Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 721, Mon Nov 6 14:03:39 2000 UTC	revision 916, Fri Aug 24 21:13:58 2001 UTC
#	Line 13	Line 13
13	Description:	Description:
14
15	----------------------------------------------------------------------	----------------------------------------------------------------------
16		Name: Matthias Blume
17		Date: 2001/08/24 17:10:00 EDT
18		Tag: Release_110_35
19		Description:
20
21		New version number (110.35).  New bootfiles.
22
23		----------------------------------------------------------------------
24		Name: Lal George
25		Date: 2001/08/24 13:47:18 EDT 2001
26		Tag: george-20010824-MLRISC-graphs
27		Description:
28
29		removed clusters from MLRISC completely and replaced with graphs.
30
31		----------------------------------------------------------------------
32		Name: Matthias Blume
33		Date: 2001/08/23 17:50:00 EDT
34		Tag: blume-20010823-toplevel
35		Description:
36
37		- some reorganization of the code that implements various kinds of
38		environments in the compiler (static, dynamic, symbolic, combined)
39		- re-implemented the EnvRef module so that evalStream works properly
40		(if the stream contains references to "use", "CM.make", etc.)
41		- cleaned up evalloop.sml and interact.sml (but they need more cleaning)
42
43		----------------------------------------------------------------------
44		Name: Matthias Blume
45		Date: 2001/08/20 15:50 EDT
46		Tag: blume20010820-slipup
47		Description:
48
49		I forgot to commit a few files.  Here they are...
50
51		----------------------------------------------------------------------
52		Name: Matthias Blume
53		Date: 2001/08/20 15:35:00 EDT
54		Tag: blume-20010820-debugprof
55		Description:
56
57		!!!! NEW BOOTFILES !!!!
58
59		This is another round of reorganizing the compiler sources.  This
60		time the main goal was to factor out all the "instrumentation"
61		passes (for profiling and backtracing) into their own library.
62		The difficulty was to do it in such a way that it does not depend
63		on elaborate.cm but only on elabdata.cm.
64
65		Therefore there have been further changes to both elaborate.cm and
66		elabdata.cm -- more "generic" things have been moved from the former
67		to the latter.  As a result, I was forced to split the assignment
68		of numbers indicating "primtyc"s into two portions: SML-generic and
69		SML/NJ-specific.  Since it would have been awkward to maintain,
70		I bit the bullet and actually _changed_ the mapping between these
71		numbers and primtycs.  The bottom line of this is that you need
72		a new set of bin- and bootfiles.
73
74		I have built new bootfiles for all architectures, so doing a fresh
75		checkout and config/install.sh should be all you need.
76
77		The newly created library's name is
78
79		$smlnj/viscomp/debugprof.cm
80
81		and its sources live under
82
83		src/compiler/DebugProf
84
85		----------------------------------------------------------------------
86		Name: Matthias Blume
87		Date: 2001/08/15 17:15:00 EDT
88		Tag: blume-20010815-compreorg
89		Description:
90
91		This is a first cut at reorganizing the CM libraries that make up the
92		core of the compiler.  The idea is to separate out pieces that could
93		be used independently by tools, e.g., the parser, the typechecker, etc.
94
95		The current status is a step in this direction, but it is not quite
96		satisfactory yet.  Expect more changes in the future.
97
98		Here is the current (new) organization...
99
100		What used to be $smlnj/viscomp/core.cm is now divided into
101		six CM libraries:
102
103		$smlnj/viscomp/basics.cm
104		/parser.cm
105		/elabdata.cm
106		/elaborate.cm
107		/execute.cm
108		/core.cm
109
110		The CM files for these libraries live under src/system/smlnj/viscomp.
111		All these libraries are proxy libraries that contain precisely
112		one CM library component.  Here are the locations of the components
113		(all within the src/compiler tree):
114
115		Basics/basics.cm
116		Parse/parser.cm
117		ElabData/elabdata.cm
118		Elaborator/elaborate.cm
119		Execution/execute.cm
120		core.cm
121
122		[This organization is the same that has been used already
123		for a while for the architecture-specific parts of the visible
124		compiler and for the old version of core.cm.]
125
126		As you will notice, many source files have been moved from their
127		respective original locations to a new home in one of the above
128		subtrees.
129
130		The division of labor between the new libraries is the following:
131
132		basics.cm:
133		- Simple, basic definitions that pertain to many (or all) of
134		the other libraries.
135		parser.cm:
136		- The SML parser, producing output of type Ast.dec.
137		- The type family for Ast is also defined and exported here.
138		elabdata.cm:
139		- The datatypes that describe input and output of the elaborator.
140		This includes types, absyn, and static environments.
141		elaborator.cm:
142		- The SML/NJ type checker and elaborator.
143		This maps an Ast.dec (with a given static environment) to
144		an Absyn.dec (with a new static environment).
145		- This libraries implements certain modules that used to be
146		structures as functors (to remove dependencies on FLINT).
147		execute.cm:
148		- Everything having to do with executing binary code objects.
149		- Dynamic environments.
150		core.cm:
151		- SML/NJ-specific instantiations of the elaborator and MLRISC.
152		- Top-level modules.
153		- FLINT (this should eventually become its own library)
154
155		Notes:
156
157		I am not 100% happy with the way I separated the elaborator (and its
158		data structures) from FLINT.  Two instances of the same problem:
159
160		1. Data structures contain certain fields that carry FLINT-specific
161		information.  I hacked around this using exn and the property list
162		module from smlnj-lib.  But the fact that there are middle-end
163		specific fields around at all is a bit annoying.
164
165		2. The elaborator calculates certain FLINT-related information.  I tried
166		to make this as abstract as I could using functorization, but, again,
167		the fact that the elaborator has to perform calculations on behalf
168		of the middle-end at all is not nice.
169
170		3. Having to used exn and property lists is unfortunate because it
171		weakens type checking.  The other alternative (parameterizing
172		nearly *everything*) is not appealing, though.
173
174		I removed the "rebinding =" warning hack because due to the new organization
175		it was awkward to maintain it.  As a result, the compiler now issues some of
176		these warnings when compiling init.cmi during bootstrap compilation. On
177		the plus side, you also get a warning when you do, for example:
178		val op = = Int32.+
179		which was not the case up to now.
180
181		I placed "assign" and "deref" into the _Core structure so that the
182		code that deals with the "lazy" keyword can find them there.  This
183		removes the need for having access to the primitive environment
184		during elaboration.
