Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 676, Sat Jun 24 03:37:03 2000 UTC	revision 749, Fri Dec 8 15:16:36 2000 UTC
#	Line 12	Line 12
12	Tag: <post-commit CVS tag>	Tag: <post-commit CVS tag>
13	Description:	Description:
14	----------------------------------------------------------------------	----------------------------------------------------------------------
15		Name: Lal George
16		Date: 2000/12/08 09:54:02 EST 2000
17		Tag: Release_110_31
18		Description:
19
20		- Version 110.31
21		----------------------------------------------------------------------
22		Name: Allen Leung
23		Date: Thu Dec  7 22:01:04 EST 2000
24		Tag:  leunga-20001207-cell-monster-hack
25		Description:
26
27		Major MLRISC internal changes.  Affect all clients.
28		Summary:
29
30		1.  Type CELLS.cell = int is now replaced by a datatype.
31		As a result, the old regmap is now gone.  Almost all interfaces
32		in MLRISC change as a consequence.
33
34		2.  A new brand version of machine description tool (v3.0) that generates
35		modules expecting the new interface.  The old version is removed.
36
37		3.  The RA interface has been further abstracted into two new functors.
38		RISC_RA and X86RA.  These functors have much simpler interfaces.
39		[See also directory MLRISC/demo.]
40
41		4.  Some other new source->source code generation tools are available:
42
43		a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
44		b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
45		I use this tool to generate the peephole optimizers---with the new
46		cell type changes, peephole rules are becoming difficult to write
47		without conditional pattern matching.
48
49		5.  More Intmap -> IntHashTable change.  Previous changes by Matthias didn't
50		cover the entire MLRISC source tree so many things broke.
51
52		6.  CM files have been moved to the subdirectory MLRISC/cm.
53		They are moved because there are a lot of them and they clutter up the
54		root dir.
55
56		7.  More detailed documentation to come...
57
58		NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
59		a makeml -rebuild first.  This is because of other other
60		changes that Matthias has made (see below).
61
62
63		----------------------------------------------------------------------
64		Name: Matthias Blume
65		Date: 2000/11/30 23:12:00 JST
66		Tag: blume-20001130-filereorg
67		Description:
68
69		Some manual updates and some file reorganizations in CM.
70
71		----------------------------------------------------------------------
72		Name: Matthias Blume
73		Date: 2000/11/24 17:45:00 JST
74		Tag: blume-20001124-link
75		Description:
76
77		Drastically improved link traversal code for the case that the dynamic
78		value was already loaded at bootstrap time.  As a result, CM and CMB
79		now both load blazingly fast -- even on a very slow machine.  Also,
80		memory consumption has been further reduced by this.
81
82		Warning: The format of the PIDMAP file has changed.  THerefore, to
83		bootstrap you have to do this:
84
85		1. Run CMB.make
86		2. Make a symbolic link for the boot directory:
87		ln -s sml.boot.ARCH-OS xxx
88		3. "Rebuild" the boot directory:
89		./makeml -boot xxx -rebuild sml ; rm xxx
90		4. Boot normally:
91		./makeml
92
93		----------------------------------------------------------------------
94		Name: Matthias Blume
95		Date: 2000/11/21 21:20:00 JST
96		Tag: blume-20001121-tools
97		Description:
98
99		Continued hacking on autoloading problem -- with success this time.
100		Also changed tool-plugin mechanism.  See new CM manual.
101
102		----------------------------------------------------------------------
103		Name: Matthias Blume
104		Date: 2000/11/19 14:30:00 JST
105		Tag:  blume-20001119-autoload
106		Description:
107
108		Some hacking to make autoloading faster.  Success for CMB, no success
109		so far for CM.  There is a reduced structure CM' that autoloads faster.
110		(This is a temporary, non-documented hack to be eliminated again when
111		the general problem is solved.)
112
113		----------------------------------------------------------------------
114		Name: Matthias Blume
115		Date: 2000/11/17 14:10:00 JST
116		Tag: blume-20001117-pickle-lib
117		Description:
118
119		1. Eliminated comp-lib.cm
120		2. Made pickle-lib.cm
121		3. Eliminated all uses of intset.sml (from comp-lib.cm)
122		4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
123		equivalent constructs from smlnj-lib.cm (INtHashTable).
