/fsubr/fmul/fdiv/fdivr  %st(n), %st
355		fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
356		fxch %st(n)
357		fld %st(n)
358		fst %st(n)
359		fst mem
360		fstp %st(n)
361		fucom %st(n)
362		fucomp %st(n)
363
364		All these are now generated when the fast fp mode is turned on.
365
366		6.  Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
367
368		----------------------------------------------------------------------
369		Name: Matthias Blume
370		Date: 2000/11/09 11:20:00 JST
371		Tag: blume-20001109-scc
372		Description:
373
374		Eliminated some code duplication:
375
376		1. Added "where" clause to GraphSCCFn in SML/NJ Library.
377		(Otherwise the functor is useless.)
378		2. Used GraphSCCFn where SCCUtilFun was used previously.
379		3. Got rid of SCCUtilFun (in comp-lib.cm).
380
381		----------------------------------------------------------------------
382		Name: Lal George
383		Date: 2000/11/06 09:02:21 EST 2000
384		Tag: Release_110_30
385		Description:
386
387		- Version 110.30
388		----------------------------------------------------------------------
389		Name: Matthias Blume
390		Date: 2000/11/04 14:45:00
391		Tag: blume-20001104-mlbuild
392		Description:
393
394		- Made ml-build faster on startup.
395		- Documentation fixes.
396
397		----------------------------------------------------------------------
398		Name: Matthias Blume
399		Date: 2000/11/02 17:00:00 JST
400		Tag: blume-20001102-condcomp
401		Description:
402
403		- Small tweaks to pickler -- new BOOTFILES!
404		- Version bumped to 110.29.2.
405		- Added conditional compilation facility to init.cmi (see comment there).
406		----------------------------------------------------------------------
407		Name: Allen Leung
408		Date: 2000/10/23 19:31:00
409		Tag: leunga-20001023-demo-ra
410
411		1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
412		2. Test programs for the graph library updated
413		3. Some new MLRISC demo programs added
414
415		----------------------------------------------------------------------
416		Name: Matthias Blume
417		Date: 2000/08/31 22:15:00 JST
418		Tag: blume-20001017-errmsg
419		Description:
420
421		More error message grief: Where there used to be no messages, there
422		now were some that had bogus error regions.  Fixed.
423
424		----------------------------------------------------------------------
425		Name: Matthias Blume
426		Date: 2000/08/31 17:30:00 JST
427		Tag: blume-20001017-v110p29p1
428		Description:
429
430		I made a version 110.29.1 with new bootfiles.
431
432		Changes:  Modified pickler/unpickler for faster and leaner unpickling.
433		CM documentation changes and a small bugfix in CM's error reporting.
434
435		----------------------------------------------------------------------
436		Name: Lal George
437		Date: 2000/09/27 14:42:35 EDT
438		Tag: george-20000927-nodestatus
439		Description:
440
441		Changed the type of the nodestatus, so that:
442
443		SPILLED(~1)             is now SPILLED
444		SPILLED(m) where m>=0   is now MEMREG(m)
445		SPILLED(s) where s<~1   is now SPILL_LOC(~s)
446
447		----------------------------------------------------------------------
448		Name: Matthias Blume
449		Date: 2000/09/07 14:45:00 JST
450		Tag: blume-20000907-cmerrmsg
451		Description:
452
453		Small tweak to CM to avoid getting ML syntax error messages twice.
454
455		----------------------------------------------------------------------
456		Name: Matthias Blume
457		Date: 2000/08/31 18:00:00 JST
458		Tag: blume-20000831-cvsbootfiles
459		Description:
460
461		New URL for boot files (because the 110.29 files on the BL server do
462		now work correctly with my updated install scripts for yacc and lex).
463
464		----------------------------------------------------------------------
465		Name: Matthias Blume
466		Date: 2000/08/08 12:33:00 JST
467		Tag: blume-20000808-manual
468		Description:
469
470		Tiny update to CM manual.
471
472		----------------------------------------------------------------------
473		Name: Allen Leung
474		Date: 2000/08/7 19:31:00
475		Tag: leunga-20000807-a-whole-bunch-of-stuff
476
477		Moby, C--, SSA, x86, machine descriptions etc.  Should only affect C--
478		and Mobdy.
479
480		1.  x86
481
482		a.  Fixes to peephole module by John and Dan.
483		b.  Assembly fix to SETcc by Allen.
484		c.  Fix to c-call by John.
485		d.  Fix to spilling by John.  (This one deals with the missing FSTPT case)
486		e.  Instruction selection optimization to SETcc as suggested by John.
487
488		For example,
489
490		MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
491
492		should generate:
493
494		MOVL a, x
495		SUBL b, x
496		SHRL 31, x
497
498		2.  IR stuff
499
500		A bunch of new DJ-graph related algorithms added.  These
501		speed up SSA construction.
502
503		3.  SSA + Scheduling
504
505		Added code for SSA and scheduling to the repository
506
507		----------------------------------------------------------------------
508		Name: Lal George
509		Date: 2000/07/27 11:53:14 EDT
510
511		Tag: lal-20000727-linux-ppc
512		Description:
513
514		Made changes to support Linux PPC.
515		p.s. I have confirmation that the 110.29 boot files work fine.
516
517		----------------------------------------------------------------------
518		Name: Matthias Blume
519		Date: 2000/07/27 17:40:00 JST
520		Tag: blume-20000727-scripts
521		Description:
522
523		!!!! WARNING !!!!
524		You must recompile the runtime system!
525		!!!! WARNING !!!!
526
527		This is basically another round of script-enhancements:
528
529		1. sml, ml-build, and ml-makedepend accept options -D and -U to define
530		and undefine CM preprocessor symbols.
531
532		2. ml-build avoids generating a new heap image if it finds that the
533		existing one is still ok.  (The condition is that no ML file had to
534		be recompiled and all ML files are found to be older that the heap
535		file.)
536
537		To make this work smoothly, I also hacked the runtime system as
538		well as SMLofNJ.SysInfo to get access to the heap image suffix
539		(.sparc-solaris, ...) that is currently being used.
540
541		Moreover, the signature of CM.mk_standalone has changed.  See the
542		CM manual.
543
544		3. ml-makedepend accepts additional options -n, -a, and -o.  (See the
545		CM manual for details.)
546
547		4. More CM manual updates:
548		- all of the above has been documented.
549		- there is now a section describing the (CM-related) command line
550		arguments that are accepted by the "sml" command
551
552		----------------------------------------------------------------------
553		Name: Matthias Blume
554		Date: 2000/07/25 16:20:00 JST
555		Tag: blume-20000725-makedepend
556		Description:
557
558		Added a script called ml-makedepend.  This can be used in makefiles
559		for Unix' make in a way very similar to the "makedepend" command for
560		C.
561
562		The script internally uses function CM.sources.
563
564		Synopsis:
565
566		ml-makedepend [-f makefile] cmfile targetname
567
568		The default for the makefile is "makefile" (or "Makefile" should
569		"makefile" not exist).