185
186		----------------------------------------------------------------------
187		Name: Matthias Blume
188		Date: 2001/08/13
189		Tag: blume-20010813-closures
190		Description:
191
192		This fix was sent to us by Zhong Shao.  It is supposed to improve the
193		performance of certain loops by avoiding needless closure allocation.
194
195		----------------------------------------------------------------------
196		Name: Lal George
197		Date: 2001/07/31 10:03:23 EDT 2001
198		Tag: george-20010731-x86-fmalloc
199		Description: Fixed bug in x86 calls
200
201		There was a bug where call instructions would mysteriously
202		vanish. The call instruction had to be one that returned
203		a floating point value.
204
205		----------------------------------------------------------------------
206		Name: Lal George
207		Date: 2001/07/19 16:36:29 EDT 2001
208		Tag: george-20010719-simple-cells
209		Description:
210
211		I have dramatically simplified the interface for CELLS in MLRISC.
212
213		In summary, the cells interface is broken up into three parts:
214
215		1. CellsBasis : CELLS_BASIS
216
217		CellsBasis is a top level structure and common for all
218		architectures.  it contains the definitions of basic datatypes
219		and utility  functions over these types.
220
221		2. functor Cells() : CELLS
222
223		Cells generates an interface for CELLS that incorporates the
224		specific resources on the target architecture, such as the
225		presence of special register classes, their number and size,
226		and various useful substructures.
227
228		3. <ARCH>CELLS
229
230		e.g. SparcCells: SPARCCELLS
231
232		<ARCH>CELLS usually contains additional bindings for special
233		registers  on the architecture, such as:
234
235		val r0 : cell           (* register zero *)
236		val y : cell            (* Y register *)
237		val psr : cell          (* processor status register *)
238		...
239
240		The structure returned by applying the Cells functor is opened
241		in this interface.
242
243		The main implication of all this is that the datatypes for cells is
244		split between CellsBasis and CELLS -- a fairly simple change for user
245		code.
246
247		In the old scheme the CELLS interface had a definitional binding of
248		the form:
249
250		signature CELLS = sig
251
252		structure CellsBasis = CellsBasis
253
254		...
255
256		end
257
258		With all the sharing constraints that goes on in MLRISC, this old
259		design  quickly leads to errors such as:
260
261		"structure definition spec inside of sharing ... "
262
263
264		and appears to require an unacceptable amount of sharing and where
265		constraint hackery.
266
267		I think this error message (the interaction of definitional specs and
268		sharing) requires more explanation on our web page.
269
270		----------------------------------------------------------------------
271		Name: Matthias Blume
272		Date: 2001/07/19 15:00:00 EDT
273		Tag: blume-20010719-libreorg
274		Description:
275
276		This update puts together a fairly extensive but straightforward change
277		to the way the libraries that implement the interactive system are
278		organized:
279
280		The biggest change is the elimination of structure Compiler.  As a
281		replacement for this structure, there is now a CM library
282		(known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
283		that exports all the substructures of the original structure Compiler
284		directly.  So instead of saying Compiler.Foo.bar one now simply
285		says Foo.bar.  (The CM libraries actually export a collection of
286		structures that is richer than the collection of substructures of
287		structure Compiler.)
288
289		To make the transition smooth, there is a separate library called
290		$smlnj/compiler/compiler.cm which puts together and exports the
291		original structure Compiler (or at least something very close to it).
292
293		There are five members of the original structure Compiler
294		that are not exported directly but which instead became members
295		of a new structure Backend (described by signature BACKEND).  These are:
296		structure Profile (: PROFILE), structure Compile (: COMPILE), structure
297		Interact (: INTERACT), structure Machine (: MACHINE), and val
298		architecture (: string).
299
300		Structure Compiler.Version has become structure CompilerVersion.
301
302		Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
303		by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
304		or x86, respectively.
305		Each of these exports the same frontend structures that
306		$smlnj/compiler.cm exports.  But they do not have a structure Backend
307		and instead export some structure <Arch>Backend where <Arch> is Alpha32,
308		Hppa, PPC, Sparc, or X86, respectively.
309
310		Library $smlnj/compiler/all.cm exports the union of the exports of
311		$smlnj/compiler/<arch>.cm
312
313		There are no structures <Arch>Compiler anymore, use
314		$smlnj/compiler/<arch>.cm instead.
315
316		Library host-compiler-0.cm is gone.  Instead, the internal library
317		that instantiates CM is now called cm0.cm.  Selection of the host
318		compiler (backend) is no longer done here but. (Responsibility for it
319		now lies with $smlnj/compiler/current.cm.  This seems to be more
320		logical.)
321
322		Many individual files have been moved or renamed.  Some files have
323		been split into multiple files, and some "dead" files have been deleted.
324
325		Aside from these changes to library organization, there are also changes
326		to the way the code itself is organized:
327
328		Structure Binfile has been re-implemented in such a way that it no
329		longer needs any knowledge of the compiler.  It exclusively deals
330		with the details of binfile layout.  It no longer invokes the
331		compiler (for the purpose of creating new prospective binfile
332		content), and it no longer has any knowledge of how to interpret
333		pickles.
334
335		Structure Compile (: COMPILE) has been stripped down to the bare
336		essentials of compilation.  It no longer deals with linking/execution.
337		The interface has been cleaned up considerably.
338
339		Utility routines for dealing with linking and execution have been
340		moved into their own substructures.
341
342		(The ultimate goal of these changes is to provide a light-weight
343		binfile loader/linker (at least for, e.g., stable libraries) that
344		does not require CM or the compiler to be present.)
345
346		CM documentation has been updated to reflect the changes to library
347		organization.
348
349		----------------------------------------------------------------------
350		Name: Matthias Blume
351		Date: 2001/07/10 17:30:00 EDT
352		Tag: Release_110_34
353		Description:
354
355		Minor tweak to 110.34 (re-tagged):
356
357		- README.html file added to CVS repository
358		- runtime compiles properly under FreeBSD 3.X and 4.X
359
360		----------------------------------------------------------------------
361		Name: Matthias Blume
362		Date: 2001/07/10 17:30:00 EDT
363		Tag: Release_110_34
364		Description:
365
366		New version number (110.34). New bootfiles.
367
368		----------------------------------------------------------------------
369		Name: Matthias Blume
370		Date: 2001/07/09 16:00:00 EDT
371		Tag: blume-20010709-more-varargs
372		Description:
373
374		I changed the handling of varargs in ml-nlffigen again:
375		The ellipsis ... will now simply be ignored (with an accompanying warning).
376
377		The immediate effect is that you can actually call a varargs function
378		from ML -- but you can't actually supply any arguments beyond the ones
379		specified explicitly.  (For example, you can call printf with its format
380		string, but you cannot pass additional arguments.)
