Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 634, Mon May 1 10:03:50 2000 UTC	revision 815, Fri May 4 05:09:10 2001 UTC
#	Line 12	Line 12
12	Tag: <post-commit CVS tag>	Tag: <post-commit CVS tag>
13	Description:	Description:
14	----------------------------------------------------------------------	----------------------------------------------------------------------
15		Name: Allen Leung
16		Date: 2001/05/04 01:57:00 EDT
17		Tag: leunga-20010504-sync
18		Description:
19
20		MLRISC features.
21
22		1. Fix to CMPXCHG instructions.
23		2. Changed RA interface to allow annotations in callbacks.
24		3. Added a new method to the stream interface to allow annotations updates.
25
26		----------------------------------------------------------------------
27		Name: Matthias Blume
28		Date: 2001/05/01 11:45:00 EDT
29		Tag: blume-20010501-pcedittmp
30		Description:
31
32		Changed install.sh to use the current working directory instead of
33		/usr/tmp for a temporary file (pcedittmp).  The previous choice
34		of /usr/tmp caused trouble with MacOS X because of file premission
35		problems.
36
37		----------------------------------------------------------------------
38		Name: Matthias Blume
39		Date: 2001/04/20 11:10:00 EDT
40		Tag: blume-20010420-inMLflag
41		Description:
42
43		- added vp_limitPtrMask to vproc-state.h
44		(for use by the raw-C-calls mechanism to implement proper interrupt
45		handling)
46		- made the ML compiler aware of various data-structure offsets so it
47		can generate code for accessing the vp_inML flag and vp_limitPtrMask
48		- tweaked mlriscGen.sml to have it emit interrupt-handling code for
49		raw C-calls
50
51		----------------------------------------------------------------------
52		Name: Lal George
53		Date: 2001/04/20 09:15:28 EDT
54		Tag: george-20010420-macosX
55		Description:
56
57		- Changes to port to Mac OS X; Darwin.
58
59		- In the process I found that sqrt was broken on the PPC, because the
60		fsqrt instruction is not implemented.
61
62		----------------------------------------------------------------------
63		Name: Matthias Blume
64		Date: 2001/04/18 12:45:00 EDT
65		Tag: blume-20010418-ccalls
66		Description:
67
68		- fixed two off-by-4 errors in the x86-specific c-calls implementation
69		(this bug prevented structure arguments containing pointers from being
70		passed correctly)
71		- changed the raw-C-call code in mlriscGen.sml in such a way that
72		structure arguments are represented as a pointer to the beginning
73		of the structure (instead of having a series of synthesized arguments,
74		one for each structure member)
75
76		- made makeml script's verbosity level configurable via environment
77		variable (MAKEML_VERBOSITY)
78
79		- eliminated placeholder implementations for f32l, w16s, i16s, and f32s
80		in rawmem-x86.sml; we are now using the real thing
81
82		----------------------------------------------------------------------
83		Name: Matthias Blume
84		Date: 2001/03/22 16:25:00 EST
85		Tag: blume-20010322-bootfiles
86		Description:
87
88		Created a new set of bootfiles (for your automatic installation convenience).
89
90		----------------------------------------------------------------------
91		Name: Matthias Blume
92		Date: 2001/03/22 15:10:00 EST
93		Tag: blume-20010322-rawmem-parcm
94		Description:
95
96		1. All "raw memory access" primitives for the new FFI are implemented now
97		(at least on the x86).
98		2. Some further cleanup of CM's parallel make mechanism.
99
100		----------------------------------------------------------------------
101		Name: Matthias Blume
102		Date: 2001/03/19 17:53:00 EST
103		Tag: blume-20010319-parallel
104		Description:
105
106		Parallel make (using compile servers) now works again.
107
108		To this end, CM.stabilize and CMB.make have been modified to work in
109		two passes when compile servers are attached:
110		1. Compile everything, do not perform stabilization; this pass
111		uses compile servers
112		2. Stabilize everything; this pass does not use compile servers
113		If there are no compile servers, the two passes are combined into one
114		(as before).  Splitting the passes increases the inherent parallelism
115		in the dependency graph because the entire graph including all
116		libraries is available at the same time.  This, in turn, improves
117		server utilization.  The downside is that the master process will
118		have to do some extra work after compilation is done (because for
119		technical reasons it must re-read all the binfiles during stabilization).
120
121		----------------------------------------------------------------------
122		Name: Matthias Blume
123		Date: 2001/03/16 12:22:00 EST
124		Tag: blume-20010316-bootfiles
125		Description:
126
127		Created a new set of bootfiles (for your automatic installation convenience).
128
129		----------------------------------------------------------------------
130		Name: Matthias Blume
131		Date: 2001/03/16 11:00:00 EST
132		Tag: blume-20010316-MLTREE-fixup
133		Description:
134
135		This is a minor fixup for an (untagged) earlier commit by Allen.
136		(A file was missing).
137
138		----------------------------------------------------------------------
139		Name: Allen Leung
140		Date: Mon Mar  5 18:54:57 EST 2001
141		Tag: leunga-20010305-cut-support
142
143		1. New support for alternative control-flow in MLTREE.
144		Currently we support
145
146		FLOW_TO(CALL ...., [k1,...,kn])
147
148		This is needed for 'cuts to' in C-- and try/handle-like constructs
149		in Moby
150
151		New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
152
153		2. Register Allocator
154
155		Changes in interface [from Fermin, John]
156
157		3. Alpha 8-bit SLL support [Fermin]
158
159		4. All architectures
160
161		A new module (ClusterExpandCopies) for expanding parallel copies.
162
163		----------------------------------------------------------------------
164		Name: Allen Leung
165		Date: 2001/02/27 23:07:00 EST
166		Tag: leunga-20010227-minor-stuff
167
168		1. Alpha bug fix for CMOVNE
169		2. Handle mltree COND(..,FCMP ...,...)
170		3. Bug fix in simplifier
171
172		----------------------------------------------------------------------
173		Name: Matthias Blume
174		Date: 2001/01/30 17:50:00 EST
175		Tag: blume-20010130-sync
176		Description:
177
178		This is just a minor update to sync my devel branch with the main brach.
179		The only visible change is the addition of some README files.
180
181		----------------------------------------------------------------------
182		Name: Matthias Blume
183		Date: 2001/01/12 23:30:00 JST
184		Tag: blume-20010112-bootfiles
185		Description:
186
187		Made a new set of bootfiles that goes with the current state of the
188		repository.
189
190		----------------------------------------------------------------------
191		Name: Matthias Blume
192		Date: 2001/01/12 21:20:00 JST
193		Tag: blume-20010112-sync
194		Description:
195
196		I am just flushing out some minor changes that had accumulated in
197		my private branch in order to sync with the main tree.  (This is
198		mainly because I had CVS trouble when trying to merge _into_ my
199		private branch.)
200
201		Most people should be completely unaffected by this.
202
203		----------------------------------------------------------------------
204		Name: Allen Leung
205		Date: Thu Jan 11 21:03:00 EST 2001
206		Tag: leunga-20010111-labexp=mltree
207		Description:
208
209		1.  Removed the type LabelExp and replace it by MLTree.
210		2.  Rewritten mltree-simplify with the pattern matcher tool.
211		3.  There were some bugs in alpha code generator which would break
212		64-bit code generation.
213		4.  Redo the tools to generate code with the
214		5.  The CM files in MLRISC (and in src/system/smlnj/MLRISC)
215		are now generated by perl scripts.
216
217		----------------------------------------------------------------------
218		Name: Matthias Blume
219		Date: 2001/01/10 21:55:00 JST
220		Tag: blume-20010110-rcc
221		Description:
222
223		The RCC stuff now seems to work (but only on the x86).
224		This required hacking of the c-calls interface (and -implementation) in
225		MLRISC.
226
227		Normal compiler users should be unaffected.
228
229		----------------------------------------------------------------------
230		Name: Matthias Blume
231		Date: 2001/01/09 01:20:00 JST
232		Tag: blume-20010109-rcc
233		Description:
234
235		This is a fairly big patch, flushing out a large number of pending
236		changes that I made to my development copy over the last couple of days.