570
571		ml-makedepend adds a cmfile/targetname-specific section to this
572		makefile (after removing the previous version of this section).  The
573		section contains a single dependency specification with targetname on
574		the LHS (targetname is an arbitrary name), and a list of files derived
575		from the cmfile on the RHS.  Some of the files on the RHS are
576		ARCH/OPSYS-specific.  Therefore, ml-makedepend inserts references to
577		"make" variables $(ARCH) and $(OPSYS) in place of the corresponding
578		path names.  The makefile writer is responsible for making sure that
579		these variables have correct at the time "make" is invoked.
580
581		----------------------------------------------------------------------
582		Name: Matthias Blume
583		Date: 2000/07/22 23:30:00 JST
584		Tag: blume-20000722-urlupdate
585		Description:
586
587		Changed BOOT and config/srcarchiveurl to point to BL server:
588
589		ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
590
591		----------------------------------------------------------------------
592		Name: Matthias Blume
593		Date: 2000/07/18 18:00:00 JST
594		Tag: blume-20000718-Version_110_29
595		Description:
596
597		1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
598
599		2. Updated config/version to 110.29
600
601		3. Updated config/srcarchiveurl
602
603		3. New boot files!
604		ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
605
606		----------------------------------------------------------------------
607		Name: Matthias Blume
608		Date: 2000/07/11 13:58:00 JST
609		Tag: blume-20000711-doctypo
610		Description:
611
612		Fixed a few typos in CM manual.
613
614		----------------------------------------------------------------------
615		Name: Allen Leung
616		Date: 2000/06/15 00:38:00
617		Tag: leunga-20000704-sparc-x86
618
619		1. x86 peephole improvement sp += k; sp -= k => nop  [from John]
620		2. fix to x86 RET bug [found by Dan Grossman]
621		3. sparc assembly bug fix for ticc instructions [found by Fermin]
622
623		Affects c-- and moby only
624
625		----------------------------------------------------------------------
626		Name: Matthias Blume
627		Date: 2000/07/04 15:26:00
628		Tag: blume-20000704-trigger
629		Description:
630
631		1. Improvements to CM manual.
632		2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
633		of getting a back-trace.  The function, when called, raises an
634		internal exception which explicitly carries the full back-trace history,
635		so it is unaffected by any intervening handle-raise pairs ("trivial"
636		or not).  The interactive loop will print that history once it arrives
637		at top level.
638		Short of having all exceptions implicitly carry the full history, the
639		recommended way of using this facility is:
640		- compile your program with instrumentation "on"
641		- run it, when it raises an exception, look at the history
642		- if the history is "cut off" because of some handler, go and modify
643		your program so that it explicitly calls BTrace.trigger
644		- recompile (still instrumented), and rerun; look at the full history
645
646		----------------------------------------------------------------------
647		Name: Matthias Blume
648		Date: 2000/07/03 15:36:00 JST
649		Tag: blume-20000702-manual
650		Description:
651
652		Small corrections and updates to CM manual.
653
654		----------------------------------------------------------------------
655		Name: Matthias Blume
656		Date: 2000/06/29 16:04:00 JST
657		Tag: blume-20000629-yacctool
658		Description:
659
660		Changes:
661
662		1. Class "mlyacc" now takes separate arguments to pass options to
663		generated .sml- and .sig-files independently.
664		2. Corresponding CM manual updates.
665		3. BTrace module now also reports call sites.  (However, for loop clusters
666		it only shows from where the cluster was entered.)  There are associated
667		modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
668
669		----------------------------------------------------------------------
670		Name: Matthias Blume
671		Date: 2000/06/27 16:51:00 JST
672		Tag: blume-20000627-noweb
673		Description:
674
675		Changes:
676
677		1. Implemented "subdir" and "witness" options for noweb tool.
678		This caused some slight internal changes in CM's tool implementation.
679		2. Fixed bug in "tool plugin" mechanism.  This is essentially cleaning
680		some remaining issues from earlier path anchor changes.
681		3. Updated CM manual accordingly.
682
683		4. Changed implementation of back-tracing so that I now consider it
684		ready for prime-time.
685
686		In particular, you don't have to explicitly trigger the back-trace
687		anymore.  Instead, if you are running BTrace-instrumented code and
688		there is an uncaught exception (regardless of whether or not it was
689		raised in instrumented code), the top-level evalloop will print
690		the back-trace.
691
692		Features:
693
694		- Instrumented and uninstrumented code work together seemlessly.
695		(Of course, uninstrumented code is never mentioned in actual
696		back-traces.)
697
698		- Asymptotic time- and space-complexity of instrumented code is
699		equal to that of uninstrumented code.  (This means that
700		tail-recursion is preserved by the instrumentation phase.)
701
702		- Modules whose code has been instrumented in different sessions
703		work together without problem.
704
705		- There is no penalty whatsoever on uninstrumented code.
706
707		- There is no penalty on "raise" expressions, even in
708		instrumented code.
709
710		A potential bug (or perhaps it is a feature, too):
711
712		A back-trace reaches no further than the outermost instrumented
713		non-trivial "raise".  Here, a "trivial" raise is one that is the
714		sole RHS of a "handle" rule.  Thus, back-traces reach trough
715
716		<exp> handle e => raise e
717
718		and even
719
720		<exp> handle Foo => raise Bar
721
722		and, of course, through
723
724		<exp> handle Foo => ...
725
726		if the exception was not Foo.
727
728		Back-traces always reach right through any un-instrumented code
729		including any of its "handle" expressions, trivial or not.
730
731		To try this out, do the following:
732
733		- Erase all existing binfiles for your program.
734		(You may keep binfiles for those modules where you think you
735		definitely don't need back-tracing.)
736		- Turn on back-trace instrumentation:
737		SMLofNJ.Internals.BTrace.mode (SOME true);
738		- Recompile your program.  (I.e., run "CM.make" or "use".)
739		- You may now turn instrumentation off again (if you want):
740		SMLofNJ.Internals.BTrace.mode (SOME false);
741		- Run your program as usual.  If it raises an exception that
742		reaches the interactive toplevel, then a back-trace will
743		automatically be printed.  After that, the toplevel loop
744		will print the exception history as usual.
745
746		----------------------------------------------------------------------
747		Name: Matthias Blume
748		Date: 2000/06/26 09:56:46 JST
749		Tag: blume-20000626-setup
750		Description:
751
752		CM: - setup-parameter to "sml" added; this can be used to run arbitrary
753		ML code before and after compiling a file (e.g., to set compiler
754		flags)
755
756		Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
757		- associated changes to btrace.sml (BTrace instrumentation pass)
758		- cleaner implementation of btimp.sml (BTrace tracing and report
759		module)
760
761		CM manual: * new path encoding documented
762		* description of setup-parameter to "sml" added
763
764		The biggest user-visible change to back-tracing is that it is no
765		longer necessary to compile all traced modules within the same
766		session.  (This was a real limitation.)