381
382		This behavior is only marginally more useful than the one before, but
383		it has the advantage that a function or, more importantly, a function
384		type never gets dropped on the floor, thus avoiding follow-up problems with
385		other types that refer to the offending one.
386
387		----------------------------------------------------------------------
388		Name: Matthias Blume
389		Date: 2001/07/09 11:25:00 EDT
390		Tag: blume-20010709-varargs
391		Description:
392
393		1. ckit-lib.cm now exports structure Error
394		2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
395		with a warning accompanied by a source location.  Moreover, it
396		merely skips the offending function or type and proceeds with the
397		rest of its work.u  As a result, one can safely feed C code containing
398		"..." to ml-nlffigen.
399		3. There are some internal improvements to CM, providing slightly
400		more general string substitutions in the tools subsystem.
401
402		----------------------------------------------------------------------
403		Name: Matthias Blume
404		Date: 2001/06/27 15:10:00 EDT
405		Tag: blume-20010627-concur
406		Description:
407
408		Fixed a small bug in CM's handling of parallel compilation.
409		(You could observe the bug by Control-C-interrupting an ordinary
410		CMB.make or CM.stabilize and then attaching some compile servers.
411		The result was that all of a sudden the previously interrupted
412		compilation would continue on its own.  This was because of
413		an over-optimization: CM did not bother to clean out certain queues
414		when no servers were attached "anyway", resulting in the contents
415		of these queues to grab control when new servers did get attached.)
416
417		There is also another minor update to the CM manual.
418
419		----------------------------------------------------------------------
420		Name: Matthias Blume
421		Date: 2001/06/26 16:15:00 EDT
422		Tag: blume-20010626-cmdoc
423		Description:
424
425		Minor typo fixed in CM manual (syntax diagram for libraries).
426
427		----------------------------------------------------------------------
428		Name: Matthias Blume
429		Date: 2001/06/25 22:55:00 EDT
430		Tag: blume-20010625-x86pc
431		Description:
432
433		Fixed a nasty bug in the X86 assembly code that caused signal
434		handlers to fail (crash) randomly.
435
436		----------------------------------------------------------------------
437		Name: Matthias Blume
438		Date: 2001/06/25 12:05:00 EDT
439		Tag: blume-20010625-nlffigen
440		Description:
441
442		This update fixes a number of minor bugs in ml-nlffigen as reported by
443		Nick Carter <nbc@andrew.cmu.edu>.
444
445		1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
446		2. Default names for generated files are now derived from the name of
447		the C file *without its directory*.  In particular, this causes generated
448		files to be placed locally even if the C file is in some system directory.
449		3. Default names for generated signatures and structures are also derived
450		from the C file name without its directory.  This avoids silly things
451		like "structure GL/GL".
452		(Other silly names are still possible because ml-nlffigen does not do
453		a thorough check of whether generated names are legal ML identifiers.
454		When in doubt, use command line arguments to force particular names.)
455
456		----------------------------------------------------------------------
457		Name: Matthias Blume
458		Date: 2001/06/21 12:25:00 EDT
459		Tag: blume-20010621-eXene
460		Description:
461
462		eXene now compiles and (sort of) works again.
463
464		The library name (for version > 110.33) is $/eXene.cm.
465
466		I also added an new example in src/eXene/examples/nbody.  See the
467		README file there for details.
468
469		----------------------------------------------------------------------
470		Name: Matthias Blume
471		Date: 2001/06/20 16:40:00 EDT
472		Tag: blume-20010620-cml
473		Description:
474
475		CML now compiles and works again.
476
477		Libraries (for version > 110.33):
478
479		$cml/cml.cm            Main CML library.
480		$cml/basis.cm          CML's version of $/basis.cm.
481		$cml/cml-internal.cm   Internal helper library.
482		$cml/core-cml.cm       Internal helper library.
483		$cml-lib/trace-cml.cm  Tracing facility.
484		$cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
485
486		The installer (config/install.sh) has been taught how to properly
487		install this stuff.
488
489		----------------------------------------------------------------------
490		Name: Matthias Blume
491		Date: 2001/06/19 17:55:00 EDT
492		Tag: blume-20010619-instantiate
493		Description:
494
495		This un-breaks the fix for bug 1432.
496		(The bug was originally fixed in 110.9 but I broke it again some
497		time after that.)
498
499		----------------------------------------------------------------------
500		Name: Matthias Blume
501		Date: 2001/06/19 17:25:00 EDT
502		Tag: blume-20010619-signals
503		Description:
504
505		This should (hopefully) fix the long-standing signal handling bug.
506		(The runtime system was constructing a continuation record with an
507		incorrect descriptor which would cause the GC to drop data on the floor...)
508
509		----------------------------------------------------------------------
510		Name: Matthias Blume
511		Date: 2001/06/15 15:05:00 EDT
512		Tag: blume-20010615-moresparc
513		Description:
514
515		Here is a short late-hour update related to Sparc c-calls:
516
517		-- made handling of double-word arguments a bit smarter
518
519		-- instruction selection phase tries to collapse certain clumsily
520		constructed ML-Trees; typical example:
521
522		ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
523
524		This currently has no further impact on SML/NJ since mlriscGen does
525		not seem to generate such patterns in the first place, and c-calls
526		(which did generate them in the beginning) has meanwhile been fixed
527		so as to avoid them as well.
528
529		----------------------------------------------------------------------
530		Name: Matthias Blume
531		Date: 2001/06/15 15:05:00 EDT
532		Tag: blume-20010615-sparc
533		Description:
534
535		The purpose of this update is to provide an implementation of NLFFI
536		on Sparc machines.
537
538		Here are the changes in detail:
539
540		* src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
541		the Sparc implementation of the c-calls API.
542		* The Sparc backend of SML/NJ has been modified to uniformely use %fp
543		for accessing the ML frame.  Thus, we have a real frame pointer and
544		can freely modify %sp without need for an omit-frame-ptr phase.