237
238		Of practical relevance at this moment is a workaround for a pickling
239		bug that Allen ran into the other day.  The cause of the bug itself is
240		still unknown and it might be hard to fix it properly, but the
241		workaround has some merits of its own (namely somewhat reducing pickling
242		overhead for certain libraries).  Therefore, I think this solution should
243		be satisfactory at this time.
244
245		The rest of the changes (i.e., the vast majority) has to do with my
246		ongoing efforts of providing direct support for C function calls from
247		ML.  At the moment there is a new primop "RAW_CCALL", typing magic
248		in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
249		case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
250		translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
251		handle RCC.
252
253		The last part (the changes to mlriscGen.sml) are still known to be
254		wrong on the x86 and not implemented on all other architectures.  But
255		the infrastructure is in place. I had to change a few functor
256		signatures in the backend to be able to route the CCalls interface
257		from MLRISC there, and I had to specialize the mltree type (on the
258		x86) to include the necessary extensions. (The extensions themselves
259		were already there and redy to go in MLRISC/x86).
260
261		Everything should be very happy as soon as someone helps me with
262		mlriscGen.sml...
263
264		In any case, nothing of this should matter to anyone as long as the
265		new primop is not being used (which is going to be the case unless you
266		find it where I hid it :). The rest of the compiler is completely
267		unaffected.
268
269		----------------------------------------------------------------------
270		Name: Matthias Blume
271		Date: 2001/01/05 00:30:00 JST
272		Tag: blume-20010105-primops
273		Description:
274
275		Added some experimental support for work that I am doing right now.
276		These changes mostly concern added primops, but there is also a new
277		experimental C library in the runtime system (but currently not enabled
278		anywhere except on Linux/X86).
279
280		In the course of adding primops (and playing with them), I discovered that
281		Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
282		broken.  (Zhong was very right he labeled this stuff as "major gross hack".)
283		To recover, I made type information in INL_PRIM mandatory and changed
284		prim.sml as well as built-in.sml accordingly.  The InLine structure now
285		has complete, correct type information (i.e., no bottom types).
286
287		Since all these changes mean that we need new binfiles, I also bumped the
288		version number to 110.32.1.
289
290		----------------------------------------------------------------------
291		Name: Matthias Blume
292		Date: 2000/12/30 22:10:00 JST
293		Tag: blume-20001230-various
294		Description:
295
296		Added proxy libraries for MLRISC and let MLRISC libraries refer
297		to each other using path anchors.  (See CM manual for explanation.)
298
299		Updated CM documentation.
300
301		Fixed some bugs in CM.
302
303		Implemented "proxy" libraries (= syntactic sugar for CM).
304
305		Added "-quiet" option to makeml and changed runtime system accordingly.
306
307		Added cleanup handler for exportML to reset timers and compiler stats.
308
309		----------------------------------------------------------------------
310		Name: Lal George
311		Date: 2000/12/22 22:22:58 EST 2000
312		Tag: Release_110_32
313		Description:
314
315		Infinite precision used throughout MLRISC.
316		see MLRISC/mltree/machine-int.sig
317
318		----------------------------------------------------------------------
319		Name: Matthias Blume
320		Date: 2000/12/22 23:16:00 JST
321		Tag: blume-20001222-warn
322		Description:
323
324		Corrected wording and formatting of some CM warning message which I
325		broke in my previous patch.
326
327		----------------------------------------------------------------------
328		Name: Matthias Blume
329		Date: 2000/12/22 21:20:00 JST
330		Tag: blume-20001222-anchorenv
331		Description:
332
333		Fixed CM's handling of anchor environments in connection with CMB.make.
334
335		----------------------------------------------------------------------
336		Name: Matthias Blume
337		Date: 2000/12/22 13:15:00 JST
338		Tag: blume-20001222-cleanup
339		Description:
340
341		Removed src/cm/ffi which does not (and did not) belong here.
342
343		----------------------------------------------------------------------
344		Name: Matthias Blume
345		Date: 2000/12/21 23:55:00 JST
346		Tag: blume-20001221-exn
347		Description:
348
349		Probably most important: CM no longer silently swallows all exceptions
350		in the compiler.
351		Plus: some other minor CM changes.  For example, CM now reports some
352		sizes for generated binfiles (code, data, envpickle, lambdapickle).
353
354		----------------------------------------------------------------------
355		Name: Matthias Blume
356		Date: 2000/12/15 00:01:05 JST
357		Tag: blume-20001215-dirtool
358		Description:
359
360		- "dir" tool added.
361		- improvements and cleanup to Tools structure
362		- documentation updates
363
364		----------------------------------------------------------------------
365		Name: Allen Leung
366		Date: Thu Dec 14 03:45:24 EST 2000
367		Description:
368		Tag:  leunga-20001214-int-inf
369		Description:
370
371		In IntInf, added these standard functions, which are missing from our
372		implementation:
373
374		andb : int * int -> int
375		xorb : int * int -> int
376		orb  : int * int -> int
377		notb : int -> int
378		<<   : int * word -> int
379		~>>  : int * word -> int
380
381		Not tested, I hope they are correct.
382
383		----------------------------------------------------------------------
384		Name: Allen Leung
385		Date: Fri Dec  8 19:23:26 EST 2000
386		Description:
387		Tag:  leunga-20001208-nowhere
388		Description:
389
390		Slight improvements to the 'nowhere' tool to handle OR-patterns,
391		to generate better error messages etc.  Plus a brief manual.
392
393		----------------------------------------------------------------------
394		Name: Lal George
395		Date: 2000/12/08 09:54:02 EST 2000
396		Tag: Release_110_31
397		Description:
398
399		- Version 110.31
400		----------------------------------------------------------------------
401		Name: Allen Leung
402		Date: Thu Dec  7 22:01:04 EST 2000
403		Tag:  leunga-20001207-cell-monster-hack
404		Description:
405
406		Major MLRISC internal changes.  Affect all clients.
407		Summary:
408
409		1.  Type CELLS.cell = int is now replaced by a datatype.
410		As a result, the old regmap is now gone.  Almost all interfaces
411		in MLRISC change as a consequence.
412
413		2.  A new brand version of machine description tool (v3.0) that generates
414		modules expecting the new interface.  The old version is removed.
415
416		3.  The RA interface has been further abstracted into two new functors.
417		RISC_RA and X86RA.  These functors have much simpler interfaces.
418		[See also directory MLRISC/demo.]
419
420		4.  Some other new source->source code generation tools are available:
421
422		a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
423		b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
424		I use this tool to generate the peephole optimizers---with the new
425		cell type changes, peephole rules are becoming difficult to write
426		without conditional pattern matching.
427
428		5.  More Intmap -> IntHashTable change.  Previous changes by Matthias didn't
429		cover the entire MLRISC source tree so many things broke.
430
431		6.  CM files have been moved to the subdirectory MLRISC/cm.
432		They are moved because there are a lot of them and they clutter up the
433		root dir.
434
435		7.  More detailed documentation to come...
436
437		NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
438		a makeml -rebuild first.  This is because of other other
439		changes that Matthias has made (see below).
440
441
442		----------------------------------------------------------------------
443		Name: Matthias Blume
444		Date: 2000/11/30 23:12:00 JST
445		Tag: blume-20001130-filereorg
446		Description:
447
448		Some manual updates and some file reorganizations in CM.
449
450		----------------------------------------------------------------------
451		Name: Matthias Blume
452		Date: 2000/11/24 17:45:00 JST
453		Tag: blume-20001124-link
454		Description:
455
456		Drastically improved link traversal code for the case that the dynamic
457		value was already loaded at bootstrap time.  As a result, CM and CMB
458		now both load blazingly fast -- even on a very slow machine.  Also,
459		memory consumption has been further reduced by this.
