Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 633, Sun Apr 30 16:44:09 2000 UTC	revision 873, Mon Jul 9 15:26:20 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		----------------------------------------------------------------------
16		Name: Matthias Blume
17		Date: 2001/07/09 11:25:00 EDT
18		Tag: blume-20010709-varargs
19		Description:
20
21		1. ckit-lib.cm now exports structure Error
22		2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
23		with a warning accompanied by a source location.  Moreover, it
24		merely skips the offending function or type and proceeds with the
25		rest of its work.u  As a result, one can safely feed C code containing
26		"..." to ml-nlffigen.
27		3. There are some internal improvements to CM, providing slightly
28		more general string substitutions in the tools subsystem.
29
30		----------------------------------------------------------------------
31		Name: Matthias Blume
32		Date: 2001/06/27 15:10:00 EDT
33		Tag: blume-20010627-concur
34		Description:
35
36		Fixed a small bug in CM's handling of parallel compilation.
37		(You could observe the bug by Control-C-interrupting an ordinary
38		CMB.make or CM.stabilize and then attaching some compile servers.
39		The result was that all of a sudden the previously interrupted
40		compilation would continue on its own.  This was because of
41		an over-optimization: CM did not bother to clean out certain queues
42		when no servers were attached "anyway", resulting in the contents
43		of these queues to grab control when new servers did get attached.)
44
45		There is also another minor update to the CM manual.
46
47		----------------------------------------------------------------------
48		Name: Matthias Blume
49		Date: 2001/06/26 16:15:00 EDT
50		Tag: blume-20010626-cmdoc
51		Description:
52
53		Minor typo fixed in CM manual (syntax diagram for libraries).
54
55		----------------------------------------------------------------------
56		Name: Matthias Blume
57		Date: 2001/06/25 22:55:00 EDT
58		Tag: blume-20010625-x86pc
59		Description:
60
61		Fixed a nasty bug in the X86 assembly code that caused signal
62		handlers to fail (crash) randomly.
63
64		----------------------------------------------------------------------
65		Name: Matthias Blume
66		Date: 2001/06/25 12:05:00 EDT
67		Tag: blume-20010625-nlffigen
68		Description:
69
70		This update fixes a number of minor bugs in ml-nlffigen as reported by
71		Nick Carter <nbc@andrew.cmu.edu>.
72
73		1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
74		2. Default names for generated files are now derived from the name of
75		the C file *without its directory*.  In particular, this causes generated
76		files to be placed locally even if the C file is in some system directory.
77		3. Default names for generated signatures and structures are also derived
78		from the C file name without its directory.  This avoids silly things
79		like "structure GL/GL".
80		(Other silly names are still possible because ml-nlffigen does not do
81		a thorough check of whether generated names are legal ML identifiers.
82		When in doubt, use command line arguments to force particular names.)
83
84		----------------------------------------------------------------------
85		Name: Matthias Blume
86		Date: 2001/06/21 12:25:00 EDT
87		Tag: blume-20010621-eXene
88		Description:
89
90		eXene now compiles and (sort of) works again.
91
92		The library name (for version > 110.33) is $/eXene.cm.
93
94		I also added an new example in src/eXene/examples/nbody.  See the
95		README file there for details.
96
97		----------------------------------------------------------------------
98		Name: Matthias Blume
99		Date: 2001/06/20 16:40:00 EDT
100		Tag: blume-20010620-cml
101		Description:
102
103		CML now compiles and works again.
104
105		Libraries (for version > 110.33):
106
107		$cml/cml.cm            Main CML library.
108		$cml/basis.cm          CML's version of $/basis.cm.
109		$cml/cml-internal.cm   Internal helper library.
110		$cml/core-cml.cm       Internal helper library.
111		$cml-lib/trace-cml.cm  Tracing facility.
112		$cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
113
114		The installer (config/install.sh) has been taught how to properly
115		install this stuff.
116
117		----------------------------------------------------------------------
118		Name: Matthias Blume
119		Date: 2001/06/19 17:55:00 EDT
120		Tag: blume-20010619-instantiate
121		Description:
122
123		This un-breaks the fix for bug 1432.
124		(The bug was originally fixed in 110.9 but I broke it again some
125		time after that.)
126
127		----------------------------------------------------------------------
128		Name: Matthias Blume
129		Date: 2001/06/19 17:25:00 EDT
130		Tag: blume-20010619-signals
131		Description:
132
133		This should (hopefully) fix the long-standing signal handling bug.
134		(The runtime system was constructing a continuation record with an
135		incorrect descriptor which would cause the GC to drop data on the floor...)
136
137		----------------------------------------------------------------------
138		Name: Matthias Blume
139		Date: 2001/06/15 15:05:00 EDT
140		Tag: blume-20010615-moresparc
141		Description:
142
143		Here is a short late-hour update related to Sparc c-calls:
144
145		-- made handling of double-word arguments a bit smarter
146
147		-- instruction selection phase tries to collapse certain clumsily
148		constructed ML-Trees; typical example:
149
150		ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
151
152		This currently has no further impact on SML/NJ since mlriscGen does
153		not seem to generate such patterns in the first place, and c-calls
154		(which did generate them in the beginning) has meanwhile been fixed
155		so as to avoid them as well.
156
157		----------------------------------------------------------------------
158		Name: Matthias Blume
159		Date: 2001/06/15 15:05:00 EDT
160		Tag: blume-20010615-sparc
161		Description:
162
163		The purpose of this update is to provide an implementation of NLFFI
164		on Sparc machines.
165
166		Here are the changes in detail:
167
168		* src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
169		the Sparc implementation of the c-calls API.
170		* The Sparc backend of SML/NJ has been modified to uniformely use %fp
171		for accessing the ML frame.  Thus, we have a real frame pointer and
172		can freely modify %sp without need for an omit-frame-ptr phase.
173		The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
174		this case.
175		* ml-nlffigen has been taught to produce code for different architectures
176		and calling conventions.
177		* In a way similar to what was done in the x86 case, the Sparc
178		backend uses its own specific extension to mltree.  (For example,
179		it needs to be able to generate UNIMP instructions which are part
180		of the calling convention.)
181		* ml-nlffi-lib was reorganized to make it more modular (in particular,
182		to make it easier to plug in new machine- and os-dependent parts).
183
184		There are some other fairly unrelated bug fixes and cleanups as well:
185
186		* I further hacked the .cm files for MLRISC tools (like MDLGen) so
187		that they properly share their libraries with existing SML/NJ libraries.
188		* I fixed a minor cosmetic bug in CM, supressing certain spurious
189		follow-up error messages.
190		* Updates to CM/CMB documentation.
191
192		TODO items:
193
194		* MLRISC should use a different register as its asmTemp on the Sparc.
195		(The current %o2 is a really bad choice because it is part of the
196		calling conventions, so things might interfere in unexpected ways.)
197
198		----------------------------------------------------------------------
199		Name: Matthias Blume
200		Date: 2001/06/07
201		Tag: blume-20010607-calls
202		Description:
203
204		A number of internal changes related to C calls and calling conventions:
205
206		1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
207		number of bytes popped implicitly (by the callee).  In most cases
208		this field is 0 but on x86/win32 it is some non-zero value.  This
209		is information provided for the benefit of the "omit-frameptr" pass.
210		2. The CALL instruction on the x86 carries a similar "pops" field.
211		The instruction selection phase copies its value from the ML-Tree
212		CALL statement.
213		3. On all other architectures, the instruction selection phase checks
214		whether "pops=0" and complains if not.
215		4. The c-calls implementation for x86 now accepts two calling conventions:
216		"ccall" and "stdcall".  When "ccall" is selected, the caller cleans
217		up after the call and pops is set to 0.  For "stdcall", the caller
218		does nothing, leaving the cleanup to the callee; pops is set to
219		the number of bytes that were pushed onto the stack.
220		5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
221		distinguish between "ccall" and "stdcall".
222		6. The UNIMP instruction has been added to the supported Sparc instruction
223		set. (This is needed for implementing the official C calling convention
224		on this architecture.)
225		7. I fixed some of the .cm files under src/MLRISC/Tools to make them
226		work with the latest CM.
227
228		----------------------------------------------------------------------
229		Name: Matthias Blume
230		Date: 2001/06/05 15:10:00 EDT
231		Tag: blume-20010605-cm-index
232		Description:
233
234		0. The "lambdasplit" parameter for class "sml" in CM has been documented.
235
236		1. CM can now generate "index files".  These are human-readable files
237		that list on a per-.cm-file basis each toplevel symbol defined or
238		imported.  The location of the index file for
239		<p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
240		To enable index-file generation, set CM.Control.generate_index to true
241		or export an environment-symbol: export CM_GENERATE_INDEX=true.
242
243		The CM manual has been updated accordingly.
244
245		2. I made some slight modifications to the c-calls API in MLRISC.
246
247		a) There is now a callback to support saving/restoring of
248		dedicated but caller-save registers around the actual call
249		instruction.
250		b) One can optionally specify a comment-annotation for the
251		call instruction.
252
253		3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
254		(For example, the comment annotation shows the C prototype of
255		the function being called.)
256
257		----------------------------------------------------------------------
258		Name: Matthias Blume
259		Date: 2001/06/01 13:30:00 EDT
260		Tag: blume-20010601-nlffi-cleanup
261		Description:
262
263		This is mostly a cleanup of MLFFI stuff:
264
265		- some signature files have been put into a more exposed place
266		- the ugly 'f type parameter is gone (simplifies types tremendously!)
267		- ml-nlffigen changed accordingly
268		- tutorial updated
269
270		Other changes:
271
272		- author's affiliation in CM manual(s) updated
273		- some more recognized keywords added to Allen's sml.sty
274
275		----------------------------------------------------------------------
276		Name: Matthias Blume
277		Date: 2001/05/25 15:30:00 EDT
278		Tag: blume-20010525-iptr
279		Description:
280
281		- put the official 110.33-README (as it appears on the ftp server) under
282		CVS
283		- fixed a small bug related to incomplete pointer types in
284		ml-nlffigen
285		- small cosmetic change to the ml-nlffi-lib's "arr" type constructor
286		(it does not need the 'f type parameter)
287
288		----------------------------------------------------------------------
289		Name: Matthias Blume
290		Date: 2001/05/23 14:30:00 EDT
291		Tag: Release_110_33
292		Description:
293
294		New version number (110.33).  New bootfiles.
295
296		----------------------------------------------------------------------
297		Name: Matthias Blume
298		Date: 2001/05/22 18:06:00 EDT
299		Tag: blume-20010522-targets
300		Description:
301
302		Made install.sh use file config/targets.customized if it exists, falling
303		back to config/targets if it doesn't.  This way one can have a customized
304		version of the targets file without touching the "real thing", thus
305		eliminating the constant fear of accidentally checking something bogus
306		back into the CVS repository...  (File config/targets.customized must
307		not be added to the repository!)