124		5. Point 4. also goes for those uses of intmap.* in MLRISC.
125		Duplicated intmap modules thrown out.
126		6. Hunted down all duplicated SCC code and replaced it with
127		equivalent stuff (GraphSCCFn from smlnj-lib.cm).
128		7. Rewrote Feedback module.
129		8. Moved sortedlist.sml into viscomp-lib.cm.  Eventually it
130		should be thrown out and equivalent modules from smlnj-lib.cm
131		should be used (IntRedBlackSet, IntListSet, ...).
132
133		Confirmed that compiler compiles to fixpoint.
134
135		----------------------------------------------------------------------
136		Name: Allen Leung
137		Date: 2000/11/10 18:00:00
138		Tag: leunga-20001110-new-x86-fp
139
140		A new x86 floating point code generator has been added.
141		By default this is turned off.  To turn this on, do:
142
143		CM.autoload "$smlnj/compiler.cm";
144		Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
145
146		Changes:
147
148		1.  Changed FTAN to FPTAN so that the assembly output is correct.
149		2.  Changed the extension callback for FTANGENT to generate:
150
151		fptan
152		fstp  %st(0)
153		instead of
154		fptan
155		fstpl ftempmem
156
157		3.  Numerous assembly fixes for x86.
158
159		5.  Cleaned up the machine code output module x86/x86MC.sml and added
160		support for a whole bunch of instructions and addressing modes:
161
162		fadd/fsub/fsubr/fmul/fdiv/fdivr  %st, %st(n)
163		faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp  %st, %st(n)
164		fadd/fsub/fsubr/fmul/fdiv/fdivr  %st(n), %st
165		fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
166		fxch %st(n)
167		fld %st(n)
168		fst %st(n)
169		fst mem
170		fstp %st(n)
171		fucom %st(n)
172		fucomp %st(n)
173
174		All these are now generated when the fast fp mode is turned on.
175
176		6.  Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
177
178		----------------------------------------------------------------------
179		Name: Matthias Blume
180		Date: 2000/11/09 11:20:00 JST
181		Tag: blume-20001109-scc
182		Description:
183
184		Eliminated some code duplication:
185
186		1. Added "where" clause to GraphSCCFn in SML/NJ Library.
187		(Otherwise the functor is useless.)
188		2. Used GraphSCCFn where SCCUtilFun was used previously.
189		3. Got rid of SCCUtilFun (in comp-lib.cm).
190
191		----------------------------------------------------------------------
192		Name: Lal George
193		Date: 2000/11/06 09:02:21 EST 2000
194		Tag: Release_110_30
195		Description:
196
197		- Version 110.30
198		----------------------------------------------------------------------
199		Name: Matthias Blume
200		Date: 2000/11/04 14:45:00
201		Tag: blume-20001104-mlbuild
202		Description:
203
204		- Made ml-build faster on startup.
205		- Documentation fixes.
206
207		----------------------------------------------------------------------
208		Name: Matthias Blume
209		Date: 2000/11/02 17:00:00 JST
210		Tag: blume-20001102-condcomp
211		Description:
212
213		- Small tweaks to pickler -- new BOOTFILES!
214		- Version bumped to 110.29.2.
215		- Added conditional compilation facility to init.cmi (see comment there).
216		----------------------------------------------------------------------
217		Name: Allen Leung
218		Date: 2000/10/23 19:31:00
219		Tag: leunga-20001023-demo-ra
220
221		1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
222		2. Test programs for the graph library updated
223		3. Some new MLRISC demo programs added
224
225		----------------------------------------------------------------------
226		Name: Matthias Blume
227		Date: 2000/08/31 22:15:00 JST
228		Tag: blume-20001017-errmsg
229		Description:
230
231		More error message grief: Where there used to be no messages, there
232		now were some that had bogus error regions.  Fixed.