767
768		----------------------------------------------------------------------
769		Name: Matthias Blume
770		Date: 2000/06/24 12:40:00 JST
771		Tag: blume-20000624-startup
772		Description:
773
774		Fixes startup slowdown problem.  (I was calling SrcPath.sync a _tad_
775		bit too often -- to put it mildly. :)
776
777		----------------------------------------------------------------------
778		Name: Matthias Blume
779		Date: 2000/06/23 18:20:00 JST
780		Tag: blume-20000623-btrace
781		Description:
782
783		This updates adds a backtrace facility to aid programmers in debugging
784		their programs.  This involves the following changes:
785
786		1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
787		keeping track of the current call stack.  When programs are compiled
788		in a special mode, the compiler will insert calls to these hooks
789		into the user program.
790		"Hook" means that it is possible for different implementations of
791		back-tracing to register themselves (at different times).
792
793		2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
794		as an Absyn.dec->Absyn.dec rewrite.  Normally this phase is turned off.
795		It can be turned on using this call:
796		SMLofNJ.Internals.BTrace.mode (SOME true);
797		Turning it off again:
798		SMLofNJ.Internals.BTrace.mode (SOME false);
799		Querying the current status:
800		SMLofNJ.Internals.BTrace.mode NONE;
801		Annotated programs are about twice as big as normal ones, and they
802		run a factor of 2 to 4 slower with a dummy back-trace plugin (one
803		where all hooks do nothing).  The slowdown with a plugin that is
804		actually useful (such as the one supplied by default) is even greater,
805		but in the case of the default plugin it is still only an constant
806		factor (amortized).
807
808		3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
809		with a sub-structure BTrace for controlling back-tracing.  In particular,
810		the above-mentioned function "mode" controls whether the annotation
811		phase is invoked by the compiler.  Another important function is
812		"trigger": when called it aborts the current execution and causes
813		the top-level loop to print a full back-trace.
814
815		4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
816		for back-tracing.  It keeps track of the dynamic call stack and in
817		addition to that it keeps a partial history at each "level" of that
818		stack.  For example, if a tail-calls b, b tail-calls c, and c tail-calls
819		d and b (at separate times, dynamically), then the report will show:
820
821		GOTO   d
822		/c
823		GOTO  \b
824		CALL   a
825
826		This shows that there was an initial non-tail call of a, then a
827		tail-call to b or c, looping behavior in a cluster of functions that
828		consist of b and c, and then a goto from that cluster (i.e., either from
829		b or from c) to d.
830
831		Note that (depending on the user program) the amount of information
832		that the back-trace module has to keep track of at each level is bounded
833		by a constant.  Thus, the whole implementation has the same asymptotical
834		complexity as the original program (both in space and in time).
835
836		5. compiler/TopLevel/interact/evalloop.sml has been modified to
837		handle the special exception SMLofNJ.Internals.BTrace.BTrace
838		which is raised by the "trigger" function mentioned above.
839
840		Notes on usage:
841
842		- Annotated code works well together with unannotated code:
843		Unannotated calls simply do not show up at all in the backtrace.
844
845		- It is not a good idea to let modules that were annotated during
846		different sessions run at the same time.  This is because the compiler
847		chooses small integers to identify individual functions, and there
848		will be clashes if different modules were compiled in separate sessions.
849		(Nothing will crash, and you will even be told about the clashes, but
850		back-trace information will in general not be useful.)
851
852		- Back-tracing can be confused by callcc and capture.
853
854		- The only way of getting a back-trace right now is to explicitly
855		invoke the "trigger" function from your user program.  Eventually, we
856		should make every exception carry back-trace information (if
857		available).  But since this creates more overhead at "raise"-time
858		(similar to the current exnHistory overhead), I have not yet
859		implemented this.  (The implementation will be rather easy.)  With
860		exceptions carrying back-trace information, this facility will be even
861		more useful because users don't need to modify their programs...
862
863		- While it is possible to compile the compiler with back-trace
864		annotations turned on (I did it to get some confidence in
865		correctness), you must make absolutely sure that core.sml and
866		btimp.sml are compiled WITHOUT annotation!  (core.sml cannot actually
867		be compiled with annotation because there is no core access yet, but
868		if you compile btimp.sml with annotation, then the system will go into
869		an infinite recursion and crash.)
870		Since CM currently does not know about BTrace, the only way to turn
871		annotations on and off for different modules of the compiler is to
872		interrupt CMB.make, change the settings, and re-invoke it.  Of course,
873		this is awkward and clumsy.
874
875		Sample sessions:
876
877		Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
878		- SMLofNJ.Internals.BTrace.mode (SOME true);
879		[autoloading]
880		[autoloading done]
881		val it = false : bool
882		- structure X = struct
883		-     fun main n = let
884		-         fun a (x, 0) = d x
885		-           | a (x, n) = b (x, n - 1)
886		-         and b (x, n) = c (x, n)
887		-         and c (x, n) = a (x, n)
888		-         and d x = e (x, 3)
889		-         and e (x, 0) = f x
890		-           | e (x, n) = e (x, n - 1)
891		-         and f 0 = SMLofNJ.Internals.BTrace.trigger ()
892		-           | f n = n * g (n - 1)
893		-         and g n = a (n, 3)
894		-     in
895		-         f n
896		-     end
897		- end;
898		structure X : sig val main : int -> int end
899		- X.main 3;
900		*** BACK-TRACE ***
901		GOTO   stdIn:4.2-13.20: X.main[2].f
902		GOTO-( stdIn:4.2-13.20: X.main[2].e
903		GOTO   stdIn:4.2-13.20: X.main[2].d
904		/ stdIn:4.2-13.20: X.main[2].a
905		| stdIn:4.2-13.20: X.main[2].b
906		GOTO-\ stdIn:4.2-13.20: X.main[2].c
907		CALL   stdIn:4.2-13.20: X.main[2].g
908		GOTO   stdIn:4.2-13.20: X.main[2].f
909		GOTO-( stdIn:4.2-13.20: X.main[2].e
910		GOTO   stdIn:4.2-13.20: X.main[2].d
911		/ stdIn:4.2-13.20: X.main[2].a
912		| stdIn:4.2-13.20: X.main[2].b
913		GOTO-\ stdIn:4.2-13.20: X.main[2].c
914		CALL   stdIn:4.2-13.20: X.main[2].g
915		GOTO   stdIn:4.2-13.20: X.main[2].f
916		GOTO-( stdIn:4.2-13.20: X.main[2].e
917		GOTO   stdIn:4.2-13.20: X.main[2].d
918		/ stdIn:4.2-13.20: X.main[2].a
919		| stdIn:4.2-13.20: X.main[2].b
920		GOTO-\ stdIn:4.2-13.20: X.main[2].c
921		CALL   stdIn:4.2-13.20: X.main[2].g
922		GOTO   stdIn:4.2-13.20: X.main[2].f
923		CALL   stdIn:2.15-17.4: X.main[2]
924		-
925
926		(Note that because of a FLINt bug the above code currently does not
927		compile without BTrace turned on.)
928
929		Here is another example, using my modified Tiger compiler:
930
931		Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
932		- SMLofNJ.Internals.BTrace.mode (SOME true);
933		[autoloading]
934		[autoloading done]
935		val it = false : bool
936		- CM.make "sources.cm";
937		[autoloading]
938		...