308
309		----------------------------------------------------------------------
310		Name: Matthias Blume
311		Date: 2001/05/22 16:30:00 EDT
312		Tag: blume-20010522-minitut
313		Description:
314
315		1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
316		struct returns.
317		2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
318		incomplete, preliminary documentation for NLFFI.
319
320		----------------------------------------------------------------------
321		Name: Matthias Blume
322		Date: 2001/05/14 11:30:00 EDT
323		Tag: blume-20010514-script
324		Description:
325
326		Some bugs in install script fixed.
327
328		In addition to that I also made a slight change to the NLFFI API:
329		Functors generated by ml-nlffigen now take the dynamic library as a
330		straight functor argument, not as a suspended one.  (The original
331		functor code used to force the suspension right away anyway, so there
332		was nothing gained by this complication of the interface.)
333
334		----------------------------------------------------------------------
335		Name: Matthias Blume
336		Date: 2001/05/11 14:35:00 EDT
337		Tag: blume-20010511-ml-nlffi
338		Description:
339
340		I finally took the plunge and added my new FFI code to the main
341		repository.  For x86-linux it is now ready for prime-time.
342
343		There are two new subdirectories of "src":
344
345		- ml-nlffi-lib:
346		The utility library for programs using the FFI interface.
347		Here is the implementation of $/c.cm and its associated low-level
348		partners $/c-int.cm and $/memory.cm.
349		- ml-nlffigen:
350		A stand-alone program for generating ML glue code from C source
351		code.
352
353		Building ml-nlffigen requires $/ckit-lib.cm.
354
355		The config/install.sh script has been updates to do the Right Thing
356		(hopefully).
357
358		Notice that the source tree for the C-Kit will not be put under "src"
359		but directly under the installation root directory.  (This is the
360		structure that currently exists on the CVS server when you check out
361		module "sml".)  Fortunately, config/install.sh knows about this oddity.
362
363		Bugs: No documentation yet.
364
365		----------------------------------------------------------------------
366		Name: Matthias Blume
367		Date: 2001/05/09 16:35:00 EDT
368		Tag: blume-20010509-cpscontract
369		Description:
370
371		Fixed a bug in the accounting code in cpsopt/contract.sml.  (The
372		wrapper/unwrapper elimination did not decrement usage counts and some
373		dead variables got overlooked by the dead-up logic.)
374
375		----------------------------------------------------------------------
376		Name: Lal George
377		Date: 2001/05/08  17:26:09 EDT
378		Tag: george-20010508-omit-frameptr
379		Description:
380
381		Changes to implement the omit-frame-pointer optimization to support
382		raw C calls. For now, there is only support on the Intel x86, but
383		other architectures will follow as more experience is gained with this.
384
385
386		----------------------------------------------------------------------
387		Name: Matthias Blume
388		Date: 2001/05/07 14:40:00 EDT
389		Tag: blume-20010507-proxies
390		Description:
391
392		I made into "proxy libraries" all libraries that qualify for such a
393		change.  (A qualifying library is a library that has another library or
394		groups as its sole member and repeats that member's export list
395		verbatim.  A proxy library avoids this repetition by omitting its export
396		list, effectively inheriting the list that its (only) member exports.
397		See the CM manual for more explanation.)
398		The main effect is that explicit export lists for these libraries
399		do not have to be kepts in sync, making maintenance a bit easier.
400
401		I also added copyright notices to many .cm-files.
402
403		Last but not least, I made a new set of bootfiles.
404
405		----------------------------------------------------------------------
406		Name: Matthias Blume
407		Date: 2001/05/04 17:00:00 EDT
408		Tag: blume-20010504-cm-lsplit
409		Description:
410
411		0. John merged pending changes to $/smlnj-lib.cm
412
413		1. Allen's previous change accidentally backed out of one of Lal's
414		earlier changes.  I undid this mistake (re-introducing Lal's change).
415
416		2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
417		within the compiler where applicable.  There is some code simplification
418		because of that.
419
420		3. The "split" phase (in FLINT) is now part of the default list of phases.
421		Compiler.Control.LambdaSplitting.* can be used to globally control the
422		lambda-splitting (cross-module-inlining) engine.  In addition to that,
423		it can now also be controlled on a per-source basis: CM has been taught
424		a new tool parameter applicable to ML source files.
425
426		- To turn lambda-splitting off completely:
427		local open Compiler.Control.LambdaSplitting in
428		val _ = set Off
429		end
430		- To make "no lambda-splitting" the global default (but allow per-source
431		overriding); this is the initial setting:
432		local open Compiler.Control.LambdaSplitting in
433		val _ = set (Default NONE)
434		end
435		- To make "lambda-splitting with aggressiveness a" the global default
436		(and allow per-source overriding):
437		local open Compiler.Control.LambdaSplitting in
438		val _ = set (Default (SOME a))
439		end
440
441		- To turn lambda-splitting off for a given ML souce file (say: a.sml)
442		write (in the respective .cm-file):
443		a.sml (lambdasplitting:off)
444		- To turn lambda-splitting for a.sml on with minimal aggressiveness:
445		a.sml (lambdasplitting:on)
446		- To turn lambda-splitting for a.sml on with aggressiveness <a> (where
447		<a> is a decimal non-negative integer):
448		a.sml (lambdasplitting:<a>)
449		- To turn lambda-splitting for a.sml on with maximal aggressiveness:
450		a.sml (lambdasplitting:infinity)
451		- To use the global default for a.sml:
452		a.sml (lambdasplitting:default)
453		or simply
454		a.sml
455
456		----------------------------------------------------------------------
457		Name: Allen Leung
458		Date: 2001/05/04 01:57:00 EDT
459		Tag: leunga-20010504-sync
460		Description:
461
462		MLRISC features.
463
464		1. Fix to CMPXCHG instructions.
465		2. Changed RA interface to allow annotations in callbacks.
466		3. Added a new method to the stream interface to allow annotations updates.
467
468		----------------------------------------------------------------------
469		Name: Matthias Blume
470		Date: 2001/05/01 11:45:00 EDT
471		Tag: blume-20010501-pcedittmp
472		Description:
473
474		Changed install.sh to use the current working directory instead of
475		/usr/tmp for a temporary file (pcedittmp).  The previous choice
476		of /usr/tmp caused trouble with MacOS X because of file premission
477		problems.
478
479		----------------------------------------------------------------------
480		Name: Matthias Blume
481		Date: 2001/04/20 11:10:00 EDT
482		Tag: blume-20010420-inMLflag
483		Description:
484
485		- added vp_limitPtrMask to vproc-state.h
486		(for use by the raw-C-calls mechanism to implement proper interrupt
487		handling)
488		- made the ML compiler aware of various data-structure offsets so it
489		can generate code for accessing the vp_inML flag and vp_limitPtrMask
490		- tweaked mlriscGen.sml to have it emit interrupt-handling code for
491		raw C-calls
492
493		----------------------------------------------------------------------
494		Name: Lal George
495		Date: 2001/04/20 09:15:28 EDT
496		Tag: george-20010420-macosX
497		Description:
498
499		- Changes to port to Mac OS X; Darwin.
500
501		- In the process I found that sqrt was broken on the PPC, because the
502		fsqrt instruction is not implemented.
503
504		----------------------------------------------------------------------
505		Name: Matthias Blume
506		Date: 2001/04/18 12:45:00 EDT
507		Tag: blume-20010418-ccalls
508		Description:
509
510		- fixed two off-by-4 errors in the x86-specific c-calls implementation
511		(this bug prevented structure arguments containing pointers from being
512		passed correctly)
513		- changed the raw-C-call code in mlriscGen.sml in such a way that
514		structure arguments are represented as a pointer to the beginning
515		of the structure (instead of having a series of synthesized arguments,
516		one for each structure member)
517
518		- made makeml script's verbosity level configurable via environment
519		variable (MAKEML_VERBOSITY)
520
521		- eliminated placeholder implementations for f32l, w16s, i16s, and f32s
522		in rawmem-x86.sml; we are now using the real thing
523
524		----------------------------------------------------------------------
525		Name: Matthias Blume
526		Date: 2001/03/22 16:25:00 EST
527		Tag: blume-20010322-bootfiles
528		Description:
529
530		Created a new set of bootfiles (for your automatic installation convenience).
531
532		----------------------------------------------------------------------
533		Name: Matthias Blume
534		Date: 2001/03/22 15:10:00 EST
535		Tag: blume-20010322-rawmem-parcm
536		Description:
537
538		1. All "raw memory access" primitives for the new FFI are implemented now
539		(at least on the x86).
540		2. Some further cleanup of CM's parallel make mechanism.
541
542		----------------------------------------------------------------------
543		Name: Matthias Blume
544		Date: 2001/03/19 17:53:00 EST
545		Tag: blume-20010319-parallel
546		Description:
547
548		Parallel make (using compile servers) now works again.
549
550		To this end, CM.stabilize and CMB.make have been modified to work in
551		two passes when compile servers are attached:
552		1. Compile everything, do not perform stabilization; this pass
553		uses compile servers
554		2. Stabilize everything; this pass does not use compile servers
555		If there are no compile servers, the two passes are combined into one
556		(as before).  Splitting the passes increases the inherent parallelism
557		in the dependency graph because the entire graph including all
558		libraries is available at the same time.  This, in turn, improves
559		server utilization.  The downside is that the master process will
560		have to do some extra work after compilation is done (because for
561		technical reasons it must re-read all the binfiles during stabilization).