460
461		Warning: The format of the PIDMAP file has changed.  THerefore, to
462		bootstrap you have to do this:
463
464		1. Run CMB.make
465		2. Make a symbolic link for the boot directory:
466		ln -s sml.boot.ARCH-OS xxx
467		3. "Rebuild" the boot directory:
468		./makeml -boot xxx -rebuild sml ; rm xxx
469		4. Boot normally:
470		./makeml
471
472		----------------------------------------------------------------------
473		Name: Matthias Blume
474		Date: 2000/11/21 21:20:00 JST
475		Tag: blume-20001121-tools
476		Description:
477
478		Continued hacking on autoloading problem -- with success this time.
479		Also changed tool-plugin mechanism.  See new CM manual.
480
481		----------------------------------------------------------------------
482		Name: Matthias Blume
483		Date: 2000/11/19 14:30:00 JST
484		Tag:  blume-20001119-autoload
485		Description:
486
487		Some hacking to make autoloading faster.  Success for CMB, no success
488		so far for CM.  There is a reduced structure CM' that autoloads faster.
489		(This is a temporary, non-documented hack to be eliminated again when
490		the general problem is solved.)
491
492		----------------------------------------------------------------------
493		Name: Matthias Blume
494		Date: 2000/11/17 14:10:00 JST
495		Tag: blume-20001117-pickle-lib
496		Description:
497
498		1. Eliminated comp-lib.cm
499		2. Made pickle-lib.cm
500		3. Eliminated all uses of intset.sml (from comp-lib.cm)
501		4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
502		equivalent constructs from smlnj-lib.cm (INtHashTable).
503		5. Point 4. also goes for those uses of intmap.* in MLRISC.
504		Duplicated intmap modules thrown out.
505		6. Hunted down all duplicated SCC code and replaced it with
506		equivalent stuff (GraphSCCFn from smlnj-lib.cm).
507		7. Rewrote Feedback module.
508		8. Moved sortedlist.sml into viscomp-lib.cm.  Eventually it
509		should be thrown out and equivalent modules from smlnj-lib.cm
510		should be used (IntRedBlackSet, IntListSet, ...).
511
512		Confirmed that compiler compiles to fixpoint.
513
514		----------------------------------------------------------------------
515		Name: Allen Leung
516		Date: 2000/11/10 18:00:00
517		Tag: leunga-20001110-new-x86-fp
518
519		A new x86 floating point code generator has been added.
520		By default this is turned off.  To turn this on, do:
521
522		CM.autoload "$smlnj/compiler.cm";
523		Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
524
525		Changes:
526
527		1.  Changed FTAN to FPTAN so that the assembly output is correct.
528		2.  Changed the extension callback for FTANGENT to generate:
529
530		fptan
531		fstp  %st(0)
532		instead of
533		fptan
534		fstpl ftempmem
535
536		3.  Numerous assembly fixes for x86.
537
538		5.  Cleaned up the machine code output module x86/x86MC.sml and added
539		support for a whole bunch of instructions and addressing modes:
540
541		fadd/fsub/fsubr/fmul/fdiv/fdivr  %st, %st(n)
542		faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp  %st, %st(n)
543		fadd/fsub/fsubr/fmul/fdiv/fdivr  %st(n), %st
544		fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
545		fxch %st(n)
546		fld %st(n)
547		fst %st(n)
548		fst mem
549		fstp %st(n)
550		fucom %st(n)
551		fucomp %st(n)
552
553		All these are now generated when the fast fp mode is turned on.
554
555		6.  Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
556
557		----------------------------------------------------------------------
558		Name: Matthias Blume
559		Date: 2000/11/09 11:20:00 JST
560		Tag: blume-20001109-scc
561		Description:
562
563		Eliminated some code duplication:
564
565		1. Added "where" clause to GraphSCCFn in SML/NJ Library.
566		(Otherwise the functor is useless.)
567		2. Used GraphSCCFn where SCCUtilFun was used previously.
568		3. Got rid of SCCUtilFun (in comp-lib.cm).
569
570		----------------------------------------------------------------------
571		Name: Lal George
572		Date: 2000/11/06 09:02:21 EST 2000
573		Tag: Release_110_30
574		Description:
575
576		- Version 110.30
577		----------------------------------------------------------------------
578		Name: Matthias Blume
579		Date: 2000/11/04 14:45:00
580		Tag: blume-20001104-mlbuild
581		Description:
582
583		- Made ml-build faster on startup.
584		- Documentation fixes.
585
586		----------------------------------------------------------------------
587		Name: Matthias Blume
588		Date: 2000/11/02 17:00:00 JST
589		Tag: blume-20001102-condcomp
590		Description:
591
592		- Small tweaks to pickler -- new BOOTFILES!
593		- Version bumped to 110.29.2.
594		- Added conditional compilation facility to init.cmi (see comment there).
595		----------------------------------------------------------------------
596		Name: Allen Leung
597		Date: 2000/10/23 19:31:00
598		Tag: leunga-20001023-demo-ra
599
600		1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
601		2. Test programs for the graph library updated
602		3. Some new MLRISC demo programs added
603
604		----------------------------------------------------------------------
605		Name: Matthias Blume
606		Date: 2000/08/31 22:15:00 JST
607		Tag: blume-20001017-errmsg
608		Description:
609
610		More error message grief: Where there used to be no messages, there
611		now were some that had bogus error regions.  Fixed.
612
613		----------------------------------------------------------------------
614		Name: Matthias Blume
615		Date: 2000/08/31 17:30:00 JST
616		Tag: blume-20001017-v110p29p1
617		Description:
618
619		I made a version 110.29.1 with new bootfiles.
620
621		Changes:  Modified pickler/unpickler for faster and leaner unpickling.
622		CM documentation changes and a small bugfix in CM's error reporting.
623
624		----------------------------------------------------------------------
625		Name: Lal George
626		Date: 2000/09/27 14:42:35 EDT
627		Tag: george-20000927-nodestatus
628		Description:
629
630		Changed the type of the nodestatus, so that:
631
632		SPILLED(~1)             is now SPILLED
633		SPILLED(m) where m>=0   is now MEMREG(m)
634		SPILLED(s) where s<~1   is now SPILL_LOC(~s)
635
636		----------------------------------------------------------------------
637		Name: Matthias Blume
638		Date: 2000/09/07 14:45:00 JST
639		Tag: blume-20000907-cmerrmsg
640		Description:
641
642		Small tweak to CM to avoid getting ML syntax error messages twice.
643
644		----------------------------------------------------------------------
645		Name: Matthias Blume
646		Date: 2000/08/31 18:00:00 JST
647		Tag: blume-20000831-cvsbootfiles
648		Description:
649
650		New URL for boot files (because the 110.29 files on the BL server do
651		now work correctly with my updated install scripts for yacc and lex).
652
653		----------------------------------------------------------------------
654		Name: Matthias Blume
655		Date: 2000/08/08 12:33:00 JST
656		Tag: blume-20000808-manual
657		Description:
658
659		Tiny update to CM manual.
660
661		----------------------------------------------------------------------
662		Name: Allen Leung
663		Date: 2000/08/7 19:31:00
664		Tag: leunga-20000807-a-whole-bunch-of-stuff
665
666		Moby, C--, SSA, x86, machine descriptions etc.  Should only affect C--
667		and Mobdy.
668
669		1.  x86
670
671		a.  Fixes to peephole module by John and Dan.
672		b.  Assembly fix to SETcc by Allen.
673		c.  Fix to c-call by John.
674		d.  Fix to spilling by John.  (This one deals with the missing FSTPT case)
675		e.  Instruction selection optimization to SETcc as suggested by John.
676
677		For example,
678
679		MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
680
681		should generate:
682
683		MOVL a, x
684		SUBL b, x
685		SHRL 31, x
686
687		2.  IR stuff
688
689		A bunch of new DJ-graph related algorithms added.  These
690		speed up SSA construction.
691
692		3.  SSA + Scheduling
693
694		Added code for SSA and scheduling to the repository
695
696		----------------------------------------------------------------------
697		Name: Lal George
698		Date: 2000/07/27 11:53:14 EDT
699
700		Tag: lal-20000727-linux-ppc
701		Description:
702
703		Made changes to support Linux PPC.
704		p.s. I have confirmation that the 110.29 boot files work fine.