545		The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
546		this case.
547		* ml-nlffigen has been taught to produce code for different architectures
548		and calling conventions.
549		* In a way similar to what was done in the x86 case, the Sparc
550		backend uses its own specific extension to mltree.  (For example,
551		it needs to be able to generate UNIMP instructions which are part
552		of the calling convention.)
553		* ml-nlffi-lib was reorganized to make it more modular (in particular,
554		to make it easier to plug in new machine- and os-dependent parts).
555
556		There are some other fairly unrelated bug fixes and cleanups as well:
557
558		* I further hacked the .cm files for MLRISC tools (like MDLGen) so
559		that they properly share their libraries with existing SML/NJ libraries.
560		* I fixed a minor cosmetic bug in CM, supressing certain spurious
561		follow-up error messages.
562		* Updates to CM/CMB documentation.
563
564		TODO items:
565
566		* MLRISC should use a different register as its asmTemp on the Sparc.
567		(The current %o2 is a really bad choice because it is part of the
568		calling conventions, so things might interfere in unexpected ways.)
569
570		----------------------------------------------------------------------
571		Name: Matthias Blume
572		Date: 2001/06/07
573		Tag: blume-20010607-calls
574		Description:
575
576		A number of internal changes related to C calls and calling conventions:
577
578		1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
579		number of bytes popped implicitly (by the callee).  In most cases
580		this field is 0 but on x86/win32 it is some non-zero value.  This
581		is information provided for the benefit of the "omit-frameptr" pass.
582		2. The CALL instruction on the x86 carries a similar "pops" field.
583		The instruction selection phase copies its value from the ML-Tree
584		CALL statement.
585		3. On all other architectures, the instruction selection phase checks
586		whether "pops=0" and complains if not.
587		4. The c-calls implementation for x86 now accepts two calling conventions:
588		"ccall" and "stdcall".  When "ccall" is selected, the caller cleans
589		up after the call and pops is set to 0.  For "stdcall", the caller
590		does nothing, leaving the cleanup to the callee; pops is set to
591		the number of bytes that were pushed onto the stack.
592		5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
593		distinguish between "ccall" and "stdcall".
594		6. The UNIMP instruction has been added to the supported Sparc instruction
595		set. (This is needed for implementing the official C calling convention
596		on this architecture.)
597		7. I fixed some of the .cm files under src/MLRISC/Tools to make them
598		work with the latest CM.
599
600		----------------------------------------------------------------------
601		Name: Matthias Blume
602		Date: 2001/06/05 15:10:00 EDT
603		Tag: blume-20010605-cm-index
604		Description:
605
606		0. The "lambdasplit" parameter for class "sml" in CM has been documented.
607
608		1. CM can now generate "index files".  These are human-readable files
609		that list on a per-.cm-file basis each toplevel symbol defined or
610		imported.  The location of the index file for
611		<p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
612		To enable index-file generation, set CM.Control.generate_index to true
613		or export an environment-symbol: export CM_GENERATE_INDEX=true.
614
615		The CM manual has been updated accordingly.
616
617		2. I made some slight modifications to the c-calls API in MLRISC.
618
619		a) There is now a callback to support saving/restoring of
620		dedicated but caller-save registers around the actual call
621		instruction.
622		b) One can optionally specify a comment-annotation for the
623		call instruction.
624
625		3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
626		(For example, the comment annotation shows the C prototype of
627		the function being called.)
628
629		----------------------------------------------------------------------
630		Name: Matthias Blume
631		Date: 2001/06/01 13:30:00 EDT
632		Tag: blume-20010601-nlffi-cleanup
633		Description:
634
635		This is mostly a cleanup of MLFFI stuff:
636
637		- some signature files have been put into a more exposed place
638		- the ugly 'f type parameter is gone (simplifies types tremendously!)
639		- ml-nlffigen changed accordingly
640		- tutorial updated
641
642		Other changes:
643
644		- author's affiliation in CM manual(s) updated
645		- some more recognized keywords added to Allen's sml.sty
646
647		----------------------------------------------------------------------
648		Name: Matthias Blume
649		Date: 2001/05/25 15:30:00 EDT
650		Tag: blume-20010525-iptr
651		Description:
652
653		- put the official 110.33-README (as it appears on the ftp server) under
654		CVS
655		- fixed a small bug related to incomplete pointer types in
656		ml-nlffigen
657		- small cosmetic change to the ml-nlffi-lib's "arr" type constructor
658		(it does not need the 'f type parameter)
659
660		----------------------------------------------------------------------
661		Name: Matthias Blume
662		Date: 2001/05/23 14:30:00 EDT
663		Tag: Release_110_33
664		Description:
665
666		New version number (110.33).  New bootfiles.
667
668		----------------------------------------------------------------------
669		Name: Matthias Blume
670		Date: 2001/05/22 18:06:00 EDT
671		Tag: blume-20010522-targets
672		Description:
673
674		Made install.sh use file config/targets.customized if it exists, falling
675		back to config/targets if it doesn't.  This way one can have a customized
676		version of the targets file without touching the "real thing", thus
677		eliminating the constant fear of accidentally checking something bogus
678		back into the CVS repository...  (File config/targets.customized must
679		not be added to the repository!)
680
681		----------------------------------------------------------------------
682		Name: Matthias Blume
683		Date: 2001/05/22 16:30:00 EDT
684		Tag: blume-20010522-minitut
685		Description:
686
687		1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
688		struct returns.
689		2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
690		incomplete, preliminary documentation for NLFFI.
691
692		----------------------------------------------------------------------
693		Name: Matthias Blume
694		Date: 2001/05/14 11:30:00 EDT
695		Tag: blume-20010514-script
696		Description:
697
698		Some bugs in install script fixed.
699
700		In addition to that I also made a slight change to the NLFFI API:
701		Functors generated by ml-nlffigen now take the dynamic library as a
702		straight functor argument, not as a suspended one.  (The original
703		functor code used to force the suspension right away anyway, so there
704		was nothing gained by this complication of the interface.)
705
706		----------------------------------------------------------------------
707		Name: Matthias Blume
708		Date: 2001/05/11 14:35:00 EDT
709		Tag: blume-20010511-ml-nlffi
710		Description:
711
712		I finally took the plunge and added my new FFI code to the main
713		repository.  For x86-linux it is now ready for prime-time.