562
563		----------------------------------------------------------------------
564		Name: Matthias Blume
565		Date: 2001/03/16 12:22:00 EST
566		Tag: blume-20010316-bootfiles
567		Description:
568
569		Created a new set of bootfiles (for your automatic installation convenience).
570
571		----------------------------------------------------------------------
572		Name: Matthias Blume
573		Date: 2001/03/16 11:00:00 EST
574		Tag: blume-20010316-MLTREE-fixup
575		Description:
576
577		This is a minor fixup for an (untagged) earlier commit by Allen.
578		(A file was missing).
579
580		----------------------------------------------------------------------
581		Name: Allen Leung
582		Date: Mon Mar  5 18:54:57 EST 2001
583		Tag: leunga-20010305-cut-support
584
585		1. New support for alternative control-flow in MLTREE.
586		Currently we support
587
588		FLOW_TO(CALL ...., [k1,...,kn])
589
590		This is needed for 'cuts to' in C-- and try/handle-like constructs
591		in Moby
592
593		New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
594
595		2. Register Allocator
596
597		Changes in interface [from Fermin, John]
598
599		3. Alpha 8-bit SLL support [Fermin]
600
601		4. All architectures
602
603		A new module (ClusterExpandCopies) for expanding parallel copies.
604
605		----------------------------------------------------------------------
606		Name: Allen Leung
607		Date: 2001/02/27 23:07:00 EST
608		Tag: leunga-20010227-minor-stuff
609
610		1. Alpha bug fix for CMOVNE
611		2. Handle mltree COND(..,FCMP ...,...)
612		3. Bug fix in simplifier
613
614		----------------------------------------------------------------------
615		Name: Matthias Blume
616		Date: 2001/01/30 17:50:00 EST
617		Tag: blume-20010130-sync
618		Description:
619
620		This is just a minor update to sync my devel branch with the main brach.
621		The only visible change is the addition of some README files.
622
623		----------------------------------------------------------------------
624		Name: Matthias Blume
625		Date: 2001/01/12 23:30:00 JST
626		Tag: blume-20010112-bootfiles
627		Description:
628
629		Made a new set of bootfiles that goes with the current state of the
630		repository.
631
632		----------------------------------------------------------------------
633		Name: Matthias Blume
634		Date: 2001/01/12 21:20:00 JST
635		Tag: blume-20010112-sync
636		Description:
637
638		I am just flushing out some minor changes that had accumulated in
639		my private branch in order to sync with the main tree.  (This is
640		mainly because I had CVS trouble when trying to merge _into_ my
641		private branch.)
642
643		Most people should be completely unaffected by this.
644
645		----------------------------------------------------------------------
646		Name: Allen Leung
647		Date: Thu Jan 11 21:03:00 EST 2001
648		Tag: leunga-20010111-labexp=mltree
649		Description:
650
651		1.  Removed the type LabelExp and replace it by MLTree.
652		2.  Rewritten mltree-simplify with the pattern matcher tool.
653		3.  There were some bugs in alpha code generator which would break
654		64-bit code generation.
655		4.  Redo the tools to generate code with the
656		5.  The CM files in MLRISC (and in src/system/smlnj/MLRISC)
657		are now generated by perl scripts.
658
659		----------------------------------------------------------------------
660		Name: Matthias Blume
661		Date: 2001/01/10 21:55:00 JST
662		Tag: blume-20010110-rcc
663		Description:
664
665		The RCC stuff now seems to work (but only on the x86).
666		This required hacking of the c-calls interface (and -implementation) in
667		MLRISC.
668
669		Normal compiler users should be unaffected.
670
671		----------------------------------------------------------------------
672		Name: Matthias Blume
673		Date: 2001/01/09 01:20:00 JST
674		Tag: blume-20010109-rcc
675		Description:
676
677		This is a fairly big patch, flushing out a large number of pending
678		changes that I made to my development copy over the last couple of days.
679
680		Of practical relevance at this moment is a workaround for a pickling
681		bug that Allen ran into the other day.  The cause of the bug itself is
682		still unknown and it might be hard to fix it properly, but the
683		workaround has some merits of its own (namely somewhat reducing pickling
684		overhead for certain libraries).  Therefore, I think this solution should
685		be satisfactory at this time.
686
687		The rest of the changes (i.e., the vast majority) has to do with my
688		ongoing efforts of providing direct support for C function calls from
689		ML.  At the moment there is a new primop "RAW_CCALL", typing magic
690		in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
691		case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
692		translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
693		handle RCC.
694
695		The last part (the changes to mlriscGen.sml) are still known to be
696		wrong on the x86 and not implemented on all other architectures.  But
697		the infrastructure is in place. I had to change a few functor
698		signatures in the backend to be able to route the CCalls interface
699		from MLRISC there, and I had to specialize the mltree type (on the
700		x86) to include the necessary extensions. (The extensions themselves
701		were already there and redy to go in MLRISC/x86).
702
703		Everything should be very happy as soon as someone helps me with
704		mlriscGen.sml...
705
706		In any case, nothing of this should matter to anyone as long as the
707		new primop is not being used (which is going to be the case unless you
708		find it where I hid it :). The rest of the compiler is completely
709		unaffected.
710
711		----------------------------------------------------------------------
712		Name: Matthias Blume
713		Date: 2001/01/05 00:30:00 JST
714		Tag: blume-20010105-primops
715		Description:
716
717		Added some experimental support for work that I am doing right now.
718		These changes mostly concern added primops, but there is also a new
719		experimental C library in the runtime system (but currently not enabled
720		anywhere except on Linux/X86).
721
722		In the course of adding primops (and playing with them), I discovered that
723		Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
724		broken.  (Zhong was very right he labeled this stuff as "major gross hack".)
725		To recover, I made type information in INL_PRIM mandatory and changed
726		prim.sml as well as built-in.sml accordingly.  The InLine structure now
727		has complete, correct type information (i.e., no bottom types).
728
729		Since all these changes mean that we need new binfiles, I also bumped the
730		version number to 110.32.1.
731
732		----------------------------------------------------------------------
733		Name: Matthias Blume
734		Date: 2000/12/30 22:10:00 JST
735		Tag: blume-20001230-various
736		Description:
737
738		Added proxy libraries for MLRISC and let MLRISC libraries refer
739		to each other using path anchors.  (See CM manual for explanation.)
740
741		Updated CM documentation.
742
743		Fixed some bugs in CM.
744
745		Implemented "proxy" libraries (= syntactic sugar for CM).
746
747		Added "-quiet" option to makeml and changed runtime system accordingly.
748
749		Added cleanup handler for exportML to reset timers and compiler stats.
750
751		----------------------------------------------------------------------
752		Name: Lal George
753		Date: 2000/12/22 22:22:58 EST 2000
754		Tag: Release_110_32
755		Description:
756
757		Infinite precision used throughout MLRISC.
758		see MLRISC/mltree/machine-int.sig
759
760		----------------------------------------------------------------------
761		Name: Matthias Blume
762		Date: 2000/12/22 23:16:00 JST
763		Tag: blume-20001222-warn
764		Description:
765
766		Corrected wording and formatting of some CM warning message which I
767		broke in my previous patch.
768
769		----------------------------------------------------------------------
770		Name: Matthias Blume
771		Date: 2000/12/22 21:20:00 JST
772		Tag: blume-20001222-anchorenv
773		Description:
774
775		Fixed CM's handling of anchor environments in connection with CMB.make.
776
777		----------------------------------------------------------------------
778		Name: Matthias Blume
779		Date: 2000/12/22 13:15:00 JST
780		Tag: blume-20001222-cleanup
781		Description:
782
783		Removed src/cm/ffi which does not (and did not) belong here.
784
785		----------------------------------------------------------------------
786		Name: Matthias Blume
787		Date: 2000/12/21 23:55:00 JST
788		Tag: blume-20001221-exn
789		Description:
790
791		Probably most important: CM no longer silently swallows all exceptions
792		in the compiler.
793		Plus: some other minor CM changes.  For example, CM now reports some
794		sizes for generated binfiles (code, data, envpickle, lambdapickle).
795
796		----------------------------------------------------------------------
797		Name: Matthias Blume
798		Date: 2000/12/15 00:01:05 JST
799		Tag: blume-20001215-dirtool
800		Description:
801
802		- "dir" tool added.
803		- improvements and cleanup to Tools structure
804		- documentation updates
805
806		----------------------------------------------------------------------
807		Name: Allen Leung
808		Date: Thu Dec 14 03:45:24 EST 2000
809		Description:
810		Tag:  leunga-20001214-int-inf
811		Description:
812
813		In IntInf, added these standard functions, which are missing from our
814		implementation:
815
816		andb : int * int -> int
817		xorb : int * int -> int
818		orb  : int * int -> int
819		notb : int -> int
820		<<   : int * word -> int
821		~>>  : int * word -> int
822
823		Not tested, I hope they are correct.
824
825		----------------------------------------------------------------------
826		Name: Allen Leung
827		Date: Fri Dec  8 19:23:26 EST 2000
828		Description:
829		Tag:  leunga-20001208-nowhere
830		Description:
831
832		Slight improvements to the 'nowhere' tool to handle OR-patterns,
833		to generate better error messages etc.  Plus a brief manual.
834
835		----------------------------------------------------------------------
836		Name: Lal George
837		Date: 2000/12/08 09:54:02 EST 2000
838		Tag: Release_110_31
839		Description:
840
841		- Version 110.31
842		----------------------------------------------------------------------
843		Name: Allen Leung
844		Date: Thu Dec  7 22:01:04 EST 2000
845		Tag:  leunga-20001207-cell-monster-hack
846		Description:
847
848		Major MLRISC internal changes.  Affect all clients.
849		Summary:
850
851		1.  Type CELLS.cell = int is now replaced by a datatype.
852		As a result, the old regmap is now gone.  Almost all interfaces
853		in MLRISC change as a consequence.
854
855		2.  A new brand version of machine description tool (v3.0) that generates
856		modules expecting the new interface.  The old version is removed.
857
858		3.  The RA interface has been further abstracted into two new functors.
859		RISC_RA and X86RA.  These functors have much simpler interfaces.
860		[See also directory MLRISC/demo.]