705
706		----------------------------------------------------------------------
707		Name: Matthias Blume
708		Date: 2000/07/27 17:40:00 JST
709		Tag: blume-20000727-scripts
710		Description:
711
712		!!!! WARNING !!!!
713		You must recompile the runtime system!
714		!!!! WARNING !!!!
715
716		This is basically another round of script-enhancements:
717
718		1. sml, ml-build, and ml-makedepend accept options -D and -U to define
719		and undefine CM preprocessor symbols.
720
721		2. ml-build avoids generating a new heap image if it finds that the
722		existing one is still ok.  (The condition is that no ML file had to
723		be recompiled and all ML files are found to be older that the heap
724		file.)
725
726		To make this work smoothly, I also hacked the runtime system as
727		well as SMLofNJ.SysInfo to get access to the heap image suffix
728		(.sparc-solaris, ...) that is currently being used.
729
730		Moreover, the signature of CM.mk_standalone has changed.  See the
731		CM manual.
732
733		3. ml-makedepend accepts additional options -n, -a, and -o.  (See the
734		CM manual for details.)
735
736		4. More CM manual updates:
737		- all of the above has been documented.
738		- there is now a section describing the (CM-related) command line
739		arguments that are accepted by the "sml" command
740
741		----------------------------------------------------------------------
742		Name: Matthias Blume
743		Date: 2000/07/25 16:20:00 JST
744		Tag: blume-20000725-makedepend
745		Description:
746
747		Added a script called ml-makedepend.  This can be used in makefiles
748		for Unix' make in a way very similar to the "makedepend" command for
749		C.
750
751		The script internally uses function CM.sources.
752
753		Synopsis:
754
755		ml-makedepend [-f makefile] cmfile targetname
756
757		The default for the makefile is "makefile" (or "Makefile" should
758		"makefile" not exist).
759
760		ml-makedepend adds a cmfile/targetname-specific section to this
761		makefile (after removing the previous version of this section).  The
762		section contains a single dependency specification with targetname on
763		the LHS (targetname is an arbitrary name), and a list of files derived
764		from the cmfile on the RHS.  Some of the files on the RHS are
765		ARCH/OPSYS-specific.  Therefore, ml-makedepend inserts references to
766		"make" variables $(ARCH) and $(OPSYS) in place of the corresponding
767		path names.  The makefile writer is responsible for making sure that
768		these variables have correct at the time "make" is invoked.
769
770		----------------------------------------------------------------------
771		Name: Matthias Blume
772		Date: 2000/07/22 23:30:00 JST
773		Tag: blume-20000722-urlupdate
774		Description:
775
776		Changed BOOT and config/srcarchiveurl to point to BL server:
777
778		ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
779
780		----------------------------------------------------------------------
781		Name: Matthias Blume
782		Date: 2000/07/18 18:00:00 JST
783		Tag: blume-20000718-Version_110_29
784		Description:
785
786		1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
787
788		2. Updated config/version to 110.29
789
790		3. Updated config/srcarchiveurl
791
792		3. New boot files!
793		ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
794
795		----------------------------------------------------------------------
796		Name: Matthias Blume
797		Date: 2000/07/11 13:58:00 JST
798		Tag: blume-20000711-doctypo
799		Description:
800
801		Fixed a few typos in CM manual.
802
803		----------------------------------------------------------------------
804		Name: Allen Leung
805		Date: 2000/06/15 00:38:00
806		Tag: leunga-20000704-sparc-x86
807
808		1. x86 peephole improvement sp += k; sp -= k => nop  [from John]
809		2. fix to x86 RET bug [found by Dan Grossman]
810		3. sparc assembly bug fix for ticc instructions [found by Fermin]
811
812		Affects c-- and moby only
813
814		----------------------------------------------------------------------
815		Name: Matthias Blume
816		Date: 2000/07/04 15:26:00
817		Tag: blume-20000704-trigger
818		Description:
819
820		1. Improvements to CM manual.
821		2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
822		of getting a back-trace.  The function, when called, raises an
823		internal exception which explicitly carries the full back-trace history,
824		so it is unaffected by any intervening handle-raise pairs ("trivial"
825		or not).  The interactive loop will print that history once it arrives
826		at top level.
827		Short of having all exceptions implicitly carry the full history, the
828		recommended way of using this facility is:
829		- compile your program with instrumentation "on"
830		- run it, when it raises an exception, look at the history
831		- if the history is "cut off" because of some handler, go and modify
832		your program so that it explicitly calls BTrace.trigger
833		- recompile (still instrumented), and rerun; look at the full history
834
835		----------------------------------------------------------------------
836		Name: Matthias Blume
837		Date: 2000/07/03 15:36:00 JST
838		Tag: blume-20000702-manual
839		Description:
840
841		Small corrections and updates to CM manual.
842
843		----------------------------------------------------------------------
844		Name: Matthias Blume
845		Date: 2000/06/29 16:04:00 JST
846		Tag: blume-20000629-yacctool
847		Description:
848
849		Changes:
850
851		1. Class "mlyacc" now takes separate arguments to pass options to
852		generated .sml- and .sig-files independently.
853		2. Corresponding CM manual updates.
854		3. BTrace module now also reports call sites.  (However, for loop clusters
855		it only shows from where the cluster was entered.)  There are associated
856		modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
857
858		----------------------------------------------------------------------
859		Name: Matthias Blume
860		Date: 2000/06/27 16:51:00 JST
861		Tag: blume-20000627-noweb
862		Description:
863
864		Changes:
865
866		1. Implemented "subdir" and "witness" options for noweb tool.
867		This caused some slight internal changes in CM's tool implementation.
868		2. Fixed bug in "tool plugin" mechanism.  This is essentially cleaning
869		some remaining issues from earlier path anchor changes.
870		3. Updated CM manual accordingly.
871
872		4. Changed implementation of back-tracing so that I now consider it
873		ready for prime-time.
874
875		In particular, you don't have to explicitly trigger the back-trace
876		anymore.  Instead, if you are running BTrace-instrumented code and
877		there is an uncaught exception (regardless of whether or not it was
878		raised in instrumented code), the top-level evalloop will print
879		the back-trace.
880
881		Features:
882
883		- Instrumented and uninstrumented code work together seemlessly.
884		(Of course, uninstrumented code is never mentioned in actual
885		back-traces.)
886
887		- Asymptotic time- and space-complexity of instrumented code is
888		equal to that of uninstrumented code.  (This means that
889		tail-recursion is preserved by the instrumentation phase.)
890
891		- Modules whose code has been instrumented in different sessions
892		work together without problem.
893
894		- There is no penalty whatsoever on uninstrumented code.
895
896		- There is no penalty on "raise" expressions, even in
897		instrumented code.
898
899		A potential bug (or perhaps it is a feature, too):
900
901		A back-trace reaches no further than the outermost instrumented
902		non-trivial "raise".  Here, a "trivial" raise is one that is the
903		sole RHS of a "handle" rule.  Thus, back-traces reach trough
904
905		<exp> handle e => raise e
906
907		and even
908
909		<exp> handle Foo => raise Bar
910
911		and, of course, through
912
913		<exp> handle Foo => ...
914
915		if the exception was not Foo.
916
917		Back-traces always reach right through any un-instrumented code
918		including any of its "handle" expressions, trivial or not.
919
920		To try this out, do the following:
921
922		- Erase all existing binfiles for your program.
923		(You may keep binfiles for those modules where you think you
924		definitely don't need back-tracing.)
925		- Turn on back-trace instrumentation:
926		SMLofNJ.Internals.BTrace.mode (SOME true);
927		- Recompile your program.  (I.e., run "CM.make" or "use".)
928		- You may now turn instrumentation off again (if you want):
929		SMLofNJ.Internals.BTrace.mode (SOME false);
930		- Run your program as usual.  If it raises an exception that
931		reaches the interactive toplevel, then a back-trace will
932		automatically be printed.  After that, the toplevel loop
933		will print the exception history as usual.