939		[autoloading done]
940		[scanning sources.cm]
941		[parsing (sources.cm):parse.sml]
942		[creating directory CM/SKEL ...]
943		[parsing (sources.cm):tiger.lex.sml]
944		...
945		[wrote CM/sparc-unix/semant.sml]
946		[compiling (sources.cm):main.sml]
947		[wrote CM/sparc-unix/main.sml]
948		[New bindings added.]
949		val it = true : bool
950		- Main.compile ("../testcases/merge.tig", "foo.out");
951		*** BACK-TRACE ***
952		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
953		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
954		CALL   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
955		GOTO   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
956		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
957		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
958		CALL   lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
959		/ lib/semant.sml:411.65-543.8: SemantFun[2].transDec
960		CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
961		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
962		CALL   lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
963		CALL   main.sml:1.18-118.4: Main.compile[2]
964		-
965
966		----------------------------------------------------------------------
967		Name: Matthias Blumen
968		Date: 2000/06/21 18:00:00 JST
969		Tag: blume-20000621-manual
970		Description:
971
972		CM manual update: Path environments documented.
973
974		----------------------------------------------------------------------
975		Name: Matthias Blume
976		Date: 2000/06/19 13:40:00
977		Tag: blume-20000619-manual
978		Description:
979
980		CM manual and system/README update.  This only covers the fact that
981		there are no more implicit anchors.  (Path environments and the "bind"
982		option to "cm" have yet to be documented.)
983
984		----------------------------------------------------------------------
985		Name: Matthias Blume
986		Date: 2000/06/19 11:05:00 JST
987		Tag: blume-20000619-chdir-bugfix
988		Description:
989
990		Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
991
992		----------------------------------------------------------------------
993		Name: Matthias Blume
994		Date: 2000/06/18 22:00:10 JST
995		Tag: blume-20000618-implicit-anchors-really-gone
996		Description:
997
998		I updates the previous HISTORY entry where I forgot to mention that
999		implicit anchors are no longer with us.
1000
1001		The current update also gets rid of the (now useless) controller
1002		CM.Control.implicit_anchors.
1003
1004		----------------------------------------------------------------------
1005		Name: Matthias Blume
1006		Date: 2000/06/16 17:30:00 JST
1007		Tag: blume-20000616-anchorenv
1008		Description:
1009
1010		This patch implements the long anticipated (just kidding :) "anchor
1011		environment" mechanism.  In the course of doing this, I also
1012		re-implemented CM's internal "SrcPath" module from scratch.  The new
1013		one should be more robust in certain boundary cases.  In any case, it
1014		is a lot cleaner than its predecessor (IMHO).
1015
1016		This time, although there is yet another boot file format change, I
1017		kept the unpickler backward-compatible.  As a result, no new bootfiles
1018		are necessary and bootstrapping is straightforward.  (You cannot read
1019		new bootfiles into an old system, but the other way around is no
1020		problem.)
1021
1022		Visible changes:
1023
1024		** 0. Implicit path anchors (without the leading $-symbol) are no
1025		longer recognized at all. This means that such path names are not
1026		illegal either.  For example, the name basis.cm simply refers to a
1027		local file called "basis.cm" (i.e, the name is an ordinary path
1028		relative to .cm-files directory).  Or, to put it differently, only
1029		names that start with $ are anchored paths.
1030
1031		** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
1032		vanished.
1033
1034		John (Reppy) had critizised this as soon as I originally proposed and
1035		implemented it, but at that time I did not really deeply believe
1036		him. :) Now I came full-circle because I need the $<singlearc> syntax
1037		in another place where it cannot be seen as an abbreviation for
1038		$/<singlearc>.  To avoid the confusion, $<singlearc> now means what it
1039		seems to mean (i.e., it "expands" into the corresponding anchor
1040		value).
1041
1042		However, when paths are used as members in CM description files, it
1043		continues to be true that there must be at least another arc after the
1044		anchor.  This is now enforced separately during semantic analysis
1045		(i.e., from a lexical/syntactical point of view, the notation is ok.)
1046
1047		** 2. The "cm" class now accepts an option "bind".  The option's value
1048		is a sub-option list of precisely two items -- one labeled "anchor"
1049		and the other one labeled "value".  As you might expect, "anchor" is
1050		used to specify an anchor name to be bound, and "value" specifies what
1051		the anchor is being bound to.
1052
1053		The value must be a directory name and can be given in either standard
1054		syntax (including the possibility that it is itself an anchored path)
1055		or native syntax.
1056
1057		Examples:
1058
1059		foo.cm (bind:(anchor:bar value:$mystuff/bar))
1060		lib.cm (bind:(anchor:a value:"H:\\x\\y\\z"))  (* only works under windows *)
1061
1062		and so on.
1063
1064		The meaning of this is that the .cm-file will be processed with an
1065		augmented anchor environment where the given anchor(s) is/are bound to
1066		the given values(s).
1067
1068		The rationale for having this feature is this: Suppose you are trying
1069		to use two different (already stable) libraries a.cm and b.cm (that
1070		you perhaps didn't write yourself).  Further, suppose each of these
1071		two libraries internally uses its own auxiliary library $aux/lib.cm.
1072		Normally you would now have a problem because the anchor "lib" can not
1073		be bound to more than one value globally.  Therefore, the project that
1074		uses both a.cm and b.cm must locally redirect the anchor to some other
1075		place:
1076
1077		a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
1078		b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
1079
1080		This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
1081		/usr/lib/smlnj/b-stuff/aux.cm, respectively.
1082
1083		Hard-wiring path names is a bit inflexible (and CM will verbosely warn
1084		you when you do so at the time of CM.stabilize).  Therefore, you can
1085		also use an anchored path as the value:
1086
1087		a.cm (bind:(anchor:lib value:$a-lib))
1088		b.cm (bind:(anchor:lib value:$b-lib))
1089
1090		Now you can globally configure (using the usual CM.Anchor.anchor or
1091		pathconfig machinery) bindings for "a-lib" and "b-lib".  Since "lib"
1092		itself is always locally bound, setting it globally is no longer
1093		meaningful or necessary (but it does not hurt either).  In fact, "lib"
1094		can still be used as a global anchor for separate purposes.  As a
1095		matter of fact, one can locally define "lib" in terms of a global
1096		"lib":
1097
1098		a.cm (bind:(anchor:lib value:$lib/a))
1099		b.cm (bind:(anchor:lib value:$lib/b))
1100
1101		** 3: The encoding of path names has changed.  This affects the way
1102		path names are shown in CM's progress report and also the internal
1103		protocol encoding used for parallel make.
1104
1105		The encoding now uses one or more ':'-separated segments.  Each
1106		segments corresponds to a file that has been specified relative to the
1107		file given by its preceding segment.  The first segment is either
1108		relative to the CWD, absolute, or anchored.  Each segment itself is
1109		basically a Unix pathname; all segments but the first are relative.
1110
1111		Example:
1112
1113		$foo/bar/baz.cm:a/b/c.sml
1114
1115		This path denotes the file bar/a/b/c.sml relative to the directory
1116		denoted by anchor "foo".  Notice that the encoding also includes
1117		baz.cm which is the .cm-file that listed a/b/c.sml.  As usual, such
1118		paths are resolved relative to the .cm-files directory, so baz.cm must
1119		be ignored to get the "real" pathname.