861
862		4.  Some other new source->source code generation tools are available:
863
864		a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
865		b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
866		I use this tool to generate the peephole optimizers---with the new
867		cell type changes, peephole rules are becoming difficult to write
868		without conditional pattern matching.
869
870		5.  More Intmap -> IntHashTable change.  Previous changes by Matthias didn't
871		cover the entire MLRISC source tree so many things broke.
872
873		6.  CM files have been moved to the subdirectory MLRISC/cm.
874		They are moved because there are a lot of them and they clutter up the
875		root dir.
876
877		7.  More detailed documentation to come...
878
879		NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
880		a makeml -rebuild first.  This is because of other other
881		changes that Matthias has made (see below).
882
883
884		----------------------------------------------------------------------
885		Name: Matthias Blume
886		Date: 2000/11/30 23:12:00 JST
887		Tag: blume-20001130-filereorg
888		Description:
889
890		Some manual updates and some file reorganizations in CM.
891
892		----------------------------------------------------------------------
893		Name: Matthias Blume
894		Date: 2000/11/24 17:45:00 JST
895		Tag: blume-20001124-link
896		Description:
897
898		Drastically improved link traversal code for the case that the dynamic
899		value was already loaded at bootstrap time.  As a result, CM and CMB
900		now both load blazingly fast -- even on a very slow machine.  Also,
901		memory consumption has been further reduced by this.
902
903		Warning: The format of the PIDMAP file has changed.  THerefore, to
904		bootstrap you have to do this:
905
906		1. Run CMB.make
907		2. Make a symbolic link for the boot directory:
908		ln -s sml.boot.ARCH-OS xxx
909		3. "Rebuild" the boot directory:
910		./makeml -boot xxx -rebuild sml ; rm xxx
911		4. Boot normally:
912		./makeml
913
914		----------------------------------------------------------------------
915		Name: Matthias Blume
916		Date: 2000/11/21 21:20:00 JST
917		Tag: blume-20001121-tools
918		Description:
919
920		Continued hacking on autoloading problem -- with success this time.
921		Also changed tool-plugin mechanism.  See new CM manual.
922
923		----------------------------------------------------------------------
924		Name: Matthias Blume
925		Date: 2000/11/19 14:30:00 JST
926		Tag:  blume-20001119-autoload
927		Description:
928
929		Some hacking to make autoloading faster.  Success for CMB, no success
930		so far for CM.  There is a reduced structure CM' that autoloads faster.
931		(This is a temporary, non-documented hack to be eliminated again when
932		the general problem is solved.)
933
934		----------------------------------------------------------------------
935		Name: Matthias Blume
936		Date: 2000/11/17 14:10:00 JST
937		Tag: blume-20001117-pickle-lib
938		Description:
939
940		1. Eliminated comp-lib.cm
941		2. Made pickle-lib.cm
942		3. Eliminated all uses of intset.sml (from comp-lib.cm)
943		4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
944		equivalent constructs from smlnj-lib.cm (INtHashTable).
945		5. Point 4. also goes for those uses of intmap.* in MLRISC.
946		Duplicated intmap modules thrown out.
947		6. Hunted down all duplicated SCC code and replaced it with
948		equivalent stuff (GraphSCCFn from smlnj-lib.cm).
949		7. Rewrote Feedback module.
950		8. Moved sortedlist.sml into viscomp-lib.cm.  Eventually it
951		should be thrown out and equivalent modules from smlnj-lib.cm
952		should be used (IntRedBlackSet, IntListSet, ...).
953
954		Confirmed that compiler compiles to fixpoint.
955
956		----------------------------------------------------------------------
957		Name: Allen Leung
958		Date: 2000/11/10 18:00:00
959		Tag: leunga-20001110-new-x86-fp
960
961		A new x86 floating point code generator has been added.
962		By default this is turned off.  To turn this on, do:
963
964		CM.autoload "$smlnj/compiler.cm";
965		Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
966
967		Changes:
968
969		1.  Changed FTAN to FPTAN so that the assembly output is correct.
970		2.  Changed the extension callback for FTANGENT to generate:
971
972		fptan
973		fstp  %st(0)
974		instead of
975		fptan
976		fstpl ftempmem
977
978		3.  Numerous assembly fixes for x86.
979
980		5.  Cleaned up the machine code output module x86/x86MC.sml and added
981		support for a whole bunch of instructions and addressing modes:
982
983		fadd/fsub/fsubr/fmul/fdiv/fdivr  %st, %st(n)
984		faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp  %st, %st(n)
985		fadd/fsub/fsubr/fmul/fdiv/fdivr  %st(n), %st
986		fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
987		fxch %st(n)
988		fld %st(n)
989		fst %st(n)
990		fst mem
991		fstp %st(n)
992		fucom %st(n)
993		fucomp %st(n)
994
995		All these are now generated when the fast fp mode is turned on.
996
997		6.  Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
998
999		----------------------------------------------------------------------
1000		Name: Matthias Blume
1001		Date: 2000/11/09 11:20:00 JST
1002		Tag: blume-20001109-scc
1003		Description:
1004
1005		Eliminated some code duplication:
1006
1007		1. Added "where" clause to GraphSCCFn in SML/NJ Library.
1008		(Otherwise the functor is useless.)
1009		2. Used GraphSCCFn where SCCUtilFun was used previously.
1010		3. Got rid of SCCUtilFun (in comp-lib.cm).
1011
1012		----------------------------------------------------------------------
1013		Name: Lal George
1014		Date: 2000/11/06 09:02:21 EST 2000
1015		Tag: Release_110_30
1016		Description:
1017
1018		- Version 110.30
1019		----------------------------------------------------------------------
1020		Name: Matthias Blume
1021		Date: 2000/11/04 14:45:00
1022		Tag: blume-20001104-mlbuild
1023		Description:
1024
1025		- Made ml-build faster on startup.
1026		- Documentation fixes.
1027
1028		----------------------------------------------------------------------
1029		Name: Matthias Blume
1030		Date: 2000/11/02 17:00:00 JST
1031		Tag: blume-20001102-condcomp
1032		Description:
1033
1034		- Small tweaks to pickler -- new BOOTFILES!
1035		- Version bumped to 110.29.2.
1036		- Added conditional compilation facility to init.cmi (see comment there).
1037		----------------------------------------------------------------------
1038		Name: Allen Leung
1039		Date: 2000/10/23 19:31:00
1040		Tag: leunga-20001023-demo-ra
1041
1042		1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
1043		2. Test programs for the graph library updated
1044		3. Some new MLRISC demo programs added
1045
1046		----------------------------------------------------------------------
1047		Name: Matthias Blume
1048		Date: 2000/08/31 22:15:00 JST
1049		Tag: blume-20001017-errmsg
1050		Description:
1051
1052		More error message grief: Where there used to be no messages, there
1053		now were some that had bogus error regions.  Fixed.
1054
1055		----------------------------------------------------------------------
1056		Name: Matthias Blume
1057		Date: 2000/08/31 17:30:00 JST
1058		Tag: blume-20001017-v110p29p1
1059		Description:
1060
1061		I made a version 110.29.1 with new bootfiles.
1062
1063		Changes:  Modified pickler/unpickler for faster and leaner unpickling.
1064		CM documentation changes and a small bugfix in CM's error reporting.
1065
1066		----------------------------------------------------------------------
1067		Name: Lal George
1068		Date: 2000/09/27 14:42:35 EDT
1069		Tag: george-20000927-nodestatus
1070		Description:
1071
1072		Changed the type of the nodestatus, so that:
1073
1074		SPILLED(~1)             is now SPILLED
1075		SPILLED(m) where m>=0   is now MEMREG(m)
1076		SPILLED(s) where s<~1   is now SPILL_LOC(~s)
1077
1078		----------------------------------------------------------------------
1079		Name: Matthias Blume
1080		Date: 2000/09/07 14:45:00 JST
1081		Tag: blume-20000907-cmerrmsg
1082		Description:
1083
1084		Small tweak to CM to avoid getting ML syntax error messages twice.
1085
1086		----------------------------------------------------------------------
1087		Name: Matthias Blume
1088		Date: 2000/08/31 18:00:00 JST
1089		Tag: blume-20000831-cvsbootfiles
1090		Description:
1091
1092		New URL for boot files (because the 110.29 files on the BL server do
1093		now work correctly with my updated install scripts for yacc and lex).
1094
1095		----------------------------------------------------------------------
1096		Name: Matthias Blume
1097		Date: 2000/08/08 12:33:00 JST
1098		Tag: blume-20000808-manual
1099		Description:
1100
1101		Tiny update to CM manual.
1102
1103		----------------------------------------------------------------------
1104		Name: Allen Leung
1105		Date: 2000/08/7 19:31:00
1106		Tag: leunga-20000807-a-whole-bunch-of-stuff
1107
1108		Moby, C--, SSA, x86, machine descriptions etc.  Should only affect C--
1109		and Mobdy.
1110
1111		1.  x86
1112
1113		a.  Fixes to peephole module by John and Dan.
1114		b.  Assembly fix to SETcc by Allen.
1115		c.  Fix to c-call by John.
1116		d.  Fix to spilling by John.  (This one deals with the missing FSTPT case)
1117		e.  Instruction selection optimization to SETcc as suggested by John.
1118
1119		For example,
1120
1121		MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
1122
1123		should generate:
1124
1125		MOVL a, x
1126		SUBL b, x
1127		SHRL 31, x
1128
1129		2.  IR stuff
1130
1131		A bunch of new DJ-graph related algorithms added.  These
1132		speed up SSA construction.
1133
1134		3.  SSA + Scheduling
1135
1136		Added code for SSA and scheduling to the repository
1137
1138		----------------------------------------------------------------------
1139		Name: Lal George
1140		Date: 2000/07/27 11:53:14 EDT
1141
1142		Tag: lal-20000727-linux-ppc
1143		Description:
1144
1145		Made changes to support Linux PPC.
1146		p.s. I have confirmation that the 110.29 boot files work fine.
1147
1148		----------------------------------------------------------------------
1149		Name: Matthias Blume
1150		Date: 2000/07/27 17:40:00 JST
1151		Tag: blume-20000727-scripts
1152		Description:
1153
1154		!!!! WARNING !!!!
1155		You must recompile the runtime system!
1156		!!!! WARNING !!!!