233
234		----------------------------------------------------------------------
235		Name: Matthias Blume
236		Date: 2000/08/31 17:30:00 JST
237		Tag: blume-20001017-v110p29p1
238		Description:
239
240		I made a version 110.29.1 with new bootfiles.
241
242		Changes:  Modified pickler/unpickler for faster and leaner unpickling.
243		CM documentation changes and a small bugfix in CM's error reporting.
244
245		----------------------------------------------------------------------
246		Name: Lal George
247		Date: 2000/09/27 14:42:35 EDT
248		Tag: george-20000927-nodestatus
249		Description:
250
251		Changed the type of the nodestatus, so that:
252
253		SPILLED(~1)             is now SPILLED
254		SPILLED(m) where m>=0   is now MEMREG(m)
255		SPILLED(s) where s<~1   is now SPILL_LOC(~s)
256
257		----------------------------------------------------------------------
258		Name: Matthias Blume
259		Date: 2000/09/07 14:45:00 JST
260		Tag: blume-20000907-cmerrmsg
261		Description:
262
263		Small tweak to CM to avoid getting ML syntax error messages twice.
264
265		----------------------------------------------------------------------
266		Name: Matthias Blume
267		Date: 2000/08/31 18:00:00 JST
268		Tag: blume-20000831-cvsbootfiles
269		Description:
270
271		New URL for boot files (because the 110.29 files on the BL server do
272		now work correctly with my updated install scripts for yacc and lex).
273
274		----------------------------------------------------------------------
275		Name: Matthias Blume
276		Date: 2000/08/08 12:33:00 JST
277		Tag: blume-20000808-manual
278		Description:
279
280		Tiny update to CM manual.
281
282		----------------------------------------------------------------------
283		Name: Allen Leung
284		Date: 2000/08/7 19:31:00
285		Tag: leunga-20000807-a-whole-bunch-of-stuff
286
287		Moby, C--, SSA, x86, machine descriptions etc.  Should only affect C--
288		and Mobdy.
289
290		1.  x86
291
292		a.  Fixes to peephole module by John and Dan.
293		b.  Assembly fix to SETcc by Allen.
294		c.  Fix to c-call by John.
295		d.  Fix to spilling by John.  (This one deals with the missing FSTPT case)
296		e.  Instruction selection optimization to SETcc as suggested by John.
297
298		For example,
299
300		MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
301
302		should generate:
303
304		MOVL a, x
305		SUBL b, x
306		SHRL 31, x
307
308		2.  IR stuff
309
310		A bunch of new DJ-graph related algorithms added.  These
311		speed up SSA construction.
312
313		3.  SSA + Scheduling
314
315		Added code for SSA and scheduling to the repository
316
317		----------------------------------------------------------------------
318		Name: Lal George
319		Date: 2000/07/27 11:53:14 EDT
320
321		Tag: lal-20000727-linux-ppc
322		Description:
323
324		Made changes to support Linux PPC.
325		p.s. I have confirmation that the 110.29 boot files work fine.
326
327		----------------------------------------------------------------------
328		Name: Matthias Blume
329		Date: 2000/07/27 17:40:00 JST
330		Tag: blume-20000727-scripts
331		Description:
332
333		!!!! WARNING !!!!
334		You must recompile the runtime system!
335		!!!! WARNING !!!!
336
337		This is basically another round of script-enhancements:
338
339		1. sml, ml-build, and ml-makedepend accept options -D and -U to define
340		and undefine CM preprocessor symbols.
341
342		2. ml-build avoids generating a new heap image if it finds that the
343		existing one is still ok.  (The condition is that no ML file had to
344		be recompiled and all ML files are found to be older that the heap
345		file.)
346
347		To make this work smoothly, I also hacked the runtime system as
348		well as SMLofNJ.SysInfo to get access to the heap image suffix
349		(.sparc-solaris, ...) that is currently being used.
350
351		Moreover, the signature of CM.mk_standalone has changed.  See the
352		CM manual.
353
354		3. ml-makedepend accepts additional options -n, -a, and -o.  (See the
355		CM manual for details.)
356
357		4. More CM manual updates:
358		- all of the above has been documented.