1120
1121		To make this fact more obvious, CM puts the names of such "virtual
1122		arcs" into parentheses when they appear in progress reports. (No
1123		parentheses will appear in the internal protocol encoding.)  Thus,
1124		what you really see is:
1125
1126		$foo/bar/(baz.cm):a/b/c.sml
1127
1128		I find this notation to be much more informative than before.
1129
1130		Another new feature of the encoding is that special characters
1131		including parentheses, colons, (back)slashes, and white space are
1132		written as \ddd (where ddd is the decimal encoding of the character).
1133
1134		*** The CM manual still needs to be updated.
1135
1136		----------------------------------------------------------------------
1137		Name: Allen Leung
1138		Date: 2000/06/15 00:38:00
1139		Tag: leunga-20000615-x86-peephole
1140
1141		x86 Peephole fix by Fermin.  Affects c-- and moby only.
1142
1143		----------------------------------------------------------------------
1144		Name: Matthias Blume
1145		Date: 2000/06/12 11:40:00
1146		Tag: blume-20000612-parmakefix
1147		Description:
1148
1149		More cleanup after changing the file naming scheme: This time I
1150		repaired the parallel make mechanism for CMB.make which I broke earlier.
1151
1152		----------------------------------------------------------------------
1153		Name: Allen Leung
1154		Date: 2000/06/09 01:25:00
1155		Tag: leunga-20000609-various
1156
1157		None of these things should affect normal SML/NJ operations
1158
1159		1. Peephole improvements provided by Fermin (c--)
1160		2. New annotation DEFUSE for adding extra dependence (moby)
1161		3. New X86 LOCK instructions (moby)
1162		4. New machine description language for reservation tables (scheduling)
1163		5. Fixes to various optimization/analysis modules (branch chaining, dominator
1164		trees etc.)
1165		6. I've changed the CM files so that they can work with versions
1166		110.0.6, 110.25 and 110.28
1167
1168		----------------------------------------------------------------------
1169		Name: Matthias Blume
1170		Date: 2000/06/09 12:40:00
1171		Tag: blume-20000609-log
1172		Description:
1173
1174		- Removed all(?) remaining RCS Log entries from sources.
1175
1176		- Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
1177		anchored paths).
1178
1179		----------------------------------------------------------------------
1180		Name: Matthias Blume
1181		Date: 2000/06/07 17:00:00 JST
1182		Tag: blume-20000607-no-implicit-anchors
1183		Description:
1184
1185		1. This update changes the default setting for
1186		CM.Control.implicit_anchors from true to false.  This means that
1187		implicit anchors are no longer permitted by default.  I also tried to
1188		make sure that nothing else still relies on implicit anchors.
1189		(This is the next step on the schedule towards a CM that does not even
1190		have the notion of implicit anchors anymore.)
1191
1192		2. More CM manual updates.
1193
1194		3. I managed to track down and fix the pickling bug I mentioned last
1195		time.  Because of the previously existing workaround, this entails no
1196		immediate practical changes.
1197
1198		----------------------------------------------------------------------
1199		Name: Matthias Blume
1200		Date: 2000/06/06 11:15:00 JST
1201		Tag: blume-20000606-lazierpickle
1202		Description:
1203
1204		!!!! NEW BOOT FILES !!!!
1205
1206		* The main purpose of this update is to make library pickles lazier in
1207		order to reduce the initial space penalty for autoloading a library.
1208		As a result, it is now possible to have $smlnj/compiler.cm
1209		pre-registered.  This should take care of the many complaints or
1210		inquiries about missing structure Compiler.  This required changes to
1211		CM's internal data structures and small tweaks to some algorithms.
1212
1213		As a neat additional effect, it is no longer necessary (for the sake
1214		of lean heap image files) to distinguish between a "minimal" CM and a
1215		"full" CM.  Now, there is only one CM (i.e., the "full" version:
1216		$smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
1217		interactive top level. ($smlnj/cm/minimal.cm is gone.)
1218
1219		To make the life of compiler-hackers easier, "makeml" now also
1220		pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm).  In other
1221		words, after you bootstrap a new sml for the first time, you will not
1222		have to autoload $smlnj/cmb.cm again afterwards.  (The first time
1223		around you will still have to do it, though.)
1224
1225		* A second change consists of major updates to the CM manual.  There
1226		are now several appendices with summary information and also a full
1227		specification of the CM description file syntax.
1228
1229		* In directory src/system I added the script "allcross".  This script
1230		invokes sml and cross-compiles the compiler for all supported
1231		architectures.  (Useful when providing a new set of boot files.)
1232
1233		* There seems to be a latent bug in my "lazy pickles" mechanism.  I
1234		added a small tweak to pickle-util.sml to work around this problem,
1235		but it is not a proper fix yet.  I will investigate further.  (The
1236		effect of the bug was an inflation of library pickle size.)
1237
1238		* Version number increased to 110.28.1 (to avoid compatibility problems).
1239
1240		----------------------------------------------------------------------
1241		Name: Allen Leung
1242		Date: 2000/05/25 17:28 EDT
1243		Tag: leunga-20000525-ra
1244		Description:
1245
1246		Fixed a bug in freezing phase of the register allocator.
1247
1248		----------------------------------------------------------------------
1249		Name: Allen Leung
1250		Date: 2000/05/15 22:53 EDT
1251		Tag: leunga-20000515-alpha-x86-ra
1252		Description:
1253
1254		1. Alpha
1255
1256		Slight cleanup.  Removed the instruction SGNXL
1257
1258		2. X86
1259
1260		Added the following instructions to the instruction set:
1261
1262		ROLx, RORx,
1263		BTx, BTSx, BTLx, BTRx,
1264		XCHGx, and variants with the LOCK prefix
1265
1266		3. Register Allocation
1267
1268		The module ra-rewrite-with-renaming has been improved.
1269
1270		These have no effect on SML/NJ.
1271
1272		----------------------------------------------------------------------
1273		Name: Matthias Blume
1274		Date: 2000/05/15 16:20:00 JST
1275		Tag: blume-20000515-lightrebuild
1276		Description:
1277
1278		1. I added an alternative to "-rebuild" to "makeml".  The difference is
1279		that prior to calling CMB.make' the CM-variable "LIGHT" will be
1280		defined.  In effect, the command will not build any cross-compiler
1281		backends and therefore finish more quickly.
1282
1283		The "fixpt" script also takes a "-light" switch to be able to use
1284		this new facility while compiling for a fixpoint.
1285
1286		2. I replaced all mentions of anchored paths in group owner specifications
1287		with simple relative paths (usually starting with "..").
1288		The rationale is that a library's internal workings should not be
1289		compromised by the lack of some anchor.  (An anchor is necessary
1290		for someone who wants to refer to the library by an anchored path,
1291		but it should not be necessary to build the same library in the first
1292		place.)