1157
1158		This is basically another round of script-enhancements:
1159
1160		1. sml, ml-build, and ml-makedepend accept options -D and -U to define
1161		and undefine CM preprocessor symbols.
1162
1163		2. ml-build avoids generating a new heap image if it finds that the
1164		existing one is still ok.  (The condition is that no ML file had to
1165		be recompiled and all ML files are found to be older that the heap
1166		file.)
1167
1168		To make this work smoothly, I also hacked the runtime system as
1169		well as SMLofNJ.SysInfo to get access to the heap image suffix
1170		(.sparc-solaris, ...) that is currently being used.
1171
1172		Moreover, the signature of CM.mk_standalone has changed.  See the
1173		CM manual.
1174
1175		3. ml-makedepend accepts additional options -n, -a, and -o.  (See the
1176		CM manual for details.)
1177
1178		4. More CM manual updates:
1179		- all of the above has been documented.
1180		- there is now a section describing the (CM-related) command line
1181		arguments that are accepted by the "sml" command
1182
1183		----------------------------------------------------------------------
1184		Name: Matthias Blume
1185		Date: 2000/07/25 16:20:00 JST
1186		Tag: blume-20000725-makedepend
1187		Description:
1188
1189		Added a script called ml-makedepend.  This can be used in makefiles
1190		for Unix' make in a way very similar to the "makedepend" command for
1191		C.
1192
1193		The script internally uses function CM.sources.
1194
1195		Synopsis:
1196
1197		ml-makedepend [-f makefile] cmfile targetname
1198
1199		The default for the makefile is "makefile" (or "Makefile" should
1200		"makefile" not exist).
1201
1202		ml-makedepend adds a cmfile/targetname-specific section to this
1203		makefile (after removing the previous version of this section).  The
1204		section contains a single dependency specification with targetname on
1205		the LHS (targetname is an arbitrary name), and a list of files derived
1206		from the cmfile on the RHS.  Some of the files on the RHS are
1207		ARCH/OPSYS-specific.  Therefore, ml-makedepend inserts references to
1208		"make" variables $(ARCH) and $(OPSYS) in place of the corresponding
1209		path names.  The makefile writer is responsible for making sure that
1210		these variables have correct at the time "make" is invoked.
1211
1212		----------------------------------------------------------------------
1213		Name: Matthias Blume
1214		Date: 2000/07/22 23:30:00 JST
1215		Tag: blume-20000722-urlupdate
1216		Description:
1217
1218		Changed BOOT and config/srcarchiveurl to point to BL server:
1219
1220		ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
1221
1222		----------------------------------------------------------------------
1223		Name: Matthias Blume
1224		Date: 2000/07/18 18:00:00 JST
1225		Tag: blume-20000718-Version_110_29
1226		Description:
1227
1228		1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
1229
1230		2. Updated config/version to 110.29
1231
1232		3. Updated config/srcarchiveurl
1233
1234		3. New boot files!
1235		ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
1236
1237		----------------------------------------------------------------------
1238		Name: Matthias Blume
1239		Date: 2000/07/11 13:58:00 JST
1240		Tag: blume-20000711-doctypo
1241		Description:
1242
1243		Fixed a few typos in CM manual.
1244
1245		----------------------------------------------------------------------
1246		Name: Allen Leung
1247		Date: 2000/06/15 00:38:00
1248		Tag: leunga-20000704-sparc-x86
1249
1250		1. x86 peephole improvement sp += k; sp -= k => nop  [from John]
1251		2. fix to x86 RET bug [found by Dan Grossman]
1252		3. sparc assembly bug fix for ticc instructions [found by Fermin]
1253
1254		Affects c-- and moby only
1255
1256		----------------------------------------------------------------------
1257		Name: Matthias Blume
1258		Date: 2000/07/04 15:26:00
1259		Tag: blume-20000704-trigger
1260		Description:
1261
1262		1. Improvements to CM manual.
1263		2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
1264		of getting a back-trace.  The function, when called, raises an
1265		internal exception which explicitly carries the full back-trace history,
1266		so it is unaffected by any intervening handle-raise pairs ("trivial"
1267		or not).  The interactive loop will print that history once it arrives
1268		at top level.
1269		Short of having all exceptions implicitly carry the full history, the
1270		recommended way of using this facility is:
1271		- compile your program with instrumentation "on"
1272		- run it, when it raises an exception, look at the history
1273		- if the history is "cut off" because of some handler, go and modify
1274		your program so that it explicitly calls BTrace.trigger
1275		- recompile (still instrumented), and rerun; look at the full history
1276
1277		----------------------------------------------------------------------
1278		Name: Matthias Blume
1279		Date: 2000/07/03 15:36:00 JST
1280		Tag: blume-20000702-manual
1281		Description:
1282
1283		Small corrections and updates to CM manual.
1284
1285		----------------------------------------------------------------------
1286		Name: Matthias Blume
1287		Date: 2000/06/29 16:04:00 JST
1288		Tag: blume-20000629-yacctool
1289		Description:
1290
1291		Changes:
1292
1293		1. Class "mlyacc" now takes separate arguments to pass options to
1294		generated .sml- and .sig-files independently.
1295		2. Corresponding CM manual updates.
1296		3. BTrace module now also reports call sites.  (However, for loop clusters
1297		it only shows from where the cluster was entered.)  There are associated
1298		modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
1299
1300		----------------------------------------------------------------------
1301		Name: Matthias Blume
1302		Date: 2000/06/27 16:51:00 JST
1303		Tag: blume-20000627-noweb
1304		Description:
1305
1306		Changes:
1307
1308		1. Implemented "subdir" and "witness" options for noweb tool.
1309		This caused some slight internal changes in CM's tool implementation.
1310		2. Fixed bug in "tool plugin" mechanism.  This is essentially cleaning
1311		some remaining issues from earlier path anchor changes.
1312		3. Updated CM manual accordingly.
1313
1314		4. Changed implementation of back-tracing so that I now consider it
1315		ready for prime-time.
1316
1317		In particular, you don't have to explicitly trigger the back-trace
1318		anymore.  Instead, if you are running BTrace-instrumented code and
1319		there is an uncaught exception (regardless of whether or not it was
1320		raised in instrumented code), the top-level evalloop will print
1321		the back-trace.
1322
1323		Features:
1324
1325		- Instrumented and uninstrumented code work together seemlessly.
1326		(Of course, uninstrumented code is never mentioned in actual
1327		back-traces.)
1328
1329		- Asymptotic time- and space-complexity of instrumented code is
1330		equal to that of uninstrumented code.  (This means that
1331		tail-recursion is preserved by the instrumentation phase.)
1332
1333		- Modules whose code has been instrumented in different sessions
1334		work together without problem.
1335
1336		- There is no penalty whatsoever on uninstrumented code.
1337
1338		- There is no penalty on "raise" expressions, even in
1339		instrumented code.
1340
1341		A potential bug (or perhaps it is a feature, too):
1342
1343		A back-trace reaches no further than the outermost instrumented
1344		non-trivial "raise".  Here, a "trivial" raise is one that is the
1345		sole RHS of a "handle" rule.  Thus, back-traces reach trough
1346
1347		<exp> handle e => raise e
1348
1349		and even
1350
1351		<exp> handle Foo => raise Bar
1352
1353		and, of course, through
1354
1355		<exp> handle Foo => ...
1356
1357		if the exception was not Foo.
1358
1359		Back-traces always reach right through any un-instrumented code
1360		including any of its "handle" expressions, trivial or not.
1361
1362		To try this out, do the following:
1363
1364		- Erase all existing binfiles for your program.
1365		(You may keep binfiles for those modules where you think you
1366		definitely don't need back-tracing.)
1367		- Turn on back-trace instrumentation:
1368		SMLofNJ.Internals.BTrace.mode (SOME true);
1369		- Recompile your program.  (I.e., run "CM.make" or "use".)
1370		- You may now turn instrumentation off again (if you want):
1371		SMLofNJ.Internals.BTrace.mode (SOME false);
1372		- Run your program as usual.  If it raises an exception that
1373		reaches the interactive toplevel, then a back-trace will
1374		automatically be printed.  After that, the toplevel loop
1375		will print the exception history as usual.
1376
1377		----------------------------------------------------------------------
1378		Name: Matthias Blume
1379		Date: 2000/06/26 09:56:46 JST
1380		Tag: blume-20000626-setup
1381		Description:
1382
1383		CM: - setup-parameter to "sml" added; this can be used to run arbitrary
1384		ML code before and after compiling a file (e.g., to set compiler
1385		flags)
1386
1387		Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
1388		- associated changes to btrace.sml (BTrace instrumentation pass)
1389		- cleaner implementation of btimp.sml (BTrace tracing and report
1390		module)
1391
1392		CM manual: * new path encoding documented
1393		* description of setup-parameter to "sml" added
1394
1395		The biggest user-visible change to back-tracing is that it is no
1396		longer necessary to compile all traced modules within the same
1397		session.  (This was a real limitation.)
1398
1399		----------------------------------------------------------------------
1400		Name: Matthias Blume
1401		Date: 2000/06/24 12:40:00 JST
1402		Tag: blume-20000624-startup
1403		Description:
1404
1405		Fixes startup slowdown problem.  (I was calling SrcPath.sync a _tad_
1406		bit too often -- to put it mildly. :)
1407
1408		----------------------------------------------------------------------
1409		Name: Matthias Blume
1410		Date: 2000/06/23 18:20:00 JST
1411		Tag: blume-20000623-btrace
1412		Description:
1413
1414		This updates adds a backtrace facility to aid programmers in debugging
1415		their programs.  This involves the following changes:
1416
1417		1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
1418		keeping track of the current call stack.  When programs are compiled
1419		in a special mode, the compiler will insert calls to these hooks
1420		into the user program.
1421		"Hook" means that it is possible for different implementations of
1422		back-tracing to register themselves (at different times).
1423
1424		2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
1425		as an Absyn.dec->Absyn.dec rewrite.  Normally this phase is turned off.