934
935		----------------------------------------------------------------------
936		Name: Matthias Blume
937		Date: 2000/06/26 09:56:46 JST
938		Tag: blume-20000626-setup
939		Description:
940
941		CM: - setup-parameter to "sml" added; this can be used to run arbitrary
942		ML code before and after compiling a file (e.g., to set compiler
943		flags)
944
945		Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
946		- associated changes to btrace.sml (BTrace instrumentation pass)
947		- cleaner implementation of btimp.sml (BTrace tracing and report
948		module)
949
950		CM manual: * new path encoding documented
951		* description of setup-parameter to "sml" added
952
953		The biggest user-visible change to back-tracing is that it is no
954		longer necessary to compile all traced modules within the same
955		session.  (This was a real limitation.)
956
957		----------------------------------------------------------------------
958		Name: Matthias Blume
959		Date: 2000/06/24 12:40:00 JST
960		Tag: blume-20000624-startup
961		Description:
962
963		Fixes startup slowdown problem.  (I was calling SrcPath.sync a _tad_
964		bit too often -- to put it mildly. :)
965
966		----------------------------------------------------------------------
967		Name: Matthias Blume
968		Date: 2000/06/23 18:20:00 JST
969		Tag: blume-20000623-btrace
970		Description:
971
972		This updates adds a backtrace facility to aid programmers in debugging
973		their programs.  This involves the following changes:
974
975		1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
976		keeping track of the current call stack.  When programs are compiled
977		in a special mode, the compiler will insert calls to these hooks
978		into the user program.
979		"Hook" means that it is possible for different implementations of
980		back-tracing to register themselves (at different times).
981
982		2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
983		as an Absyn.dec->Absyn.dec rewrite.  Normally this phase is turned off.
984		It can be turned on using this call:
985		SMLofNJ.Internals.BTrace.mode (SOME true);
986		Turning it off again:
987		SMLofNJ.Internals.BTrace.mode (SOME false);
988		Querying the current status:
989		SMLofNJ.Internals.BTrace.mode NONE;
990		Annotated programs are about twice as big as normal ones, and they
991		run a factor of 2 to 4 slower with a dummy back-trace plugin (one
992		where all hooks do nothing).  The slowdown with a plugin that is
993		actually useful (such as the one supplied by default) is even greater,
994		but in the case of the default plugin it is still only an constant
995		factor (amortized).
996
997		3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
998		with a sub-structure BTrace for controlling back-tracing.  In particular,
999		the above-mentioned function "mode" controls whether the annotation
1000		phase is invoked by the compiler.  Another important function is
1001		"trigger": when called it aborts the current execution and causes
1002		the top-level loop to print a full back-trace.
1003
1004		4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
1005		for back-tracing.  It keeps track of the dynamic call stack and in
1006		addition to that it keeps a partial history at each "level" of that
1007		stack.  For example, if a tail-calls b, b tail-calls c, and c tail-calls
1008		d and b (at separate times, dynamically), then the report will show:
1009
1010		GOTO   d
1011		/c
1012		GOTO  \b
1013		CALL   a
1014
1015		This shows that there was an initial non-tail call of a, then a
1016		tail-call to b or c, looping behavior in a cluster of functions that
1017		consist of b and c, and then a goto from that cluster (i.e., either from
1018		b or from c) to d.
1019
1020		Note that (depending on the user program) the amount of information
1021		that the back-trace module has to keep track of at each level is bounded
1022		by a constant.  Thus, the whole implementation has the same asymptotical
1023		complexity as the original program (both in space and in time).
1024
1025		5. compiler/TopLevel/interact/evalloop.sml has been modified to
1026		handle the special exception SMLofNJ.Internals.BTrace.BTrace
1027		which is raised by the "trigger" function mentioned above.
1028
1029		Notes on usage:
1030
1031		- Annotated code works well together with unannotated code:
1032		Unannotated calls simply do not show up at all in the backtrace.
1033
1034		- It is not a good idea to let modules that were annotated during
1035		different sessions run at the same time.  This is because the compiler
1036		chooses small integers to identify individual functions, and there
1037		will be clashes if different modules were compiled in separate sessions.
1038		(Nothing will crash, and you will even be told about the clashes, but
1039		back-trace information will in general not be useful.)
1040
1041		- Back-tracing can be confused by callcc and capture.
1042
1043		- The only way of getting a back-trace right now is to explicitly
1044		invoke the "trigger" function from your user program.  Eventually, we
1045		should make every exception carry back-trace information (if
1046		available).  But since this creates more overhead at "raise"-time
1047		(similar to the current exnHistory overhead), I have not yet
1048		implemented this.  (The implementation will be rather easy.)  With
1049		exceptions carrying back-trace information, this facility will be even
1050		more useful because users don't need to modify their programs...
1051
1052		- While it is possible to compile the compiler with back-trace
1053		annotations turned on (I did it to get some confidence in
1054		correctness), you must make absolutely sure that core.sml and
1055		btimp.sml are compiled WITHOUT annotation!  (core.sml cannot actually
1056		be compiled with annotation because there is no core access yet, but
1057		if you compile btimp.sml with annotation, then the system will go into
1058		an infinite recursion and crash.)
1059		Since CM currently does not know about BTrace, the only way to turn
1060		annotations on and off for different modules of the compiler is to
1061		interrupt CMB.make, change the settings, and re-invoke it.  Of course,
1062		this is awkward and clumsy.
1063
1064		Sample sessions:
1065
1066		Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
1067		- SMLofNJ.Internals.BTrace.mode (SOME true);
1068		[autoloading]
1069		[autoloading done]
1070		val it = false : bool
1071		- structure X = struct
1072		-     fun main n = let
1073		-         fun a (x, 0) = d x
1074		-           | a (x, n) = b (x, n - 1)
1075		-         and b (x, n) = c (x, n)
1076		-         and c (x, n) = a (x, n)
1077		-         and d x = e (x, 3)
1078		-         and e (x, 0) = f x
1079		-           | e (x, n) = e (x, n - 1)
1080		-         and f 0 = SMLofNJ.Internals.BTrace.trigger ()
1081		-           | f n = n * g (n - 1)
1082		-         and g n = a (n, 3)
1083		-     in
1084		-         f n
1085		-     end
1086		- end;
1087		structure X : sig val main : int -> int end
1088		- X.main 3;
1089		*** BACK-TRACE ***
1090		GOTO   stdIn:4.2-13.20: X.main[2].f
1091		GOTO-( stdIn:4.2-13.20: X.main[2].e
1092		GOTO   stdIn:4.2-13.20: X.main[2].d
1093		/ stdIn:4.2-13.20: X.main[2].a
1094		| stdIn:4.2-13.20: X.main[2].b
1095		GOTO-\ stdIn:4.2-13.20: X.main[2].c
1096		CALL   stdIn:4.2-13.20: X.main[2].g
1097		GOTO   stdIn:4.2-13.20: X.main[2].f
1098		GOTO-( stdIn:4.2-13.20: X.main[2].e
1099		GOTO   stdIn:4.2-13.20: X.main[2].d
1100		/ stdIn:4.2-13.20: X.main[2].a
1101		| stdIn:4.2-13.20: X.main[2].b
1102		GOTO-\ stdIn:4.2-13.20: X.main[2].c
1103		CALL   stdIn:4.2-13.20: X.main[2].g
1104		GOTO   stdIn:4.2-13.20: X.main[2].f
1105		GOTO-( stdIn:4.2-13.20: X.main[2].e
1106		GOTO   stdIn:4.2-13.20: X.main[2].d
1107		/ stdIn:4.2-13.20: X.main[2].a
1108		| stdIn:4.2-13.20: X.main[2].b
1109		GOTO-\ stdIn:4.2-13.20: X.main[2].c
1110		CALL   stdIn:4.2-13.20: X.main[2].g
1111		GOTO   stdIn:4.2-13.20: X.main[2].f
1112		CALL   stdIn:2.15-17.4: X.main[2]
1113		-
1114
1115		(Note that because of a FLINt bug the above code currently does not
1116		compile without BTrace turned on.)