1293
1294		3. I changed the way CM's tool mechanism determines the shell command
1295		string used for things like ml-yacc etc. so that it does not break
1296		when CM.Control.implicit_anchors is turned off.
1297
1298		----------------------------------------------------------------------
1299		Name: Matthias Blume
1300		Date: 2000/05/12 18:20:00 JST
1301		Tag: blume-20000512-ml-build
1302		Description:
1303
1304		Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
1305		getting installed properly (by config/install.sh).
1306
1307		----------------------------------------------------------------------
1308		Name: Matthias Blume
1309		Date: 2000/05/12 17:30:00 JST
1310		Tag: blume-20000512-anchors
1311		Description:
1312
1313		!!! NEW BOOT FILES !!!
1314
1315		This change is in preparation of fading out support for "implicitly
1316		anchored path names".  I went through all sources and used the
1317		explicit (and relatively new) $-notation.  See system/README and the
1318		CM manual for more info on this.
1319
1320		I also modified the anchoring scheme for some things such as "smlnj",
1321		"MLRISC", "cm", etc. to take advantage of the fact that explicit
1322		anchors are more expressive: anchor name and first arc do not have to
1323		coincide.  This entails the following user-visible change:
1324
1325		You have to write $smlnj/foo/bar instead of smlnj/foo/bar.  In
1326		particular, when you fire up sml with a command-line argument, say,
1327		e.g.:
1328
1329		sml '$smlnj/cmb.cm'
1330
1331		At the ML toplevel prompt:
1332
1333		CM.autoload "$smlnj/cmb.cm";
1334
1335		There is also a new controller in CM.Control that can be used to turn
1336		off all remaining support for implicit anchors by saying:
1337
1338		CM.autoload "$smlnj/
1339		#set CM.Control.implicit_anchors false;
1340
1341		This causes CM to reject implicitly anchored paths.  This is (for the
1342		time being) less permissive than the "final" version where there will
1343		be no more such implicit anchors and relative paths will be just that:
1344		relative.
1345
1346		The next step (version after next version?) will be to make the
1347		default for CM.Control.implicit_anchors false.  After the dust has
1348		settled, I can then produce the "final" version of this...
1349
1350		Note: Since bootstrapping is a bit tricky, I provided new boot files.
1351
1352		----------------------------------------------------------------------
1353		Name: Matthias Blume
1354		Date: 2000/05/11 16:30:00 JST
1355		Tag: blume-20000511-sources
1356		Description:
1357
1358		The main change is that I added function CM.sources as a generalized
1359		version of the earlier CM.makedepend.  This entails the following
1360		additional changes:
1361
1362		- CM.makedepend has been dropped.
1363
1364		- CM manual has been updated.
1365
1366		- TOOLS signature and API have been changed.
1367
1368		----------------------------------------------------------------------
1369		Name: Allen Leung
1370		Date: 2000/05/10 21:17 EDT
1371		Tag: leunga-20000510-moby-c--ssa
1372		Description:
1373
1374		Various bug fixes and new features for C--, Moby and MLRISC optimizations.
1375		None of these affect SML/NJ.
1376
1377		1. Register Allocation
1378
1379		a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
1380		This module tries to remove local (i.e. basic block level) redundancies
1381		during spilling.
1382
1383		b. A new framework for performing region based register allocation.
1384		Not yet entirely functional.
1385
1386		2. X86
1387
1388		a. DefUse for POP was missing the stack pointer [found by Lal]
1389		b. Reload for CALL was incorrect in X86Spill [found by John]
1390		c. Various fixes in X86Spill so that it can be used correctly for
1391		the new spilling module.
1392
1393		3. SSA/IR
1394
1395		a. New module ir/dj-dataflow.sml implements elimination based
1396		data flow analysis.
1397
1398		4. MLRiscGen
1399
1400		a. Fix for gc type annotation
1401
1402		5. MDGen
1403
1404		Various fixes for machine description -> ml code translation.  For ssa
1405		only.
1406
1407		----------------------------------------------------------------------
1408		Name: Allen Leung
1409		Date: 2000/05/08 22:17 EDT
1410		Tag: leunga-20000508-labexp
1411		Description:
1412
1413		Fermin has found a few assembly problems with constant expressions
1414		generated in LabelExp.  Mostly, the problems involve extra parentheses,
1415		which choke on dumb assemblers.  This is his fix.
1416
1417		----------------------------------------------------------------------
1418		Name: Dave MacQueen
1419		Date: 2000/04/09 14:00 EDT
1420		Tag: dbm-20000502-Version_110_28
1421		Description:
1422
1423		1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
1424
1425		2. Updated config/version to 110.28
1426
1427		3. Updated config/srcarchiveurl
1428
1429		3. New boot files!
1430		ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
1431
1432		----------------------------------------------------------------------
1433		Name: Matthias Blume
1434		Date: 2000/05/01 19:05:00 JST
1435		Tag: blume-20000501-noweb
1436		Description:
1437
1438		A new noweb tool has been added.  The existing system is entirely
1439		unaffected by this, but some CM users have asked for renewed noweb
1440		support.  Everything is documented in the CM manual.
1441
1442		New (plugin) libraries:
1443
1444		noweb-tool.cm
1445		nw-ext.cm
1446
1447		----------------------------------------------------------------------
1448		Name: Dave MacQueen
1449		Date: 2000/04/30 12:40PM EDT
1450		Tag: dbm-20000430-bug_fixes
1451		Description:
1452
1453		1. Fix for bug 1498
1454		smlnj/src/system/Basis/Implementation/Unsafe/object.sig
1455		smlnj/src/system/Basis/Implementation/Unsafe/object.sml
1456		added toRealArray function
1457		smlnj/src/compiler/MiscUtil/print/ppobj.sml
1458		added check for tag Obj.RealArray to array printing case in ppObj
1459
1460		2. Fix for bug 1510
1461		smlnj/src/compiler/Semant/types/typesutil.sml
1462		fixed definition of dummyargs (used by equalTycon) so that
1463		dummy args are distinct types
1464
1465		----------------------------------------------------------------------
1466		Name: Matthias Blume
1467		Date: 2000/04/30 01:00:00 JST
1468		Tag: blume-20000430-versions
1469		Description:
1470
1471		1. CM version numbering added.  This is an implementation of Lal's
1472		proposal for adding version numbers and version checking to .cm
1473		files.  Lal said that his proposal was just that -- a proposal.
1474		For the time being I went ahead and implemented it so that people
1475		can comment on it.  Everything is completely backward-compatible
1476		(except for the stable library format, i.e., new bootfiles!).
1477
1478		As usual, see the CM manual for details.
1479
1480		2. An alternative syntax for anchored paths has been implemented.
1481		Dave has recently voiced the same concerns that I had when I did
1482		this, so there should be some support.  My take is that eventually
1483		I will let support for the current syntax (where anchors are
1484		"implicit") fade out in favor of the new, explicit syntax.
1485		In order to be backward-compatible, both old and new syntax are
1486		currently supported.
1487
1488		Again, see the CM manual for details.
1489
1490		3. Parallel make is trying to be slightly smarter:  When the master
1491		process finds a "bottleneck", i.e., when there is only one
1492		compilation unit that can be compiled and everybody else is
1493		waiting on it, then it will simply compile it directly instead
1494		of clumsily telling one of the slaves to do it.