1426		It can be turned on using this call:
1427		SMLofNJ.Internals.BTrace.mode (SOME true);
1428		Turning it off again:
1429		SMLofNJ.Internals.BTrace.mode (SOME false);
1430		Querying the current status:
1431		SMLofNJ.Internals.BTrace.mode NONE;
1432		Annotated programs are about twice as big as normal ones, and they
1433		run a factor of 2 to 4 slower with a dummy back-trace plugin (one
1434		where all hooks do nothing).  The slowdown with a plugin that is
1435		actually useful (such as the one supplied by default) is even greater,
1436		but in the case of the default plugin it is still only an constant
1437		factor (amortized).
1438
1439		3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
1440		with a sub-structure BTrace for controlling back-tracing.  In particular,
1441		the above-mentioned function "mode" controls whether the annotation
1442		phase is invoked by the compiler.  Another important function is
1443		"trigger": when called it aborts the current execution and causes
1444		the top-level loop to print a full back-trace.
1445
1446		4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
1447		for back-tracing.  It keeps track of the dynamic call stack and in
1448		addition to that it keeps a partial history at each "level" of that
1449		stack.  For example, if a tail-calls b, b tail-calls c, and c tail-calls
1450		d and b (at separate times, dynamically), then the report will show:
1451
1452		GOTO   d
1453		/c
1454		GOTO  \b
1455		CALL   a
1456
1457		This shows that there was an initial non-tail call of a, then a
1458		tail-call to b or c, looping behavior in a cluster of functions that
1459		consist of b and c, and then a goto from that cluster (i.e., either from
1460		b or from c) to d.
1461
1462		Note that (depending on the user program) the amount of information
1463		that the back-trace module has to keep track of at each level is bounded
1464		by a constant.  Thus, the whole implementation has the same asymptotical
1465		complexity as the original program (both in space and in time).
1466
1467		5. compiler/TopLevel/interact/evalloop.sml has been modified to
1468		handle the special exception SMLofNJ.Internals.BTrace.BTrace
1469		which is raised by the "trigger" function mentioned above.
1470
1471		Notes on usage:
1472
1473		- Annotated code works well together with unannotated code:
1474		Unannotated calls simply do not show up at all in the backtrace.
1475
1476		- It is not a good idea to let modules that were annotated during
1477		different sessions run at the same time.  This is because the compiler
1478		chooses small integers to identify individual functions, and there
1479		will be clashes if different modules were compiled in separate sessions.
1480		(Nothing will crash, and you will even be told about the clashes, but
1481		back-trace information will in general not be useful.)
1482
1483		- Back-tracing can be confused by callcc and capture.
1484
1485		- The only way of getting a back-trace right now is to explicitly
1486		invoke the "trigger" function from your user program.  Eventually, we
1487		should make every exception carry back-trace information (if
1488		available).  But since this creates more overhead at "raise"-time
1489		(similar to the current exnHistory overhead), I have not yet
1490		implemented this.  (The implementation will be rather easy.)  With
1491		exceptions carrying back-trace information, this facility will be even
1492		more useful because users don't need to modify their programs...
1493
1494		- While it is possible to compile the compiler with back-trace
1495		annotations turned on (I did it to get some confidence in
1496		correctness), you must make absolutely sure that core.sml and
1497		btimp.sml are compiled WITHOUT annotation!  (core.sml cannot actually
1498		be compiled with annotation because there is no core access yet, but
1499		if you compile btimp.sml with annotation, then the system will go into
1500		an infinite recursion and crash.)
1501		Since CM currently does not know about BTrace, the only way to turn
1502		annotations on and off for different modules of the compiler is to
1503		interrupt CMB.make, change the settings, and re-invoke it.  Of course,
1504		this is awkward and clumsy.
1505
1506		Sample sessions:
1507
1508		Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
1509		- SMLofNJ.Internals.BTrace.mode (SOME true);
1510		[autoloading]
1511		[autoloading done]
1512		val it = false : bool
1513		- structure X = struct
1514		-     fun main n = let
1515		-         fun a (x, 0) = d x
1516		-           | a (x, n) = b (x, n - 1)
1517		-         and b (x, n) = c (x, n)
1518		-         and c (x, n) = a (x, n)
1519		-         and d x = e (x, 3)
1520		-         and e (x, 0) = f x
1521		-           | e (x, n) = e (x, n - 1)
1522		-         and f 0 = SMLofNJ.Internals.BTrace.trigger ()
1523		-           | f n = n * g (n - 1)
1524		-         and g n = a (n, 3)
1525		-     in
1526		-         f n
1527		-     end
1528		- end;
1529		structure X : sig val main : int -> int end
1530		- X.main 3;
1531		*** BACK-TRACE ***
1532		GOTO   stdIn:4.2-13.20: X.main[2].f
1533		GOTO-( stdIn:4.2-13.20: X.main[2].e
1534		GOTO   stdIn:4.2-13.20: X.main[2].d
1535		/ stdIn:4.2-13.20: X.main[2].a
1536		| stdIn:4.2-13.20: X.main[2].b
1537		GOTO-\ stdIn:4.2-13.20: X.main[2].c
1538		CALL   stdIn:4.2-13.20: X.main[2].g
1539		GOTO   stdIn:4.2-13.20: X.main[2].f
1540		GOTO-( stdIn:4.2-13.20: X.main[2].e
1541		GOTO   stdIn:4.2-13.20: X.main[2].d
1542		/ stdIn:4.2-13.20: X.main[2].a
1543		| stdIn:4.2-13.20: X.main[2].b
1544		GOTO-\ stdIn:4.2-13.20: X.main[2].c
1545		CALL   stdIn:4.2-13.20: X.main[2].g
1546		GOTO   stdIn:4.2-13.20: X.main[2].f
1547		GOTO-( stdIn:4.2-13.20: X.main[2].e
1548		GOTO   stdIn:4.2-13.20: X.main[2].d
1549		/ stdIn:4.2-13.20: X.main[2].a
1550		| stdIn:4.2-13.20: X.main[2].b
1551		GOTO-\ stdIn:4.2-13.20: X.main[2].c
1552		CALL   stdIn:4.2-13.20: X.main[2].g
1553		GOTO   stdIn:4.2-13.20: X.main[2].f
1554		CALL   stdIn:2.15-17.4: X.main[2]
1555		-
1556
1557		(Note that because of a FLINt bug the above code currently does not
1558		compile without BTrace turned on.)
1559
1560		Here is another example, using my modified Tiger compiler:
1561
1562		Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
1563		- SMLofNJ.Internals.BTrace.mode (SOME true);
1564		[autoloading]
1565		[autoloading done]
1566		val it = false : bool
1567		- CM.make "sources.cm";
1568		[autoloading]
1569		...
1570		[autoloading done]
1571		[scanning sources.cm]
1572		[parsing (sources.cm):parse.sml]
1573		[creating directory CM/SKEL ...]
1574		[parsing (sources.cm):tiger.lex.sml]
1575		...
1576		[wrote CM/sparc-unix/semant.sml]
1577		[compiling (sources.cm):main.sml]
1578		[wrote CM/sparc-unix/main.sml]
1579		[New bindings added.]
1580		val it = true : bool
1581		- Main.compile ("../testcases/merge.tig", "foo.out");
1582		*** BACK-TRACE ***
1583		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
1584		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1585		CALL   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
1586		GOTO   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
1587		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1588		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1589		CALL   lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
1590		/ lib/semant.sml:411.65-543.8: SemantFun[2].transDec
1591		CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
1592		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1593		CALL   lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
1594		CALL   main.sml:1.18-118.4: Main.compile[2]
1595		-
1596
1597		----------------------------------------------------------------------
1598		Name: Matthias Blumen
1599		Date: 2000/06/21 18:00:00 JST
1600		Tag: blume-20000621-manual
1601		Description:
1602
1603		CM manual update: Path environments documented.
1604
1605		----------------------------------------------------------------------
1606		Name: Matthias Blume
1607		Date: 2000/06/19 13:40:00
1608		Tag: blume-20000619-manual
1609		Description:
1610
1611		CM manual and system/README update.  This only covers the fact that
1612		there are no more implicit anchors.  (Path environments and the "bind"
1613		option to "cm" have yet to be documented.)
1614
1615		----------------------------------------------------------------------
1616		Name: Matthias Blume
1617		Date: 2000/06/19 11:05:00 JST
1618		Tag: blume-20000619-chdir-bugfix
1619		Description:
1620
1621		Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
1622
1623		----------------------------------------------------------------------
1624		Name: Matthias Blume
1625		Date: 2000/06/18 22:00:10 JST
1626		Tag: blume-20000618-implicit-anchors-really-gone
1627		Description:
1628
1629		I updates the previous HISTORY entry where I forgot to mention that
1630		implicit anchors are no longer with us.
1631
1632		The current update also gets rid of the (now useless) controller
1633		CM.Control.implicit_anchors.
1634
1635		----------------------------------------------------------------------
1636		Name: Matthias Blume
1637		Date: 2000/06/16 17:30:00 JST
1638		Tag: blume-20000616-anchorenv
1639		Description:
1640
1641		This patch implements the long anticipated (just kidding :) "anchor
1642		environment" mechanism.  In the course of doing this, I also
1643		re-implemented CM's internal "SrcPath" module from scratch.  The new
1644		one should be more robust in certain boundary cases.  In any case, it
1645		is a lot cleaner than its predecessor (IMHO).
1646
1647		This time, although there is yet another boot file format change, I
1648		kept the unpickler backward-compatible.  As a result, no new bootfiles
1649		are necessary and bootstrapping is straightforward.  (You cannot read
1650		new bootfiles into an old system, but the other way around is no
1651		problem.)
1652
1653		Visible changes:
1654
1655		** 0. Implicit path anchors (without the leading $-symbol) are no
1656		longer recognized at all. This means that such path names are not
1657		illegal either.  For example, the name basis.cm simply refers to a
1658		local file called "basis.cm" (i.e, the name is an ordinary path
1659		relative to .cm-files directory).  Or, to put it differently, only
1660		names that start with $ are anchored paths.
1661
1662		** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
1663		vanished.
1664
1665		John (Reppy) had critizised this as soon as I originally proposed and
1666		implemented it, but at that time I did not really deeply believe
1667		him. :) Now I came full-circle because I need the $<singlearc> syntax
1668		in another place where it cannot be seen as an abbreviation for
1669		$/<singlearc>.  To avoid the confusion, $<singlearc> now means what it
1670		seems to mean (i.e., it "expands" into the corresponding anchor
1671		value).