1117
1118		Here is another example, using my modified Tiger compiler:
1119
1120		Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
1121		- SMLofNJ.Internals.BTrace.mode (SOME true);
1122		[autoloading]
1123		[autoloading done]
1124		val it = false : bool
1125		- CM.make "sources.cm";
1126		[autoloading]
1127		...
1128		[autoloading done]
1129		[scanning sources.cm]
1130		[parsing (sources.cm):parse.sml]
1131		[creating directory CM/SKEL ...]
1132		[parsing (sources.cm):tiger.lex.sml]
1133		...
1134		[wrote CM/sparc-unix/semant.sml]
1135		[compiling (sources.cm):main.sml]
1136		[wrote CM/sparc-unix/main.sml]
1137		[New bindings added.]
1138		val it = true : bool
1139		- Main.compile ("../testcases/merge.tig", "foo.out");
1140		*** BACK-TRACE ***
1141		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
1142		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1143		CALL   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
1144		GOTO   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
1145		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1146		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1147		CALL   lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
1148		/ lib/semant.sml:411.65-543.8: SemantFun[2].transDec
1149		CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
1150		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1151		CALL   lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
1152		CALL   main.sml:1.18-118.4: Main.compile[2]
1153		-
1154
1155		----------------------------------------------------------------------
1156		Name: Matthias Blumen
1157		Date: 2000/06/21 18:00:00 JST
1158		Tag: blume-20000621-manual
1159		Description:
1160
1161		CM manual update: Path environments documented.
1162
1163		----------------------------------------------------------------------
1164		Name: Matthias Blume
1165		Date: 2000/06/19 13:40:00
1166		Tag: blume-20000619-manual
1167		Description:
1168
1169		CM manual and system/README update.  This only covers the fact that
1170		there are no more implicit anchors.  (Path environments and the "bind"
1171		option to "cm" have yet to be documented.)
1172
1173		----------------------------------------------------------------------
1174		Name: Matthias Blume
1175		Date: 2000/06/19 11:05:00 JST
1176		Tag: blume-20000619-chdir-bugfix
1177		Description:
1178
1179		Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
1180
1181		----------------------------------------------------------------------
1182		Name: Matthias Blume
1183		Date: 2000/06/18 22:00:10 JST
1184		Tag: blume-20000618-implicit-anchors-really-gone
1185		Description:
1186
1187		I updates the previous HISTORY entry where I forgot to mention that
1188		implicit anchors are no longer with us.
1189
1190		The current update also gets rid of the (now useless) controller
1191		CM.Control.implicit_anchors.
1192
1193		----------------------------------------------------------------------
1194		Name: Matthias Blume
1195		Date: 2000/06/16 17:30:00 JST
1196		Tag: blume-20000616-anchorenv
1197		Description:
1198
1199		This patch implements the long anticipated (just kidding :) "anchor
1200		environment" mechanism.  In the course of doing this, I also
1201		re-implemented CM's internal "SrcPath" module from scratch.  The new
1202		one should be more robust in certain boundary cases.  In any case, it
1203		is a lot cleaner than its predecessor (IMHO).
1204
1205		This time, although there is yet another boot file format change, I
1206		kept the unpickler backward-compatible.  As a result, no new bootfiles
1207		are necessary and bootstrapping is straightforward.  (You cannot read
1208		new bootfiles into an old system, but the other way around is no
1209		problem.)
1210
1211		Visible changes:
1212
1213		** 0. Implicit path anchors (without the leading $-symbol) are no
1214		longer recognized at all. This means that such path names are not
1215		illegal either.  For example, the name basis.cm simply refers to a
1216		local file called "basis.cm" (i.e, the name is an ordinary path
1217		relative to .cm-files directory).  Or, to put it differently, only
1218		names that start with $ are anchored paths.
1219
1220		** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
1221		vanished.
1222
1223		John (Reppy) had critizised this as soon as I originally proposed and
1224		implemented it, but at that time I did not really deeply believe
1225		him. :) Now I came full-circle because I need the $<singlearc> syntax
1226		in another place where it cannot be seen as an abbreviation for
1227		$/<singlearc>.  To avoid the confusion, $<singlearc> now means what it
1228		seems to mean (i.e., it "expands" into the corresponding anchor
1229		value).
1230
1231		However, when paths are used as members in CM description files, it
1232		continues to be true that there must be at least another arc after the
1233		anchor.  This is now enforced separately during semantic analysis
1234		(i.e., from a lexical/syntactical point of view, the notation is ok.)
1235
1236		** 2. The "cm" class now accepts an option "bind".  The option's value
1237		is a sub-option list of precisely two items -- one labeled "anchor"
1238		and the other one labeled "value".  As you might expect, "anchor" is
1239		used to specify an anchor name to be bound, and "value" specifies what
1240		the anchor is being bound to.
1241
1242		The value must be a directory name and can be given in either standard
1243		syntax (including the possibility that it is itself an anchored path)
1244		or native syntax.
1245
1246		Examples:
1247
1248		foo.cm (bind:(anchor:bar value:$mystuff/bar))
1249		lib.cm (bind:(anchor:a value:"H:\\x\\y\\z"))  (* only works under windows *)
1250
1251		and so on.
1252
1253		The meaning of this is that the .cm-file will be processed with an
1254		augmented anchor environment where the given anchor(s) is/are bound to
1255		the given values(s).
1256
1257		The rationale for having this feature is this: Suppose you are trying
1258		to use two different (already stable) libraries a.cm and b.cm (that
1259		you perhaps didn't write yourself).  Further, suppose each of these
1260		two libraries internally uses its own auxiliary library $aux/lib.cm.
1261		Normally you would now have a problem because the anchor "lib" can not
1262		be bound to more than one value globally.  Therefore, the project that
1263		uses both a.cm and b.cm must locally redirect the anchor to some other
1264		place:
1265
1266		a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
1267		b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
1268
1269		This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
1270		/usr/lib/smlnj/b-stuff/aux.cm, respectively.
1271
1272		Hard-wiring path names is a bit inflexible (and CM will verbosely warn
1273		you when you do so at the time of CM.stabilize).  Therefore, you can
1274		also use an anchored path as the value:
1275
1276		a.cm (bind:(anchor:lib value:$a-lib))
1277		b.cm (bind:(anchor:lib value:$b-lib))
1278
1279		Now you can globally configure (using the usual CM.Anchor.anchor or
1280		pathconfig machinery) bindings for "a-lib" and "b-lib".  Since "lib"
1281		itself is always locally bound, setting it globally is no longer
1282		meaningful or necessary (but it does not hurt either).  In fact, "lib"
1283		can still be used as a global anchor for separate purposes.  As a
1284		matter of fact, one can locally define "lib" in terms of a global
1285		"lib":
1286
1287		a.cm (bind:(anchor:lib value:$lib/a))
1288		b.cm (bind:(anchor:lib value:$lib/b))
1289
1290		** 3: The encoding of path names has changed.  This affects the way
1291		path names are shown in CM's progress report and also the internal
1292		protocol encoding used for parallel make.
1293
1294		The encoding now uses one or more ':'-separated segments.  Each
1295		segments corresponds to a file that has been specified relative to the
1296		file given by its preceding segment.  The first segment is either
1297		relative to the CWD, absolute, or anchored.  Each segment itself is
1298		basically a Unix pathname; all segments but the first are relative.
1299
1300		Example:
1301
1302		$foo/bar/baz.cm:a/b/c.sml
1303
1304		This path denotes the file bar/a/b/c.sml relative to the directory
1305		denoted by anchor "foo".  Notice that the encoding also includes
1306		baz.cm which is the .cm-file that listed a/b/c.sml.  As usual, such
1307		paths are resolved relative to the .cm-files directory, so baz.cm must
1308		be ignored to get the "real" pathname.