1495
1496		4. Support for "unsharing" added.  This is necessary in order to be
1497		able to have two different versions of the same library running
1498		at the same time (e.g., for trying out a new MLRISC while still
1499		having the old MLRISC linked into the current compiler, etc.)
1500		See the CM manual.
1501
1502		5. Simple "makedepend" functionality added for generating Makefile
1503		dependency information.  (This is rather crude at the moment.
1504		Expect some changes here in the future.)
1505
1506		6. ".fun" added as a recognized suffix for ML files. Also documented
1507		explicitly in the manual that the fallback behavior (unknown suffix
1508		-> ML file) is not an official feature!
1509
1510		7. Small changes to the pickler for stable libraries.
1511
1512		8. Several internal changes to CM (for cleanup/improvement).
1513
1514
1515		!!!! NEW BINFILES !!!!
1516
1517		----------------------------------------------------------------------
1518		Name: Matthias Blume
1519		Date: 2000/04/28 17:30:00 JST
1520		Tag: blume-20000428-pathconfig
1521		Description:
1522
1523		1. I changed config/install.sh to remove duplicate entries from the
1524		lib/pathconfig file at the end.  Moreover, the final version of
1525		lib/pathconfig is sorted alphabetically.  The same (sorting) is done
1526		in src/system/installml.
1527
1528		2. The config/install.sh script now consistently uses relative
1529		pathnames in lib/pathconfig whenever the anchor is in the lib
1530		directory.  (So far this was true for the libraries that come
1531		pre-compiled and bundled as part of the bootfiles but not for
1532		libraries that are compiled by the script itself.)
1533
1534		----------------------------------------------------------------------
1535		Name: Matthias Blume
1536		Date: 2000/04/26 13:10:00 JST
1537		Tag: blume-20000426-fun_suffix
1538		Description:
1539
1540		Added ".fun" as a recognized file name suffix (for ML code).
1541
1542		----------------------------------------------------------------------
1543		Name: Allen Leung
1544		Date: 2000/04/25 17:00:00 EST
1545		Tag: leunga-20000425-alpha-ra
1546		Description:
1547
1548		1. Alpha
1549
1550		PSEUDOARITH was missing in AlphaRewrite.  This causes an endless loop
1551		in C--.
1552
1553		2. RA
1554
1555		Added a flag "ra-dump-size" to print out the size of the flowgraph
1556		and the interference graph.
1557
1558		----------------------------------------------------------------------
1559		Name: Dave MacQueen
1560		Date: 2000/04/25/
1561		Tag: dbm-20000425-mlyacc_doc_examples
1562		Description:
1563		Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
1564		Updated all three examples in src/ml-yacc/examples to run
1565		under 110.* using CM.make.
1566
1567		----------------------------------------------------------------------
1568		Name: Allen Leung
1569		Date: 2000/04/20 23:04:00 EST
1570		Tag: leunga-20000420-ssa-c---stuff
1571		Description:
1572
1573		This update synchronizes my repository with Yale's.  Most of these
1574		changes, however, do not affect SML/NJ at all (the RA is an exception).
1575
1576		1. Register Allocator
1577
1578		a. An improvement in the interference graph construction:
1579		Given a copy
1580
1581		s <- t
1582
1583		no interference edge between s and t is added for this definition of s.
1584
1585		b. I've added two new spill heuristic modules that Fermin and I developed
1586		(in the new library RA.cm). These are unused in SML/NJ but maybe
1587		useful for others (Moby?)
1588
1589		2. X86
1590
1591		a. Various fixes in the backend provided by Fermin [C--] and Lal.
1592
1593		3. Alpha
1594
1595		a. Added the BSR instruction and code generation that goes with it [C--]
1596		b. Other fixes too numerous to recount provided by Fermin [C--]
1597
1598		4. Regmaps
1599
1600		a. The regmaps are not initialized with the identity physical bindings
1601		at creation time.  This is unneeded.
1602
1603		5. MLRISC Optimizations
1604
1605		a. The DJ-Graph module can now compute the iterated dominance frontiers
1606		intersects with liveness incrementally in linear time! Woohoo!
1607		This is now used in my new SSA construction algorithm.
1608
1609		b. THe branch reorganization module is now smarter about linear chains of
1610		basic blocks.
1611
1612
1613		----------------------------------------------------------------------
1614		Name: Matthias Blume
1615		Date: 2000/04/12 13:52:00 JST
1616		Tag: blume_main_v110p27_1
1617		Description:
1618
1619		Changed install.sh script to handle archive files without version number
1620		and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
1621		name of the boot file archive.
1622
1623		----------------------------------------------------------------------
1624		Name: Dave MacQueen
1625		Date: 2000/04/09 14:00 EDT
1626		Tag: dbm-20000410-Version_110_27
1627		Description:
1628
1629		1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
1630
1631		2. Updated src/config/version to 110.27
1632
1633		3. New boot files!
1634
1635		----------------------------------------------------------------------
1636		Name: Allen Leung
1637		Date: 2000/04/09 19:09:00 EST
1638		Tag: leunga-20000409-misc
1639		Description:
1640
1641		1.  Yet another fix for x86 assembly for idivl, imull, mull and friends.
1642
1643		2.  Miscellaneous improvements to MLRISC (unused in sml/nj)
1644
1645		----------------------------------------------------------------------
1646		Name: Stefan
1647		Date: 2000/04/07 10:00:00 EDT
1648		Tag: monnier-20000406-branch-handling
1649		Description:
1650
1651		Improved handling of branches (mostly those generated from
1652		polymorphic equality), removed switchoff and changed the
1653		default optimization settings (more cpsopt and less flintopt).
1654
1655		----------------------------------------------------------------------
1656		Name: Allen Leung
1657		Date: 2000/04/06 01:30:00 EST
1658		Tag: leunga-20000406-peephole-x86-SSA-2
1659		Description:
1660
1661		Forgot a few files.
1662
1663		----------------------------------------------------------------------
1664		Name: Allen Leung
1665		Date: 2000/04/06 00:36:00 EST
1666		Tag: leunga-20000406-peephole-x86-SSA
1667		Description:
1668
1669		1.  New Peephole code
1670
1671		2.  Minor improvement to X86 instruction selection
1672
1673		3.  Various fixes to SSA and machine description -> code translator
1674
1675		----------------------------------------------------------------------
1676		Name: Matthias Blume
1677		Date: 2000/04/05 12:30:00 JST
1678		Tag: blume_main_v110p26p2_3
1679		Description:
1680
1681		This update just merges three minor cosmetic updates to CM's sources
1682		to get ready for the 110.27 code freeze on Friday.  No functionality
1683		has changed.
1684
1685		----------------------------------------------------------------------
1686		Name: Allen Leung
1687		Date: 2000/04/04 19:39:00 EST
1688		Tag: leunga-20000404-x86-asm
1689		Description:
1690
1691		1.  Fixed a problem in X86 assembly.