1672
1673		However, when paths are used as members in CM description files, it
1674		continues to be true that there must be at least another arc after the
1675		anchor.  This is now enforced separately during semantic analysis
1676		(i.e., from a lexical/syntactical point of view, the notation is ok.)
1677
1678		** 2. The "cm" class now accepts an option "bind".  The option's value
1679		is a sub-option list of precisely two items -- one labeled "anchor"
1680		and the other one labeled "value".  As you might expect, "anchor" is
1681		used to specify an anchor name to be bound, and "value" specifies what
1682		the anchor is being bound to.
1683
1684		The value must be a directory name and can be given in either standard
1685		syntax (including the possibility that it is itself an anchored path)
1686		or native syntax.
1687
1688		Examples:
1689
1690		foo.cm (bind:(anchor:bar value:$mystuff/bar))
1691		lib.cm (bind:(anchor:a value:"H:\\x\\y\\z"))  (* only works under windows *)
1692
1693		and so on.
1694
1695		The meaning of this is that the .cm-file will be processed with an
1696		augmented anchor environment where the given anchor(s) is/are bound to
1697		the given values(s).
1698
1699		The rationale for having this feature is this: Suppose you are trying
1700		to use two different (already stable) libraries a.cm and b.cm (that
1701		you perhaps didn't write yourself).  Further, suppose each of these
1702		two libraries internally uses its own auxiliary library $aux/lib.cm.
1703		Normally you would now have a problem because the anchor "lib" can not
1704		be bound to more than one value globally.  Therefore, the project that
1705		uses both a.cm and b.cm must locally redirect the anchor to some other
1706		place:
1707
1708		a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
1709		b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
1710
1711		This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
1712		/usr/lib/smlnj/b-stuff/aux.cm, respectively.
1713
1714		Hard-wiring path names is a bit inflexible (and CM will verbosely warn
1715		you when you do so at the time of CM.stabilize).  Therefore, you can
1716		also use an anchored path as the value:
1717
1718		a.cm (bind:(anchor:lib value:$a-lib))
1719		b.cm (bind:(anchor:lib value:$b-lib))
1720
1721		Now you can globally configure (using the usual CM.Anchor.anchor or
1722		pathconfig machinery) bindings for "a-lib" and "b-lib".  Since "lib"
1723		itself is always locally bound, setting it globally is no longer
1724		meaningful or necessary (but it does not hurt either).  In fact, "lib"
1725		can still be used as a global anchor for separate purposes.  As a
1726		matter of fact, one can locally define "lib" in terms of a global
1727		"lib":
1728
1729		a.cm (bind:(anchor:lib value:$lib/a))
1730		b.cm (bind:(anchor:lib value:$lib/b))
1731
1732		** 3: The encoding of path names has changed.  This affects the way
1733		path names are shown in CM's progress report and also the internal
1734		protocol encoding used for parallel make.
1735
1736		The encoding now uses one or more ':'-separated segments.  Each
1737		segments corresponds to a file that has been specified relative to the
1738		file given by its preceding segment.  The first segment is either
1739		relative to the CWD, absolute, or anchored.  Each segment itself is
1740		basically a Unix pathname; all segments but the first are relative.
1741
1742		Example:
1743
1744		$foo/bar/baz.cm:a/b/c.sml
1745
1746		This path denotes the file bar/a/b/c.sml relative to the directory
1747		denoted by anchor "foo".  Notice that the encoding also includes
1748		baz.cm which is the .cm-file that listed a/b/c.sml.  As usual, such
1749		paths are resolved relative to the .cm-files directory, so baz.cm must
1750		be ignored to get the "real" pathname.
1751
1752		To make this fact more obvious, CM puts the names of such "virtual
1753		arcs" into parentheses when they appear in progress reports. (No
1754		parentheses will appear in the internal protocol encoding.)  Thus,
1755		what you really see is:
1756
1757		$foo/bar/(baz.cm):a/b/c.sml
1758
1759		I find this notation to be much more informative than before.
1760
1761		Another new feature of the encoding is that special characters
1762		including parentheses, colons, (back)slashes, and white space are
1763		written as \ddd (where ddd is the decimal encoding of the character).
1764
1765		*** The CM manual still needs to be updated.
1766
1767		----------------------------------------------------------------------
1768		Name: Allen Leung
1769		Date: 2000/06/15 00:38:00
1770		Tag: leunga-20000615-x86-peephole
1771
1772		x86 Peephole fix by Fermin.  Affects c-- and moby only.
1773
1774		----------------------------------------------------------------------
1775		Name: Matthias Blume
1776		Date: 2000/06/12 11:40:00
1777		Tag: blume-20000612-parmakefix
1778		Description:
1779
1780		More cleanup after changing the file naming scheme: This time I
1781		repaired the parallel make mechanism for CMB.make which I broke earlier.
1782
1783		----------------------------------------------------------------------
1784		Name: Allen Leung
1785		Date: 2000/06/09 01:25:00
1786		Tag: leunga-20000609-various
1787
1788		None of these things should affect normal SML/NJ operations
1789
1790		1. Peephole improvements provided by Fermin (c--)
1791		2. New annotation DEFUSE for adding extra dependence (moby)
1792		3. New X86 LOCK instructions (moby)
1793		4. New machine description language for reservation tables (scheduling)
1794		5. Fixes to various optimization/analysis modules (branch chaining, dominator
1795		trees etc.)
1796		6. I've changed the CM files so that they can work with versions
1797		110.0.6, 110.25 and 110.28
1798
1799		----------------------------------------------------------------------
1800		Name: Matthias Blume
1801		Date: 2000/06/09 12:40:00
1802		Tag: blume-20000609-log
1803		Description:
1804
1805		- Removed all(?) remaining RCS Log entries from sources.
1806
1807		- Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
1808		anchored paths).
1809
1810		----------------------------------------------------------------------
1811		Name: Matthias Blume
1812		Date: 2000/06/07 17:00:00 JST
1813		Tag: blume-20000607-no-implicit-anchors
1814		Description:
1815
1816		1. This update changes the default setting for
1817		CM.Control.implicit_anchors from true to false.  This means that
1818		implicit anchors are no longer permitted by default.  I also tried to
1819		make sure that nothing else still relies on implicit anchors.
1820		(This is the next step on the schedule towards a CM that does not even
1821		have the notion of implicit anchors anymore.)
1822
1823		2. More CM manual updates.
1824
1825		3. I managed to track down and fix the pickling bug I mentioned last
1826		time.  Because of the previously existing workaround, this entails no
1827		immediate practical changes.
1828
1829		----------------------------------------------------------------------
1830		Name: Matthias Blume
1831		Date: 2000/06/06 11:15:00 JST
1832		Tag: blume-20000606-lazierpickle
1833		Description:
1834
1835		!!!! NEW BOOT FILES !!!!
1836
1837		* The main purpose of this update is to make library pickles lazier in
1838		order to reduce the initial space penalty for autoloading a library.
1839		As a result, it is now possible to have $smlnj/compiler.cm
1840		pre-registered.  This should take care of the many complaints or
1841		inquiries about missing structure Compiler.  This required changes to
1842		CM's internal data structures and small tweaks to some algorithms.
1843
1844		As a neat additional effect, it is no longer necessary (for the sake
1845		of lean heap image files) to distinguish between a "minimal" CM and a
1846		"full" CM.  Now, there is only one CM (i.e., the "full" version:
1847		$smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
1848		interactive top level. ($smlnj/cm/minimal.cm is gone.)
1849
1850		To make the life of compiler-hackers easier, "makeml" now also
1851		pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm).  In other
1852		words, after you bootstrap a new sml for the first time, you will not
1853		have to autoload $smlnj/cmb.cm again afterwards.  (The first time
1854		around you will still have to do it, though.)
1855
1856		* A second change consists of major updates to the CM manual.  There
1857		are now several appendices with summary information and also a full
1858		specification of the CM description file syntax.
1859
1860		* In directory src/system I added the script "allcross".  This script
1861		invokes sml and cross-compiles the compiler for all supported
1862		architectures.  (Useful when providing a new set of boot files.)
1863
1864		* There seems to be a latent bug in my "lazy pickles" mechanism.  I
1865		added a small tweak to pickle-util.sml to work around this problem,
1866		but it is not a proper fix yet.  I will investigate further.  (The
1867		effect of the bug was an inflation of library pickle size.)
1868
1869		* Version number increased to 110.28.1 (to avoid compatibility problems).
1870
1871		----------------------------------------------------------------------
1872		Name: Allen Leung
1873		Date: 2000/05/25 17:28 EDT
1874		Tag: leunga-20000525-ra
1875		Description:
1876
1877		Fixed a bug in freezing phase of the register allocator.
1878
1879		----------------------------------------------------------------------
1880		Name: Allen Leung
1881		Date: 2000/05/15 22:53 EDT
1882		Tag: leunga-20000515-alpha-x86-ra
1883		Description:
1884
1885		1. Alpha
1886
1887		Slight cleanup.  Removed the instruction SGNXL
1888
1889		2. X86
1890
1891		Added the following instructions to the instruction set:
1892
1893		ROLx, RORx,
1894		BTx, BTSx, BTLx, BTRx,
1895		XCHGx, and variants with the LOCK prefix
1896
1897		3. Register Allocation
1898
1899		The module ra-rewrite-with-renaming has been improved.
1900
1901		These have no effect on SML/NJ.
1902
1903		----------------------------------------------------------------------
1904		Name: Matthias Blume
1905		Date: 2000/05/15 16:20:00 JST
1906		Tag: blume-20000515-lightrebuild
1907		Description:
1908
1909		1. I added an alternative to "-rebuild" to "makeml".  The difference is
1910		that prior to calling CMB.make' the CM-variable "LIGHT" will be
1911		defined.  In effect, the command will not build any cross-compiler
1912		backends and therefore finish more quickly.
1913
1914		The "fixpt" script also takes a "-light" switch to be able to use
1915		this new facility while compiling for a fixpoint.
1916
1917		2. I replaced all mentions of anchored paths in group owner specifications
1918		with simple relative paths (usually starting with "..").