714
715		There are two new subdirectories of "src":
716
717		- ml-nlffi-lib:
718		The utility library for programs using the FFI interface.
719		Here is the implementation of $/c.cm and its associated low-level
720		partners $/c-int.cm and $/memory.cm.
721		- ml-nlffigen:
722		A stand-alone program for generating ML glue code from C source
723		code.
724
725		Building ml-nlffigen requires $/ckit-lib.cm.
726
727		The config/install.sh script has been updates to do the Right Thing
728		(hopefully).
729
730		Notice that the source tree for the C-Kit will not be put under "src"
731		but directly under the installation root directory.  (This is the
732		structure that currently exists on the CVS server when you check out
733		module "sml".)  Fortunately, config/install.sh knows about this oddity.
734
735		Bugs: No documentation yet.
736
737		----------------------------------------------------------------------
738		Name: Matthias Blume
739		Date: 2001/05/09 16:35:00 EDT
740		Tag: blume-20010509-cpscontract
741		Description:
742
743		Fixed a bug in the accounting code in cpsopt/contract.sml.  (The
744		wrapper/unwrapper elimination did not decrement usage counts and some
745		dead variables got overlooked by the dead-up logic.)
746
747		----------------------------------------------------------------------
748		Name: Lal George
749		Date: 2001/05/08  17:26:09 EDT
750		Tag: george-20010508-omit-frameptr
751		Description:
752
753		Changes to implement the omit-frame-pointer optimization to support
754		raw C calls. For now, there is only support on the Intel x86, but
755		other architectures will follow as more experience is gained with this.
756
757
758		----------------------------------------------------------------------
759		Name: Matthias Blume
760		Date: 2001/05/07 14:40:00 EDT
761		Tag: blume-20010507-proxies
762		Description:
763
764		I made into "proxy libraries" all libraries that qualify for such a
765		change.  (A qualifying library is a library that has another library or
766		groups as its sole member and repeats that member's export list
767		verbatim.  A proxy library avoids this repetition by omitting its export
768		list, effectively inheriting the list that its (only) member exports.
769		See the CM manual for more explanation.)
770		The main effect is that explicit export lists for these libraries
771		do not have to be kepts in sync, making maintenance a bit easier.
772
773		I also added copyright notices to many .cm-files.
774
775		Last but not least, I made a new set of bootfiles.
776
777		----------------------------------------------------------------------
778		Name: Matthias Blume
779		Date: 2001/05/04 17:00:00 EDT
780		Tag: blume-20010504-cm-lsplit
781		Description:
782
783		0. John merged pending changes to $/smlnj-lib.cm
784
785		1. Allen's previous change accidentally backed out of one of Lal's
786		earlier changes.  I undid this mistake (re-introducing Lal's change).
787
788		2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
789		within the compiler where applicable.  There is some code simplification
790		because of that.
791
792		3. The "split" phase (in FLINT) is now part of the default list of phases.
793		Compiler.Control.LambdaSplitting.* can be used to globally control the
794		lambda-splitting (cross-module-inlining) engine.  In addition to that,
795		it can now also be controlled on a per-source basis: CM has been taught
796		a new tool parameter applicable to ML source files.
797
798		- To turn lambda-splitting off completely:
799		local open Compiler.Control.LambdaSplitting in
800		val _ = set Off
801		end
802		- To make "no lambda-splitting" the global default (but allow per-source
803		overriding); this is the initial setting:
804		local open Compiler.Control.LambdaSplitting in
805		val _ = set (Default NONE)
806		end
807		- To make "lambda-splitting with aggressiveness a" the global default
808		(and allow per-source overriding):
809		local open Compiler.Control.LambdaSplitting in
810		val _ = set (Default (SOME a))
811		end
812
813		- To turn lambda-splitting off for a given ML souce file (say: a.sml)
814		write (in the respective .cm-file):
815		a.sml (lambdasplitting:off)
816		- To turn lambda-splitting for a.sml on with minimal aggressiveness:
817		a.sml (lambdasplitting:on)
818		- To turn lambda-splitting for a.sml on with aggressiveness <a> (where
819		<a> is a decimal non-negative integer):
820		a.sml (lambdasplitting:<a>)
821		- To turn lambda-splitting for a.sml on with maximal aggressiveness:
822		a.sml (lambdasplitting:infinity)
823		- To use the global default for a.sml:
824		a.sml (lambdasplitting:default)
825		or simply
826		a.sml
827
828		----------------------------------------------------------------------
829		Name: Allen Leung
830		Date: 2001/05/04 01:57:00 EDT
831		Tag: leunga-20010504-sync
832		Description:
833
834		MLRISC features.
835
836		1. Fix to CMPXCHG instructions.
837		2. Changed RA interface to allow annotations in callbacks.
838		3. Added a new method to the stream interface to allow annotations updates.
839
840		----------------------------------------------------------------------
841		Name: Matthias Blume
842		Date: 2001/05/01 11:45:00 EDT
843		Tag: blume-20010501-pcedittmp
844		Description:
845
846		Changed install.sh to use the current working directory instead of
847		/usr/tmp for a temporary file (pcedittmp).  The previous choice
848		of /usr/tmp caused trouble with MacOS X because of file premission
849		problems.
850
851		----------------------------------------------------------------------
852		Name: Matthias Blume
853		Date: 2001/04/20 11:10:00 EDT
854		Tag: blume-20010420-inMLflag
855		Description:
856
857		- added vp_limitPtrMask to vproc-state.h
858		(for use by the raw-C-calls mechanism to implement proper interrupt
859		handling)
860		- made the ML compiler aware of various data-structure offsets so it
861		can generate code for accessing the vp_inML flag and vp_limitPtrMask
862		- tweaked mlriscGen.sml to have it emit interrupt-handling code for
863		raw C-calls
864
865		----------------------------------------------------------------------
866		Name: Lal George
867		Date: 2001/04/20 09:15:28 EDT
868		Tag: george-20010420-macosX
869		Description:
870
871		- Changes to port to Mac OS X; Darwin.
872
873		- In the process I found that sqrt was broken on the PPC, because the
874		fsqrt instruction is not implemented.