1692
1693		Things like
1694
1695		jmp %eax
1696		jmp (%eax)
1697
1698		should be output as
1699
1700		jmp *%eax
1701		jmp *(%eax)
1702
1703		2.  Assembly output
1704
1705		Added a new flag
1706
1707		"asm-indent-copies" (default to false)
1708
1709		When this flag is on, parallel copies will be indented an extra level.
1710
1711		----------------------------------------------------------------------
1712		Name: Allen Leung
1713		Date: 2000/04/04 03:18:00 EST
1714		Tag: leunga-20000404-C--Moby
1715		Description:
1716
1717		All of these fixes are related to C--, Moby, and my own optimization
1718		stuff; so they shouldn't affect SML/NJ.
1719
1720		1.  X86
1721
1722		Various fixes related floating point, and extensions.
1723
1724		2.  Alpha
1725
1726		Some extra patterns related to loads with signed/zero extension
1727		provided by Fermin.
1728
1729		3.  Assembly
1730
1731		When generating assembly, resolve the value of client defined constants,
1732		instead of generating symbolic values.  This is controlled by the
1733		new flag "asm-resolve-constants", which is default to true.
1734
1735		4.  Machine Descriptions
1736
1737		a. The precedence parser was slightly broken when parsing infixr symbols.
1738		b. The type generalizing code had the bound variables reversed, resulting
1739		in a problem during arity raising.
1740		c. Various fixes in machine descriptions.
1741
1742		----------------------------------------------------------------------
1743		Name: Matthias Blume
1744		Date: 2000/04/03 16:05:00 JST
1745		Tag: blume_main_v110p26p2_2
1746		Description:
1747
1748		I eliminated coreEnv from compInfo.  Access to the "Core" structure is
1749		now done via the ordinary static environment that is context to each
1750		compilation unit.
1751
1752		To this end, I arranged that instead of "structure Core" as "structure
1753		_Core" is bound in the pervasive environment.  Core access is done via
1754		_Core (which can never be accidentally rebound because _Core is not a
1755		legal surface-syntax symbol).
1756
1757		The current solution is much cleaner because the core environment is
1758		now simply part of the pervasive environment which is part of every
1759		compilation unit's context anyway.  In particular, this eliminates all
1760		special-case handling that was necessary until now in order to deal
1761		with dynamic and symbolic parts of the core environment.
1762
1763		Remaining hackery (to bind the "magic" symbol _Core) is localized in the
1764		compilation manager's bootstrap compiler (actually: in the "init group"
1765		handling).  See the comments in src/system/smlnj/init/init.cmi for
1766		more details.
1767
1768		I also tried to track down all mentions of "Core" (as string argument
1769		to Symbol.strSymbol) in the compiler and replaced them with a
1770		reference to the new CoreSym.coreSym.  Seems cleaner since the actual
1771		name appears in one place only.
1772
1773		Binfile and bootfile format have not changed, but the switchover from
1774		the old "init.cmi" to the new one is a bit tricky, so I supplied new
1775		bootfiles anyway.
1776
1777		----------------------------------------------------------------------
1778		Name: Allen Leung
1779		Date: 2000/04/02 21:17:00 EST
1780		Tag: leunga-20000402-mltree
1781		Description:
1782
1783		1. Renamed the constructor CALL in MLTREE by popular demand.
1784		2. Added a bunch of files from my repository.  These are currently
1785		used by other non-SMLNJ backends.
1786
1787		----------------------------------------------------------------------
1788		Name: Allen Leung
1789		Date: 2000/03/31 21:15:00 EST
1790		Tag: leunga-20000331-aliasing
1791		Description:
1792
1793		This update contains a rewritten (and hopefully more correct) module
1794		for extracting aliasing information from CPS.
1795
1796		To turn on this feature:
1797
1798		Compiler.Control.CG.memDisambiguate := true
1799
1800		To pretty print the region information with assembly
1801
1802		Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
1803
1804		To control how many levels of aliasing information are printed, use:
1805
1806		Compiler.Control.MLRISC.getInt "points-to-show-level" := n
1807
1808		The default of n is 3.
1809
1810	----------------------------------------------------------------------	----------------------------------------------------------------------
1811	Name: David MacQueen	Name: David MacQueen
#	Line 81	Line 1876
1876	(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
1877	used).	used).
1878
1879	This was done to accomodate the new "make" and "shell" tools which	This was done to accommodate the new "make" and "shell" tools which
1880	facilitate fairly seemless hookup to portions of code managed using	facilitate fairly seamless hookup to portions of code managed using
1881	Makefiles or Shell scripts.	Makefiles or Shell scripts.
1882
1883	There are no classes "shared" or "private" anymore.  Instead, the	There are no classes "shared" or "private" anymore.  Instead, the
#	Line 94	Line 1889
1889
1890	All existing tools are described in the CM manual.	All existing tools are described in the CM manual.
1891
1892	- Slightly better error handling.  (CM now surpresses many followup	- Slightly better error handling.  (CM now suppresses many followup
1893	error messages that tended to be more annoying than helpful.)	error messages that tended to be more annoying than helpful.)
1894
1895	2. Major changes to the compiler's static environment data structures.	2. Major changes to the compiler's static environment data structures.
#	Line 228	Line 2023
2023
2024	I've changed andl to testl in the floating point test sequence	I've changed andl to testl in the floating point test sequence
2025	whenever appropriate.  The Intel optimization guide states that	whenever appropriate.  The Intel optimization guide states that
2026	testl is perferable to andl.	testl is preferable to andl.
2027
2028	7. RA (x86 only)	7. RA (x86 only)
2029
#	Line 410	Line 2205
2205
2206	1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an	1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
2207	additional argument called "template" which is an optional string that	additional argument called "template" which is an optional string that
2208	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
2209	explanation.	explanation.
2210
2211	2. A special-purpose tool can be "regisitered" by simply dropping the	2. A special-purpose tool can be "registered" by simply dropping the
2212	corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same	corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
2213	directory where the .cm file lives that uses this tool.  (The	directory where the .cm file lives that uses this tool.  (The
2214	behavior/misfeature until now was to look for the tool description	behavior/misfeature until now was to look for the tool description
#	Line 457	Line 2252
2252	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
2253	<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,
2254	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
2255	perhaps got accidentially killed).  In this case, fixpt will quickly	perhaps got accidentally killed).  In this case, fixpt will quickly
2256	move through what exists before continuing where it left off earlier,	move through what exists before continuing where it left off earlier,
2257	and, thus, saves a lot of time.	and, thus, saves a lot of time.
2258
#	Line 507	Line 2302
2302	it from that remote directory.	it from that remote directory.
2303	This should simplify installation further:  For machines that have	This should simplify installation further:  For machines that have
2304	access to the internet, just fetch <version>-config.tgz, unpack it,	access to the internet, just fetch <version>-config.tgz, unpack it,
2305	edit config/targets, and go (run config/install.sh).  The scipt will	edit config/targets, and go (run config/install.sh).  The script will
2306	fetch everything else that it might need all by itself.	fetch everything else that it might need all by itself.
2307
2308	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.589   changed lines   Added in v.775

---

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