359		- there is now a section describing the (CM-related) command line
360		arguments that are accepted by the "sml" command
361
362		----------------------------------------------------------------------
363		Name: Matthias Blume
364		Date: 2000/07/25 16:20:00 JST
365		Tag: blume-20000725-makedepend
366		Description:
367
368		Added a script called ml-makedepend.  This can be used in makefiles
369		for Unix' make in a way very similar to the "makedepend" command for
370		C.
371
372		The script internally uses function CM.sources.
373
374		Synopsis:
375
376		ml-makedepend [-f makefile] cmfile targetname
377
378		The default for the makefile is "makefile" (or "Makefile" should
379		"makefile" not exist).
380
381		ml-makedepend adds a cmfile/targetname-specific section to this
382		makefile (after removing the previous version of this section).  The
383		section contains a single dependency specification with targetname on
384		the LHS (targetname is an arbitrary name), and a list of files derived
385		from the cmfile on the RHS.  Some of the files on the RHS are
386		ARCH/OPSYS-specific.  Therefore, ml-makedepend inserts references to
387		"make" variables $(ARCH) and $(OPSYS) in place of the corresponding
388		path names.  The makefile writer is responsible for making sure that
389		these variables have correct at the time "make" is invoked.
390
391		----------------------------------------------------------------------
392		Name: Matthias Blume
393		Date: 2000/07/22 23:30:00 JST
394		Tag: blume-20000722-urlupdate
395		Description:
396
397		Changed BOOT and config/srcarchiveurl to point to BL server:
398
399		ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
400
401		----------------------------------------------------------------------
402		Name: Matthias Blume
403		Date: 2000/07/18 18:00:00 JST
404		Tag: blume-20000718-Version_110_29
405		Description:
406
407		1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
408
409		2. Updated config/version to 110.29
410
411		3. Updated config/srcarchiveurl
412
413		3. New boot files!
414		ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
415
416		----------------------------------------------------------------------
417		Name: Matthias Blume
418		Date: 2000/07/11 13:58:00 JST
419		Tag: blume-20000711-doctypo
420		Description:
421
422		Fixed a few typos in CM manual.
423
424		----------------------------------------------------------------------
425		Name: Allen Leung
426		Date: 2000/06/15 00:38:00
427		Tag: leunga-20000704-sparc-x86
428
429		1. x86 peephole improvement sp += k; sp -= k => nop  [from John]
430		2. fix to x86 RET bug [found by Dan Grossman]
431		3. sparc assembly bug fix for ticc instructions [found by Fermin]
432
433		Affects c-- and moby only
434
435		----------------------------------------------------------------------
436		Name: Matthias Blume
437		Date: 2000/07/04 15:26:00
438		Tag: blume-20000704-trigger
439		Description:
440
441		1. Improvements to CM manual.
442		2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
443		of getting a back-trace.  The function, when called, raises an
444		internal exception which explicitly carries the full back-trace history,
445		so it is unaffected by any intervening handle-raise pairs ("trivial"
446		or not).  The interactive loop will print that history once it arrives
447		at top level.
448		Short of having all exceptions implicitly carry the full history, the
449		recommended way of using this facility is:
450		- compile your program with instrumentation "on"
451		- run it, when it raises an exception, look at the history
452		- if the history is "cut off" because of some handler, go and modify
453		your program so that it explicitly calls BTrace.trigger
454		- recompile (still instrumented), and rerun; look at the full history
455
456		----------------------------------------------------------------------
457		Name: Matthias Blume
458		Date: 2000/07/03 15:36:00 JST
459		Tag: blume-20000702-manual
460		Description:
461
462		Small corrections and updates to CM manual.
463
464		----------------------------------------------------------------------
465		Name: Matthias Blume
466		Date: 2000/06/29 16:04:00 JST
467		Tag: blume-20000629-yacctool
468		Description:
469
470		Changes:
471
472		1. Class "mlyacc" now takes separate arguments to pass options to
473		generated .sml- and .sig-files independently.
474		2. Corresponding CM manual updates.