1309
1310		To make this fact more obvious, CM puts the names of such "virtual
1311		arcs" into parentheses when they appear in progress reports. (No
1312		parentheses will appear in the internal protocol encoding.)  Thus,
1313		what you really see is:
1314
1315		$foo/bar/(baz.cm):a/b/c.sml
1316
1317		I find this notation to be much more informative than before.
1318
1319		Another new feature of the encoding is that special characters
1320		including parentheses, colons, (back)slashes, and white space are
1321		written as \ddd (where ddd is the decimal encoding of the character).
1322
1323		*** The CM manual still needs to be updated.
1324
1325		----------------------------------------------------------------------
1326		Name: Allen Leung
1327		Date: 2000/06/15 00:38:00
1328		Tag: leunga-20000615-x86-peephole
1329
1330		x86 Peephole fix by Fermin.  Affects c-- and moby only.
1331
1332		----------------------------------------------------------------------
1333		Name: Matthias Blume
1334		Date: 2000/06/12 11:40:00
1335		Tag: blume-20000612-parmakefix
1336		Description:
1337
1338		More cleanup after changing the file naming scheme: This time I
1339		repaired the parallel make mechanism for CMB.make which I broke earlier.
1340
1341		----------------------------------------------------------------------
1342		Name: Allen Leung
1343		Date: 2000/06/09 01:25:00
1344		Tag: leunga-20000609-various
1345
1346		None of these things should affect normal SML/NJ operations
1347
1348		1. Peephole improvements provided by Fermin (c--)
1349		2. New annotation DEFUSE for adding extra dependence (moby)
1350		3. New X86 LOCK instructions (moby)
1351		4. New machine description language for reservation tables (scheduling)
1352		5. Fixes to various optimization/analysis modules (branch chaining, dominator
1353		trees etc.)
1354		6. I've changed the CM files so that they can work with versions
1355		110.0.6, 110.25 and 110.28
1356
1357		----------------------------------------------------------------------
1358		Name: Matthias Blume
1359		Date: 2000/06/09 12:40:00
1360		Tag: blume-20000609-log
1361		Description:
1362
1363		- Removed all(?) remaining RCS Log entries from sources.
1364
1365		- Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
1366		anchored paths).
1367
1368		----------------------------------------------------------------------
1369		Name: Matthias Blume
1370		Date: 2000/06/07 17:00:00 JST
1371		Tag: blume-20000607-no-implicit-anchors
1372		Description:
1373
1374		1. This update changes the default setting for
1375		CM.Control.implicit_anchors from true to false.  This means that
1376		implicit anchors are no longer permitted by default.  I also tried to
1377		make sure that nothing else still relies on implicit anchors.
1378		(This is the next step on the schedule towards a CM that does not even
1379		have the notion of implicit anchors anymore.)
1380
1381		2. More CM manual updates.
1382
1383		3. I managed to track down and fix the pickling bug I mentioned last
1384		time.  Because of the previously existing workaround, this entails no
1385		immediate practical changes.
1386
1387		----------------------------------------------------------------------
1388		Name: Matthias Blume
1389		Date: 2000/06/06 11:15:00 JST
1390		Tag: blume-20000606-lazierpickle
1391		Description:
1392
1393		!!!! NEW BOOT FILES !!!!
1394
1395		* The main purpose of this update is to make library pickles lazier in
1396		order to reduce the initial space penalty for autoloading a library.
1397		As a result, it is now possible to have $smlnj/compiler.cm
1398		pre-registered.  This should take care of the many complaints or
1399		inquiries about missing structure Compiler.  This required changes to
1400		CM's internal data structures and small tweaks to some algorithms.
1401
1402		As a neat additional effect, it is no longer necessary (for the sake
1403		of lean heap image files) to distinguish between a "minimal" CM and a
1404		"full" CM.  Now, there is only one CM (i.e., the "full" version:
1405		$smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
1406		interactive top level. ($smlnj/cm/minimal.cm is gone.)
1407
1408		To make the life of compiler-hackers easier, "makeml" now also
1409		pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm).  In other
1410		words, after you bootstrap a new sml for the first time, you will not
1411		have to autoload $smlnj/cmb.cm again afterwards.  (The first time
1412		around you will still have to do it, though.)
1413
1414		* A second change consists of major updates to the CM manual.  There
1415		are now several appendices with summary information and also a full
1416		specification of the CM description file syntax.
1417
1418		* In directory src/system I added the script "allcross".  This script
1419		invokes sml and cross-compiles the compiler for all supported
1420		architectures.  (Useful when providing a new set of boot files.)
1421
1422		* There seems to be a latent bug in my "lazy pickles" mechanism.  I
1423		added a small tweak to pickle-util.sml to work around this problem,
1424		but it is not a proper fix yet.  I will investigate further.  (The
1425		effect of the bug was an inflation of library pickle size.)
1426
1427		* Version number increased to 110.28.1 (to avoid compatibility problems).
1428
1429		----------------------------------------------------------------------
1430		Name: Allen Leung
1431		Date: 2000/05/25 17:28 EDT
1432		Tag: leunga-20000525-ra
1433		Description:
1434
1435		Fixed a bug in freezing phase of the register allocator.
1436
1437		----------------------------------------------------------------------
1438		Name: Allen Leung
1439		Date: 2000/05/15 22:53 EDT
1440		Tag: leunga-20000515-alpha-x86-ra
1441		Description:
1442
1443		1. Alpha
1444
1445		Slight cleanup.  Removed the instruction SGNXL
1446
1447		2. X86
1448
1449		Added the following instructions to the instruction set:
1450
1451		ROLx, RORx,
1452		BTx, BTSx, BTLx, BTRx,
1453		XCHGx, and variants with the LOCK prefix
1454
1455		3. Register Allocation
1456
1457		The module ra-rewrite-with-renaming has been improved.
1458
1459		These have no effect on SML/NJ.
1460
1461		----------------------------------------------------------------------
1462		Name: Matthias Blume
1463		Date: 2000/05/15 16:20:00 JST
1464		Tag: blume-20000515-lightrebuild
1465		Description:
1466
1467		1. I added an alternative to "-rebuild" to "makeml".  The difference is
1468		that prior to calling CMB.make' the CM-variable "LIGHT" will be
1469		defined.  In effect, the command will not build any cross-compiler
1470		backends and therefore finish more quickly.
1471
1472		The "fixpt" script also takes a "-light" switch to be able to use
1473		this new facility while compiling for a fixpoint.
1474
1475		2. I replaced all mentions of anchored paths in group owner specifications
1476		with simple relative paths (usually starting with "..").
1477		The rationale is that a library's internal workings should not be
1478		compromised by the lack of some anchor.  (An anchor is necessary
1479		for someone who wants to refer to the library by an anchored path,
1480		but it should not be necessary to build the same library in the first
1481		place.)
1482
1483		3. I changed the way CM's tool mechanism determines the shell command
1484		string used for things like ml-yacc etc. so that it does not break
1485		when CM.Control.implicit_anchors is turned off.
1486
1487		----------------------------------------------------------------------
1488		Name: Matthias Blume
1489		Date: 2000/05/12 18:20:00 JST
1490		Tag: blume-20000512-ml-build
1491		Description:
1492
1493		Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
1494		getting installed properly (by config/install.sh).
1495
1496		----------------------------------------------------------------------
1497		Name: Matthias Blume
1498		Date: 2000/05/12 17:30:00 JST
1499		Tag: blume-20000512-anchors
1500		Description:
1501
1502		!!! NEW BOOT FILES !!!
1503
1504		This change is in preparation of fading out support for "implicitly
1505		anchored path names".  I went through all sources and used the
1506		explicit (and relatively new) $-notation.  See system/README and the
1507		CM manual for more info on this.
1508
1509		I also modified the anchoring scheme for some things such as "smlnj",
1510		"MLRISC", "cm", etc. to take advantage of the fact that explicit
1511		anchors are more expressive: anchor name and first arc do not have to
1512		coincide.  This entails the following user-visible change:
1513
1514		You have to write $smlnj/foo/bar instead of smlnj/foo/bar.  In
1515		particular, when you fire up sml with a command-line argument, say,
1516		e.g.:
1517
1518		sml '$smlnj/cmb.cm'
1519
1520		At the ML toplevel prompt:
1521
1522		CM.autoload "$smlnj/cmb.cm";
1523
1524		There is also a new controller in CM.Control that can be used to turn
1525		off all remaining support for implicit anchors by saying:
1526
1527		CM.autoload "$smlnj/
1528		#set CM.Control.implicit_anchors false;
1529
1530		This causes CM to reject implicitly anchored paths.  This is (for the
1531		time being) less permissive than the "final" version where there will
1532		be no more such implicit anchors and relative paths will be just that:
1533		relative.