875
876		----------------------------------------------------------------------
877		Name: Matthias Blume
878		Date: 2001/04/18 12:45:00 EDT
879		Tag: blume-20010418-ccalls
880		Description:
881
882		- fixed two off-by-4 errors in the x86-specific c-calls implementation
883		(this bug prevented structure arguments containing pointers from being
884		passed correctly)
885		- changed the raw-C-call code in mlriscGen.sml in such a way that
886		structure arguments are represented as a pointer to the beginning
887		of the structure (instead of having a series of synthesized arguments,
888		one for each structure member)
889
890		- made makeml script's verbosity level configurable via environment
891		variable (MAKEML_VERBOSITY)
892
893		- eliminated placeholder implementations for f32l, w16s, i16s, and f32s
894		in rawmem-x86.sml; we are now using the real thing
895
896		----------------------------------------------------------------------
897		Name: Matthias Blume
898		Date: 2001/03/22 16:25:00 EST
899		Tag: blume-20010322-bootfiles
900		Description:
901
902		Created a new set of bootfiles (for your automatic installation convenience).
903
904		----------------------------------------------------------------------
905		Name: Matthias Blume
906		Date: 2001/03/22 15:10:00 EST
907		Tag: blume-20010322-rawmem-parcm
908		Description:
909
910		1. All "raw memory access" primitives for the new FFI are implemented now
911		(at least on the x86).
912		2. Some further cleanup of CM's parallel make mechanism.
913
914		----------------------------------------------------------------------
915		Name: Matthias Blume
916		Date: 2001/03/19 17:53:00 EST
917		Tag: blume-20010319-parallel
918		Description:
919
920		Parallel make (using compile servers) now works again.
921
922		To this end, CM.stabilize and CMB.make have been modified to work in
923		two passes when compile servers are attached:
924		1. Compile everything, do not perform stabilization; this pass
925		uses compile servers
926		2. Stabilize everything; this pass does not use compile servers
927		If there are no compile servers, the two passes are combined into one
928		(as before).  Splitting the passes increases the inherent parallelism
929		in the dependency graph because the entire graph including all
930		libraries is available at the same time.  This, in turn, improves
931		server utilization.  The downside is that the master process will
932		have to do some extra work after compilation is done (because for
933		technical reasons it must re-read all the binfiles during stabilization).
934
935		----------------------------------------------------------------------
936		Name: Matthias Blume
937		Date: 2001/03/16 12:22:00 EST
938		Tag: blume-20010316-bootfiles
939		Description:
940
941		Created a new set of bootfiles (for your automatic installation convenience).
942
943		----------------------------------------------------------------------
944		Name: Matthias Blume
945		Date: 2001/03/16 11:00:00 EST
946		Tag: blume-20010316-MLTREE-fixup
947		Description:
948
949		This is a minor fixup for an (untagged) earlier commit by Allen.
950		(A file was missing).
951
952		----------------------------------------------------------------------
953		Name: Allen Leung
954		Date: Mon Mar  5 18:54:57 EST 2001
955		Tag: leunga-20010305-cut-support
956
957		1. New support for alternative control-flow in MLTREE.
958		Currently we support
959
960		FLOW_TO(CALL ...., [k1,...,kn])
961
962		This is needed for 'cuts to' in C-- and try/handle-like constructs
963		in Moby
964
965		New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
966
967		2. Register Allocator
968
969		Changes in interface [from Fermin, John]
970
971		3. Alpha 8-bit SLL support [Fermin]
972
973		4. All architectures
974
975		A new module (ClusterExpandCopies) for expanding parallel copies.
976
977		----------------------------------------------------------------------
978		Name: Allen Leung
979		Date: 2001/02/27 23:07:00 EST
980		Tag: leunga-20010227-minor-stuff
981
982		1. Alpha bug fix for CMOVNE
983		2. Handle mltree COND(..,FCMP ...,...)
984		3. Bug fix in simplifier
985
986		----------------------------------------------------------------------
987		Name: Matthias Blume
988		Date: 2001/01/30 17:50:00 EST
989		Tag: blume-20010130-sync
990		Description:
991
992		This is just a minor update to sync my devel branch with the main brach.
993		The only visible change is the addition of some README files.
994
995		----------------------------------------------------------------------
996		Name: Matthias Blume
997		Date: 2001/01/12 23:30:00 JST
998		Tag: blume-20010112-bootfiles
999		Description:
1000
1001		Made a new set of bootfiles that goes with the current state of the
1002		repository.
1003
1004		----------------------------------------------------------------------
1005		Name: Matthias Blume
1006		Date: 2001/01/12 21:20:00 JST
1007		Tag: blume-20010112-sync
1008		Description:
1009
1010		I am just flushing out some minor changes that had accumulated in
1011		my private branch in order to sync with the main tree.  (This is
1012		mainly because I had CVS trouble when trying to merge _into_ my
1013		private branch.)
1014
1015		Most people should be completely unaffected by this.
1016
1017		----------------------------------------------------------------------
1018		Name: Allen Leung
1019		Date: Thu Jan 11 21:03:00 EST 2001
1020		Tag: leunga-20010111-labexp=mltree
1021		Description:
1022
1023		1.  Removed the type LabelExp and replace it by MLTree.
1024		2.  Rewritten mltree-simplify with the pattern matcher tool.
1025		3.  There were some bugs in alpha code generator which would break
1026		64-bit code generation.
1027		4.  Redo the tools to generate code with the
1028		5.  The CM files in MLRISC (and in src/system/smlnj/MLRISC)
1029		are now generated by perl scripts.
1030
1031		----------------------------------------------------------------------
1032		Name: Matthias Blume
1033		Date: 2001/01/10 21:55:00 JST
1034		Tag: blume-20010110-rcc
1035		Description:
1036
1037		The RCC stuff now seems to work (but only on the x86).
1038		This required hacking of the c-calls interface (and -implementation) in
1039		MLRISC.
1040
1041		Normal compiler users should be unaffected.
1042
1043		----------------------------------------------------------------------
1044		Name: Matthias Blume
1045		Date: 2001/01/09 01:20:00 JST
1046		Tag: blume-20010109-rcc
1047		Description:
1048
1049		This is a fairly big patch, flushing out a large number of pending
1050		changes that I made to my development copy over the last couple of days.
1051
1052		Of practical relevance at this moment is a workaround for a pickling
1053		bug that Allen ran into the other day.  The cause of the bug itself is
1054		still unknown and it might be hard to fix it properly, but the
1055		workaround has some merits of its own (namely somewhat reducing pickling
1056		overhead for certain libraries).  Therefore, I think this solution should
1057		be satisfactory at this time.