475		3. BTrace module now also reports call sites.  (However, for loop clusters
476		it only shows from where the cluster was entered.)  There are associated
477		modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
478
479		----------------------------------------------------------------------
480		Name: Matthias Blume
481		Date: 2000/06/27 16:51:00 JST
482		Tag: blume-20000627-noweb
483		Description:
484
485		Changes:
486
487		1. Implemented "subdir" and "witness" options for noweb tool.
488		This caused some slight internal changes in CM's tool implementation.
489		2. Fixed bug in "tool plugin" mechanism.  This is essentially cleaning
490		some remaining issues from earlier path anchor changes.
491		3. Updated CM manual accordingly.
492
493		4. Changed implementation of back-tracing so that I now consider it
494		ready for prime-time.
495
496		In particular, you don't have to explicitly trigger the back-trace
497		anymore.  Instead, if you are running BTrace-instrumented code and
498		there is an uncaught exception (regardless of whether or not it was
499		raised in instrumented code), the top-level evalloop will print
500		the back-trace.
501
502		Features:
503
504		- Instrumented and uninstrumented code work together seemlessly.
505		(Of course, uninstrumented code is never mentioned in actual
506		back-traces.)
507
508		- Asymptotic time- and space-complexity of instrumented code is
509		equal to that of uninstrumented code.  (This means that
510		tail-recursion is preserved by the instrumentation phase.)
511
512		- Modules whose code has been instrumented in different sessions
513		work together without problem.
514
515		- There is no penalty whatsoever on uninstrumented code.
516
517		- There is no penalty on "raise" expressions, even in
518		instrumented code.
519
520		A potential bug (or perhaps it is a feature, too):
521
522		A back-trace reaches no further than the outermost instrumented
523		non-trivial "raise".  Here, a "trivial" raise is one that is the
524		sole RHS of a "handle" rule.  Thus, back-traces reach trough
525
526		<exp> handle e => raise e
527
528		and even
529
530		<exp> handle Foo => raise Bar
531
532		and, of course, through
533
534		<exp> handle Foo => ...
535
536		if the exception was not Foo.
537
538		Back-traces always reach right through any un-instrumented code
539		including any of its "handle" expressions, trivial or not.
540
541		To try this out, do the following:
542
543		- Erase all existing binfiles for your program.
544		(You may keep binfiles for those modules where you think you
545		definitely don't need back-tracing.)
546		- Turn on back-trace instrumentation:
547		SMLofNJ.Internals.BTrace.mode (SOME true);
548		- Recompile your program.  (I.e., run "CM.make" or "use".)
549		- You may now turn instrumentation off again (if you want):
550		SMLofNJ.Internals.BTrace.mode (SOME false);
551		- Run your program as usual.  If it raises an exception that
552		reaches the interactive toplevel, then a back-trace will
553		automatically be printed.  After that, the toplevel loop
554		will print the exception history as usual.
555
556		----------------------------------------------------------------------
557		Name: Matthias Blume
558		Date: 2000/06/26 09:56:46 JST
559		Tag: blume-20000626-setup
560		Description:
561
562		CM: - setup-parameter to "sml" added; this can be used to run arbitrary
563		ML code before and after compiling a file (e.g., to set compiler
564		flags)
565
566		Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
567		- associated changes to btrace.sml (BTrace instrumentation pass)
568		- cleaner implementation of btimp.sml (BTrace tracing and report
569		module)
570
571		CM manual: * new path encoding documented
572		* description of setup-parameter to "sml" added
573
574		The biggest user-visible change to back-tracing is that it is no
575		longer necessary to compile all traced modules within the same
576		session.  (This was a real limitation.)
577
578		----------------------------------------------------------------------
579	Name: Matthias Blume	Name: Matthias Blume
580	Date: 2000/06/24 12:40:00 JST	Date: 2000/06/24 12:40:00 JST
581	Tag: blume-20000624-startup	Tag: blume-20000624-startup

---

Colored Diff Long Colored Diff Unidiff Context Diff Side by Side

Legend: Removed from v.676   changed lines   Added in v.749

---

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