1919		The rationale is that a library's internal workings should not be
1920		compromised by the lack of some anchor.  (An anchor is necessary
1921		for someone who wants to refer to the library by an anchored path,
1922		but it should not be necessary to build the same library in the first
1923		place.)
1924
1925		3. I changed the way CM's tool mechanism determines the shell command
1926		string used for things like ml-yacc etc. so that it does not break
1927		when CM.Control.implicit_anchors is turned off.
1928
1929		----------------------------------------------------------------------
1930		Name: Matthias Blume
1931		Date: 2000/05/12 18:20:00 JST
1932		Tag: blume-20000512-ml-build
1933		Description:
1934
1935		Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
1936		getting installed properly (by config/install.sh).
1937
1938		----------------------------------------------------------------------
1939		Name: Matthias Blume
1940		Date: 2000/05/12 17:30:00 JST
1941		Tag: blume-20000512-anchors
1942		Description:
1943
1944		!!! NEW BOOT FILES !!!
1945
1946		This change is in preparation of fading out support for "implicitly
1947		anchored path names".  I went through all sources and used the
1948		explicit (and relatively new) $-notation.  See system/README and the
1949		CM manual for more info on this.
1950
1951		I also modified the anchoring scheme for some things such as "smlnj",
1952		"MLRISC", "cm", etc. to take advantage of the fact that explicit
1953		anchors are more expressive: anchor name and first arc do not have to
1954		coincide.  This entails the following user-visible change:
1955
1956		You have to write $smlnj/foo/bar instead of smlnj/foo/bar.  In
1957		particular, when you fire up sml with a command-line argument, say,
1958		e.g.:
1959
1960		sml '$smlnj/cmb.cm'
1961
1962		At the ML toplevel prompt:
1963
1964		CM.autoload "$smlnj/cmb.cm";
1965
1966		There is also a new controller in CM.Control that can be used to turn
1967		off all remaining support for implicit anchors by saying:
1968
1969		CM.autoload "$smlnj/
1970		#set CM.Control.implicit_anchors false;
1971
1972		This causes CM to reject implicitly anchored paths.  This is (for the
1973		time being) less permissive than the "final" version where there will
1974		be no more such implicit anchors and relative paths will be just that:
1975		relative.
1976
1977		The next step (version after next version?) will be to make the
1978		default for CM.Control.implicit_anchors false.  After the dust has
1979		settled, I can then produce the "final" version of this...
1980
1981		Note: Since bootstrapping is a bit tricky, I provided new boot files.
1982
1983		----------------------------------------------------------------------
1984		Name: Matthias Blume
1985		Date: 2000/05/11 16:30:00 JST
1986		Tag: blume-20000511-sources
1987		Description:
1988
1989		The main change is that I added function CM.sources as a generalized
1990		version of the earlier CM.makedepend.  This entails the following
1991		additional changes:
1992
1993		- CM.makedepend has been dropped.
1994
1995		- CM manual has been updated.
1996
1997		- TOOLS signature and API have been changed.
1998
1999		----------------------------------------------------------------------
2000		Name: Allen Leung
2001		Date: 2000/05/10 21:17 EDT
2002		Tag: leunga-20000510-moby-c--ssa
2003		Description:
2004
2005		Various bug fixes and new features for C--, Moby and MLRISC optimizations.
2006		None of these affect SML/NJ.
2007
2008		1. Register Allocation
2009
2010		a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
2011		This module tries to remove local (i.e. basic block level) redundancies
2012		during spilling.
2013
2014		b. A new framework for performing region based register allocation.
2015		Not yet entirely functional.
2016
2017		2. X86
2018
2019		a. DefUse for POP was missing the stack pointer [found by Lal]
2020		b. Reload for CALL was incorrect in X86Spill [found by John]
2021		c. Various fixes in X86Spill so that it can be used correctly for
2022		the new spilling module.
2023
2024		3. SSA/IR
2025
2026		a. New module ir/dj-dataflow.sml implements elimination based
2027		data flow analysis.
2028
2029		4. MLRiscGen
2030
2031		a. Fix for gc type annotation
2032
2033		5. MDGen
2034
2035		Various fixes for machine description -> ml code translation.  For ssa
2036		only.
2037
2038		----------------------------------------------------------------------
2039		Name: Allen Leung
2040		Date: 2000/05/08 22:17 EDT
2041		Tag: leunga-20000508-labexp
2042		Description:
2043
2044		Fermin has found a few assembly problems with constant expressions
2045		generated in LabelExp.  Mostly, the problems involve extra parentheses,
2046		which choke on dumb assemblers.  This is his fix.
2047
2048		----------------------------------------------------------------------
2049		Name: Dave MacQueen
2050		Date: 2000/04/09 14:00 EDT
2051		Tag: dbm-20000502-Version_110_28
2052		Description:
2053
2054		1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
2055
2056		2. Updated config/version to 110.28
2057
2058		3. Updated config/srcarchiveurl
2059
2060		3. New boot files!
2061		ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
2062
2063		----------------------------------------------------------------------
2064		Name: Matthias Blume
2065		Date: 2000/05/01 19:05:00 JST
2066		Tag: blume-20000501-noweb
2067		Description:
2068
2069		A new noweb tool has been added.  The existing system is entirely
2070		unaffected by this, but some CM users have asked for renewed noweb
2071		support.  Everything is documented in the CM manual.
2072
2073		New (plugin) libraries:
2074
2075		noweb-tool.cm
2076		nw-ext.cm
2077
2078	----------------------------------------------------------------------	----------------------------------------------------------------------
2079	Name: Dave MacQueen	Name: Dave MacQueen
2080	Date: 2000/04/30 12:40PM EDT	Date: 2000/04/30 12:40PM EDT
#	Line 295	Line 2359
2359
2360	3.  Assembly	3.  Assembly
2361
2362	When generating assemby, resolve the value of client defined constants,	When generating assembly, resolve the value of client defined constants,
2363	instead of generating symbolic values.  This is controlled by the	instead of generating symbolic values.  This is controlled by the
2364	new flag "asm-resolve-constants", which is default to true.	new flag "asm-resolve-constants", which is default to true.
2365
#	Line 318	Line 2382
2382
2383	To this end, I arranged that instead of "structure Core" as "structure	To this end, I arranged that instead of "structure Core" as "structure
2384	_Core" is bound in the pervasive environment.  Core access is done via	_Core" is bound in the pervasive environment.  Core access is done via
2385	_Core (which can never be accidentially rebound because _Core is not a	_Core (which can never be accidentally rebound because _Core is not a
2386	legal surface-syntax symbol).	legal surface-syntax symbol).
2387
2388	The current solution is much cleaner because the core environment is	The current solution is much cleaner because the core environment is
#	Line 328	Line 2392
2392	with dynamic and symbolic parts of the core environment.	with dynamic and symbolic parts of the core environment.
2393
2394	Remaining hackery (to bind the "magic" symbol _Core) is localized in the	Remaining hackery (to bind the "magic" symbol _Core) is localized in the
2395	compilation mananger's bootstrap compiler (actually: in the "init group"	compilation manager's bootstrap compiler (actually: in the "init group"
2396	handling).  See the comments in src/system/smlnj/init/init.cmi for	handling).  See the comments in src/system/smlnj/init/init.cmi for
2397	more details.	more details.
2398
#	Line 443	Line 2507
2507	(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
2508	used).	used).
2509
2510	This was done to accomodate the new "make" and "shell" tools which	This was done to accommodate the new "make" and "shell" tools which
2511	facilitate fairly seemless hookup to portions of code managed using	facilitate fairly seamless hookup to portions of code managed using
2512	Makefiles or Shell scripts.	Makefiles or Shell scripts.
2513
2514	There are no classes "shared" or "private" anymore.  Instead, the	There are no classes "shared" or "private" anymore.  Instead, the
#	Line 456	Line 2520
2520
2521	All existing tools are described in the CM manual.	All existing tools are described in the CM manual.
2522
2523	- Slightly better error handling.  (CM now surpresses many followup	- Slightly better error handling.  (CM now suppresses many followup
2524	error messages that tended to be more annoying than helpful.)	error messages that tended to be more annoying than helpful.)
2525
2526	2. Major changes to the compiler's static environment data structures.	2. Major changes to the compiler's static environment data structures.
#	Line 590	Line 2654
2654
2655	I've changed andl to testl in the floating point test sequence	I've changed andl to testl in the floating point test sequence
2656	whenever appropriate.  The Intel optimization guide states that	whenever appropriate.  The Intel optimization guide states that
2657	testl is perferable to andl.	testl is preferable to andl.
2658
2659	7. RA (x86 only)	7. RA (x86 only)
2660
#	Line 772	Line 2836
2836
2837	1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an	1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
2838	additional argument called "template" which is an optional string that	additional argument called "template" which is an optional string that
2839	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
2840	explanation.	explanation.
2841
2842	2. A special-purpose tool can be "regisitered" by simply dropping the	2. A special-purpose tool can be "registered" by simply dropping the
2843	corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same	corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
2844	directory where the .cm file lives that uses this tool.  (The	directory where the .cm file lives that uses this tool.  (The
2845	behavior/misfeature until now was to look for the tool description	behavior/misfeature until now was to look for the tool description
#	Line 819	Line 2883
2883	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
2884	<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,
2885	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
2886	perhaps got accidentially killed).  In this case, fixpt will quickly	perhaps got accidentally killed).  In this case, fixpt will quickly
2887	move through what exists before continuing where it left off earlier,	move through what exists before continuing where it left off earlier,
2888	and, thus, saves a lot of time.	and, thus, saves a lot of time.
2889
#	Line 869	Line 2933
2933	it from that remote directory.	it from that remote directory.
2934	This should simplify installation further:  For machines that have	This should simplify installation further:  For machines that have
2935	access to the internet, just fetch <version>-config.tgz, unpack it,	access to the internet, just fetch <version>-config.tgz, unpack it,
2936	edit config/targets, and go (run config/install.sh).  The scipt will	edit config/targets, and go (run config/install.sh).  The script will
2937	fetch everything else that it might need all by itself.	fetch everything else that it might need all by itself.
2938
2939	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.633   changed lines   Added in v.873

---

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