1058
1059		The rest of the changes (i.e., the vast majority) has to do with my
1060		ongoing efforts of providing direct support for C function calls from
1061		ML.  At the moment there is a new primop "RAW_CCALL", typing magic
1062		in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
1063		case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
1064		translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
1065		handle RCC.
1066
1067		The last part (the changes to mlriscGen.sml) are still known to be
1068		wrong on the x86 and not implemented on all other architectures.  But
1069		the infrastructure is in place. I had to change a few functor
1070		signatures in the backend to be able to route the CCalls interface
1071		from MLRISC there, and I had to specialize the mltree type (on the
1072		x86) to include the necessary extensions. (The extensions themselves
1073		were already there and redy to go in MLRISC/x86).
1074
1075		Everything should be very happy as soon as someone helps me with
1076		mlriscGen.sml...
1077
1078		In any case, nothing of this should matter to anyone as long as the
1079		new primop is not being used (which is going to be the case unless you
1080		find it where I hid it :). The rest of the compiler is completely
1081		unaffected.
1082
1083		----------------------------------------------------------------------
1084		Name: Matthias Blume
1085		Date: 2001/01/05 00:30:00 JST
1086		Tag: blume-20010105-primops
1087		Description:
1088
1089		Added some experimental support for work that I am doing right now.
1090		These changes mostly concern added primops, but there is also a new
1091		experimental C library in the runtime system (but currently not enabled
1092		anywhere except on Linux/X86).
1093
1094		In the course of adding primops (and playing with them), I discovered that
1095		Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
1096		broken.  (Zhong was very right he labeled this stuff as "major gross hack".)
1097		To recover, I made type information in INL_PRIM mandatory and changed
1098		prim.sml as well as built-in.sml accordingly.  The InLine structure now
1099		has complete, correct type information (i.e., no bottom types).
1100
1101		Since all these changes mean that we need new binfiles, I also bumped the
1102		version number to 110.32.1.
1103
1104		----------------------------------------------------------------------
1105		Name: Matthias Blume
1106		Date: 2000/12/30 22:10:00 JST
1107		Tag: blume-20001230-various
1108		Description:
1109
1110		Added proxy libraries for MLRISC and let MLRISC libraries refer
1111		to each other using path anchors.  (See CM manual for explanation.)
1112
1113		Updated CM documentation.
1114
1115		Fixed some bugs in CM.
1116
1117		Implemented "proxy" libraries (= syntactic sugar for CM).
1118
1119		Added "-quiet" option to makeml and changed runtime system accordingly.
1120
1121		Added cleanup handler for exportML to reset timers and compiler stats.
1122
1123		----------------------------------------------------------------------
1124		Name: Lal George
1125		Date: 2000/12/22 22:22:58 EST 2000
1126		Tag: Release_110_32
1127		Description:
1128
1129		Infinite precision used throughout MLRISC.
1130		see MLRISC/mltree/machine-int.sig
1131
1132		----------------------------------------------------------------------
1133		Name: Matthias Blume
1134		Date: 2000/12/22 23:16:00 JST
1135		Tag: blume-20001222-warn
1136		Description:
1137
1138		Corrected wording and formatting of some CM warning message which I
1139		broke in my previous patch.
1140
1141		----------------------------------------------------------------------
1142		Name: Matthias Blume
1143		Date: 2000/12/22 21:20:00 JST
1144		Tag: blume-20001222-anchorenv
1145		Description:
1146
1147		Fixed CM's handling of anchor environments in connection with CMB.make.
1148
1149		----------------------------------------------------------------------
1150		Name: Matthias Blume
1151		Date: 2000/12/22 13:15:00 JST
1152		Tag: blume-20001222-cleanup
1153		Description:
1154
1155		Removed src/cm/ffi which does not (and did not) belong here.
1156
1157		----------------------------------------------------------------------
1158		Name: Matthias Blume
1159		Date: 2000/12/21 23:55:00 JST
1160		Tag: blume-20001221-exn
1161		Description:
1162
1163		Probably most important: CM no longer silently swallows all exceptions
1164		in the compiler.
1165		Plus: some other minor CM changes.  For example, CM now reports some
1166		sizes for generated binfiles (code, data, envpickle, lambdapickle).
1167
1168		----------------------------------------------------------------------
1169		Name: Matthias Blume
1170		Date: 2000/12/15 00:01:05 JST
1171		Tag: blume-20001215-dirtool
1172		Description:
1173
1174		- "dir" tool added.
1175		- improvements and cleanup to Tools structure
1176		- documentation updates
1177
1178		----------------------------------------------------------------------
1179		Name: Allen Leung
1180		Date: Thu Dec 14 03:45:24 EST 2000
1181		Description:
1182		Tag:  leunga-20001214-int-inf
1183		Description:
1184
1185		In IntInf, added these standard functions, which are missing from our
1186		implementation:
1187
1188		andb : int * int -> int
1189		xorb : int * int -> int
1190		orb  : int * int -> int
1191		notb : int -> int
1192		<<   : int * word -> int
1193		~>>  : int * word -> int
1194
1195		Not tested, I hope they are correct.
1196
1197		----------------------------------------------------------------------
1198		Name: Allen Leung
1199		Date: Fri Dec  8 19:23:26 EST 2000
1200		Description:
1201		Tag:  leunga-20001208-nowhere
1202		Description:
1203
1204		Slight improvements to the 'nowhere' tool to handle OR-patterns,
1205		to generate better error messages etc.  Plus a brief manual.
1206
1207		----------------------------------------------------------------------
1208		Name: Lal George
1209		Date: 2000/12/08 09:54:02 EST 2000
1210		Tag: Release_110_31
1211		Description:
1212
1213		- Version 110.31
1214		----------------------------------------------------------------------
1215		Name: Allen Leung
1216		Date: Thu Dec  7 22:01:04 EST 2000
1217		Tag:  leunga-20001207-cell-monster-hack
1218		Description:
1219
1220		Major MLRISC internal changes.  Affect all clients.
1221		Summary:
1222
1223		1.  Type CELLS.cell = int is now replaced by a datatype.
1224		As a result, the old regmap is now gone.  Almost all interfaces
1225		in MLRISC change as a consequence.
1226
1227		2.  A new brand version of machine description tool (v3.0) that generates
1228		modules expecting the new interface.  The old version is removed.