1534
1535		The next step (version after next version?) will be to make the
1536		default for CM.Control.implicit_anchors false.  After the dust has
1537		settled, I can then produce the "final" version of this...
1538
1539		Note: Since bootstrapping is a bit tricky, I provided new boot files.
1540
1541		----------------------------------------------------------------------
1542		Name: Matthias Blume
1543		Date: 2000/05/11 16:30:00 JST
1544		Tag: blume-20000511-sources
1545		Description:
1546
1547		The main change is that I added function CM.sources as a generalized
1548		version of the earlier CM.makedepend.  This entails the following
1549		additional changes:
1550
1551		- CM.makedepend has been dropped.
1552
1553		- CM manual has been updated.
1554
1555		- TOOLS signature and API have been changed.
1556
1557		----------------------------------------------------------------------
1558		Name: Allen Leung
1559		Date: 2000/05/10 21:17 EDT
1560		Tag: leunga-20000510-moby-c--ssa
1561		Description:
1562
1563		Various bug fixes and new features for C--, Moby and MLRISC optimizations.
1564		None of these affect SML/NJ.
1565
1566		1. Register Allocation
1567
1568		a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
1569		This module tries to remove local (i.e. basic block level) redundancies
1570		during spilling.
1571
1572		b. A new framework for performing region based register allocation.
1573		Not yet entirely functional.
1574
1575		2. X86
1576
1577		a. DefUse for POP was missing the stack pointer [found by Lal]
1578		b. Reload for CALL was incorrect in X86Spill [found by John]
1579		c. Various fixes in X86Spill so that it can be used correctly for
1580		the new spilling module.
1581
1582		3. SSA/IR
1583
1584		a. New module ir/dj-dataflow.sml implements elimination based
1585		data flow analysis.
1586
1587		4. MLRiscGen
1588
1589		a. Fix for gc type annotation
1590
1591		5. MDGen
1592
1593		Various fixes for machine description -> ml code translation.  For ssa
1594		only.
1595
1596		----------------------------------------------------------------------
1597		Name: Allen Leung
1598		Date: 2000/05/08 22:17 EDT
1599		Tag: leunga-20000508-labexp
1600		Description:
1601
1602		Fermin has found a few assembly problems with constant expressions
1603		generated in LabelExp.  Mostly, the problems involve extra parentheses,
1604		which choke on dumb assemblers.  This is his fix.
1605
1606		----------------------------------------------------------------------
1607		Name: Dave MacQueen
1608		Date: 2000/04/09 14:00 EDT
1609		Tag: dbm-20000502-Version_110_28
1610		Description:
1611
1612		1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
1613
1614		2. Updated config/version to 110.28
1615
1616		3. Updated config/srcarchiveurl
1617
1618		3. New boot files!
1619		ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
1620
1621		----------------------------------------------------------------------
1622	Name: Matthias Blume	Name: Matthias Blume
1623	Date: 2000/05/01 19:05:00 JST	Date: 2000/05/01 19:05:00 JST
1624	Tag: blume-20000501-noweb	Tag: blume-20000501-noweb
#	Line 310	Line 1917
1917
1918	3.  Assembly	3.  Assembly
1919
1920	When generating assemby, resolve the value of client defined constants,	When generating assembly, resolve the value of client defined constants,
1921	instead of generating symbolic values.  This is controlled by the	instead of generating symbolic values.  This is controlled by the
1922	new flag "asm-resolve-constants", which is default to true.	new flag "asm-resolve-constants", which is default to true.
1923
#	Line 333	Line 1940
1940
1941	To this end, I arranged that instead of "structure Core" as "structure	To this end, I arranged that instead of "structure Core" as "structure
1942	_Core" is bound in the pervasive environment.  Core access is done via	_Core" is bound in the pervasive environment.  Core access is done via
1943	_Core (which can never be accidentially rebound because _Core is not a	_Core (which can never be accidentally rebound because _Core is not a
1944	legal surface-syntax symbol).	legal surface-syntax symbol).
1945
1946	The current solution is much cleaner because the core environment is	The current solution is much cleaner because the core environment is
#	Line 343	Line 1950
1950	with dynamic and symbolic parts of the core environment.	with dynamic and symbolic parts of the core environment.
1951
1952	Remaining hackery (to bind the "magic" symbol _Core) is localized in the	Remaining hackery (to bind the "magic" symbol _Core) is localized in the
1953	compilation mananger's bootstrap compiler (actually: in the "init group"	compilation manager's bootstrap compiler (actually: in the "init group"
1954	handling).  See the comments in src/system/smlnj/init/init.cmi for	handling).  See the comments in src/system/smlnj/init/init.cmi for
1955	more details.	more details.
1956
#	Line 458	Line 2065
2065	(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
2066	used).	used).
2067
2068	This was done to accomodate the new "make" and "shell" tools which	This was done to accommodate the new "make" and "shell" tools which
2069	facilitate fairly seemless hookup to portions of code managed using	facilitate fairly seamless hookup to portions of code managed using
2070	Makefiles or Shell scripts.	Makefiles or Shell scripts.
2071
2072	There are no classes "shared" or "private" anymore.  Instead, the	There are no classes "shared" or "private" anymore.  Instead, the
#	Line 471	Line 2078
2078
2079	All existing tools are described in the CM manual.	All existing tools are described in the CM manual.
2080
2081	- Slightly better error handling.  (CM now surpresses many followup	- Slightly better error handling.  (CM now suppresses many followup
2082	error messages that tended to be more annoying than helpful.)	error messages that tended to be more annoying than helpful.)
2083
2084	2. Major changes to the compiler's static environment data structures.	2. Major changes to the compiler's static environment data structures.
#	Line 605	Line 2212
2212
2213	I've changed andl to testl in the floating point test sequence	I've changed andl to testl in the floating point test sequence
2214	whenever appropriate.  The Intel optimization guide states that	whenever appropriate.  The Intel optimization guide states that
2215	testl is perferable to andl.	testl is preferable to andl.
2216
2217	7. RA (x86 only)	7. RA (x86 only)
2218
#	Line 787	Line 2394
2394
2395	1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an	1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
2396	additional argument called "template" which is an optional string that	additional argument called "template" which is an optional string that
2397	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
2398	explanation.	explanation.
2399
2400	2. A special-purpose tool can be "regisitered" by simply dropping the	2. A special-purpose tool can be "registered" by simply dropping the
2401	corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same	corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
2402	directory where the .cm file lives that uses this tool.  (The	directory where the .cm file lives that uses this tool.  (The
2403	behavior/misfeature until now was to look for the tool description	behavior/misfeature until now was to look for the tool description
#	Line 834	Line 2441
2441	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
2442	<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,
2443	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
2444	perhaps got accidentially killed).  In this case, fixpt will quickly	perhaps got accidentally killed).  In this case, fixpt will quickly
2445	move through what exists before continuing where it left off earlier,	move through what exists before continuing where it left off earlier,
2446	and, thus, saves a lot of time.	and, thus, saves a lot of time.
2447
#	Line 884	Line 2491
2491	it from that remote directory.	it from that remote directory.
2492	This should simplify installation further:  For machines that have	This should simplify installation further:  For machines that have
2493	access to the internet, just fetch <version>-config.tgz, unpack it,	access to the internet, just fetch <version>-config.tgz, unpack it,
2494	edit config/targets, and go (run config/install.sh).  The scipt will	edit config/targets, and go (run config/install.sh).  The script will
2495	fetch everything else that it might need all by itself.	fetch everything else that it might need all by itself.
2496
2497	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.634   changed lines   Added in v.815

---

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