1229
1230		3.  The RA interface has been further abstracted into two new functors.
1231		RISC_RA and X86RA.  These functors have much simpler interfaces.
1232		[See also directory MLRISC/demo.]
1233
1234		4.  Some other new source->source code generation tools are available:
1235
1236		a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
1237		b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
1238		I use this tool to generate the peephole optimizers---with the new
1239		cell type changes, peephole rules are becoming difficult to write
1240		without conditional pattern matching.
1241
1242		5.  More Intmap -> IntHashTable change.  Previous changes by Matthias didn't
1243		cover the entire MLRISC source tree so many things broke.
1244
1245		6.  CM files have been moved to the subdirectory MLRISC/cm.
1246		They are moved because there are a lot of them and they clutter up the
1247		root dir.
1248
1249		7.  More detailed documentation to come...
1250
1251		NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
1252		a makeml -rebuild first.  This is because of other other
1253		changes that Matthias has made (see below).
1254
1255
1256		----------------------------------------------------------------------
1257		Name: Matthias Blume
1258		Date: 2000/11/30 23:12:00 JST
1259		Tag: blume-20001130-filereorg
1260		Description:
1261
1262		Some manual updates and some file reorganizations in CM.
1263
1264		----------------------------------------------------------------------
1265		Name: Matthias Blume
1266		Date: 2000/11/24 17:45:00 JST
1267		Tag: blume-20001124-link
1268		Description:
1269
1270		Drastically improved link traversal code for the case that the dynamic
1271		value was already loaded at bootstrap time.  As a result, CM and CMB
1272		now both load blazingly fast -- even on a very slow machine.  Also,
1273		memory consumption has been further reduced by this.
1274
1275		Warning: The format of the PIDMAP file has changed.  THerefore, to
1276		bootstrap you have to do this:
1277
1278		1. Run CMB.make
1279		2. Make a symbolic link for the boot directory:
1280		ln -s sml.boot.ARCH-OS xxx
1281		3. "Rebuild" the boot directory:
1282		./makeml -boot xxx -rebuild sml ; rm xxx
1283		4. Boot normally:
1284		./makeml
1285
1286		----------------------------------------------------------------------
1287		Name: Matthias Blume
1288		Date: 2000/11/21 21:20:00 JST
1289		Tag: blume-20001121-tools
1290		Description:
1291
1292		Continued hacking on autoloading problem -- with success this time.
1293		Also changed tool-plugin mechanism.  See new CM manual.
1294
1295		----------------------------------------------------------------------
1296		Name: Matthias Blume
1297		Date: 2000/11/19 14:30:00 JST
1298		Tag:  blume-20001119-autoload
1299		Description:
1300
1301		Some hacking to make autoloading faster.  Success for CMB, no success
1302		so far for CM.  There is a reduced structure CM' that autoloads faster.
1303		(This is a temporary, non-documented hack to be eliminated again when
1304		the general problem is solved.)
1305
1306		----------------------------------------------------------------------
1307		Name: Matthias Blume
1308		Date: 2000/11/17 14:10:00 JST
1309		Tag: blume-20001117-pickle-lib
1310		Description:
1311
1312		1. Eliminated comp-lib.cm
1313		2. Made pickle-lib.cm
1314		3. Eliminated all uses of intset.sml (from comp-lib.cm)
1315		4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
1316		equivalent constructs from smlnj-lib.cm (INtHashTable).
1317		5. Point 4. also goes for those uses of intmap.* in MLRISC.
1318		Duplicated intmap modules thrown out.
1319		6. Hunted down all duplicated SCC code and replaced it with
1320		equivalent stuff (GraphSCCFn from smlnj-lib.cm).
1321		7. Rewrote Feedback module.
1322		8. Moved sortedlist.sml into viscomp-lib.cm.  Eventually it
1323		should be thrown out and equivalent modules from smlnj-lib.cm
1324		should be used (IntRedBlackSet, IntListSet, ...).
1325
1326		Confirmed that compiler compiles to fixpoint.
1327
1328		----------------------------------------------------------------------
1329		Name: Allen Leung
1330		Date: 2000/11/10 18:00:00
1331		Tag: leunga-20001110-new-x86-fp
1332
1333		A new x86 floating point code generator has been added.
1334		By default this is turned off.  To turn this on, do:
1335
1336		CM.autoload "$smlnj/compiler.cm";
1337		Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
1338
1339		Changes:
1340
1341		1.  Changed FTAN to FPTAN so that the assembly output is correct.
1342		2.  Changed the extension callback for FTANGENT to generate:
1343
1344		fptan
1345		fstp  %st(0)
1346		instead of
1347		fptan
1348		fstpl ftempmem
1349
1350		3.  Numerous assembly fixes for x86.
1351
1352		5.  Cleaned up the machine code output module x86/x86MC.sml and added
1353		support for a whole bunch of instructions and addressing modes:
1354
1355		fadd/fsub/fsubr/fmul/fdiv/fdivr  %st, %st(n)
1356		faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp  %st, %st(n)
1357		fadd/fsub/fsubr/fmul/fdiv/fdivr  %st(n), %st
1358		fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
1359		fxch %st(n)
1360		fld %st(n)
1361		fst %st(n)
1362		fst mem
1363		fstp %st(n)
1364		fucom %st(n)
1365		fucomp %st(n)
1366
1367		All these are now generated when the fast fp mode is turned on.
1368
1369		6.  Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
1370
1371		----------------------------------------------------------------------
1372		Name: Matthias Blume
1373		Date: 2000/11/09 11:20:00 JST
1374		Tag: blume-20001109-scc
1375		Description:
1376
1377		Eliminated some code duplication:
1378
1379		1. Added "where" clause to GraphSCCFn in SML/NJ Library.
1380		(Otherwise the functor is useless.)
1381		2. Used GraphSCCFn where SCCUtilFun was used previously.
1382		3. Got rid of SCCUtilFun (in comp-lib.cm).
1383
1384		----------------------------------------------------------------------
1385	Name: Lal George	Name: Lal George
1386	Date: 2000/11/06 09:02:21 EST 2000	Date: 2000/11/06 09:02:21 EST 2000
1387	Tag: Release_110_30	Tag: Release_110_30

---

Colored Diff Long Colored Diff Unidiff Context Diff Side by Side

Legend: Removed from v.721   changed lines   Added in v.916

---

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