Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 756, Thu Dec 14 16:01:33 2000 UTC	revision 890, Thu Jul 19 20:38:56 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: Lal George
17		Date: 2001/07/19 16:36:29 EDT 2001
18		Tag: george-20010719-simple-cells
19		Description:
20
21		I have dramatically simplified the interface for CELLS in MLRISC.
22
23		In summary, the cells interface is broken up into three parts:
24
25		1. CellsBasis : CELLS_BASIS
26
27		CellsBasis is a top level structure and common for all
28		architectures.  it contains the definitions of basic datatypes
29		and utility  functions over these types.
30
31		2. functor Cells() : CELLS
32
33		Cells generates an interface for CELLS that incorporates the
34		specific resources on the target architecture, such as the
35		presence of special register classes, their number and size,
36		and various useful substructures.
37
38		3. <ARCH>CELLS
39
40		e.g. SparcCells: SPARCCELLS
41
42		<ARCH>CELLS usually contains additional bindings for special
43		registers  on the architecture, such as:
44
45		val r0 : cell           (* register zero *)
46		val y : cell            (* Y register *)
47		val psr : cell          (* processor status register *)
48		...
49
50		The structure returned by applying the Cells functor is opened
51		in this interface.
52
53		The main implication of all this is that the datatypes for cells is
54		split between CellsBasis and CELLS -- a fairly simple change for user
55		code.
56
57		In the old scheme the CELLS interface had a definitional binding of
58		the form:
59
60		signature CELLS = sig
61
62		structure CellsBasis = CellsBasis
63
64		...
65
66		end
67
68		With all the sharing constraints that goes on in MLRISC, this old
69		design  quickly leads to errors such as:
70
71		"structure definition spec inside of sharing ... "
72
73
74		and appears to require an unacceptable amount of sharing and where
75		constraint hackery.
76
77		I think this error message (the interaction of definitional specs and
78		sharing) requires more explanation on our web page.
79
80		----------------------------------------------------------------------
81		Name: Matthias Blume
82		Date: 2001/07/19 15:00:00 EDT
83		Tag: blume-20010719-libreorg
84		Description:
85
86		This update puts together a fairly extensive but straightforward change
87		to the way the libraries that implement the interactive system are
88		organized:
89
90		The biggest change is the elimination of structure Compiler.  As a
91		replacement for this structure, there is now a CM library
92		(known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
93		that exports all the substructures of the original structure Compiler
94		directly.  So instead of saying Compiler.Foo.bar one now simply
95		says Foo.bar.  (The CM libraries actually export a collection of
96		structures that is richer than the collection of substructures of
97		structure Compiler.)
98
99		To make the transition smooth, there is a separate library called
100		$smlnj/compiler/compiler.cm which puts together and exports the
101		original structure Compiler (or at least something very close to it).
102
103		There are five members of the original structure Compiler
104		that are not exported directly but which instead became members
105		of a new structure Backend (described by signature BACKEND).  These are:
106		structure Profile (: PROFILE), structure Compile (: COMPILE), structure
107		Interact (: INTERACT), structure Machine (: MACHINE), and val
108		architecture (: string).
109
110		Structure Compiler.Version has become structure CompilerVersion.
111
112		Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
113		by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
114		or x86, respectively.
115		Each of these exports the same frontend structures that
116		$smlnj/compiler.cm exports.  But they do not have a structure Backend
117		and instead export some structure <Arch>Backend where <Arch> is Alpha32,
118		Hppa, PPC, Sparc, or X86, respectively.
119
120		Library $smlnj/compiler/all.cm exports the union of the exports of
121		$smlnj/compiler/<arch>.cm
122
123		There are no structures <Arch>Compiler anymore, use
124		$smlnj/compiler/<arch>.cm instead.
125
126		Library host-compiler-0.cm is gone.  Instead, the internal library
127		that instantiates CM is now called cm0.cm.  Selection of the host
128		compiler (backend) is no longer done here but. (Responsibility for it
129		now lies with $smlnj/compiler/current.cm.  This seems to be more
130		logical.)
131
132		Many individual files have been moved or renamed.  Some files have
133		been split into multiple files, and some "dead" files have been deleted.
134
135		Aside from these changes to library organization, there are also changes
136		to the way the code itself is organized:
137
138		Structure Binfile has been re-implemented in such a way that it no
139		longer needs any knowledge of the compiler.  It exclusively deals
140		with the details of binfile layout.  It no longer invokes the
141		compiler (for the purpose of creating new prospective binfile
142		content), and it no longer has any knowledge of how to interpret
143		pickles.
144
145		Structure Compile (: COMPILE) has been stripped down to the bare
146		essentials of compilation.  It no longer deals with linking/execution.
147		The interface has been cleaned up considerably.
148
149		Utility routines for dealing with linking and execution have been
150		moved into their own substructures.
151
152		(The ultimate goal of these changes is to provide a light-weight
153		binfile loader/linker (at least for, e.g., stable libraries) that
154		does not require CM or the compiler to be present.)
155
156		CM documentation has been updated to reflect the changes to library
157		organization.
158
159		----------------------------------------------------------------------
160		Name: Matthias Blume
161		Date: 2001/07/10 17:30:00 EDT
162		Tag: Release_110_34
163		Description:
164
165		Minor tweak to 110.34 (re-tagged):
166
167		- README.html file added to CVS repository
168		- runtime compiles properly under FreeBSD 3.X and 4.X
169
170		----------------------------------------------------------------------
171		Name: Matthias Blume
172		Date: 2001/07/10 17:30:00 EDT
173		Tag: Release_110_34
174		Description:
175
176		New version number (110.34). New bootfiles.
177
178		----------------------------------------------------------------------
179		Name: Matthias Blume
180		Date: 2001/07/09 16:00:00 EDT
181		Tag: blume-20010709-more-varargs
182		Description:
183
184		I changed the handling of varargs in ml-nlffigen again:
185		The ellipsis ... will now simply be ignored (with an accompanying warning).
186
187		The immediate effect is that you can actually call a varargs function
188		from ML -- but you can't actually supply any arguments beyond the ones
189		specified explicitly.  (For example, you can call printf with its format
190		string, but you cannot pass additional arguments.)
191
192		This behavior is only marginally more useful than the one before, but
193		it has the advantage that a function or, more importantly, a function
194		type never gets dropped on the floor, thus avoiding follow-up problems with
195		other types that refer to the offending one.
196
197		----------------------------------------------------------------------
198		Name: Matthias Blume
199		Date: 2001/07/09 11:25:00 EDT
200		Tag: blume-20010709-varargs
201		Description:
202
203		1. ckit-lib.cm now exports structure Error
204		2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
205		with a warning accompanied by a source location.  Moreover, it
206		merely skips the offending function or type and proceeds with the
207		rest of its work.u  As a result, one can safely feed C code containing
208		"..." to ml-nlffigen.
209		3. There are some internal improvements to CM, providing slightly
210		more general string substitutions in the tools subsystem.
211
212		----------------------------------------------------------------------
213		Name: Matthias Blume
214		Date: 2001/06/27 15:10:00 EDT
215		Tag: blume-20010627-concur
216		Description:
217
218		Fixed a small bug in CM's handling of parallel compilation.
219		(You could observe the bug by Control-C-interrupting an ordinary
220		CMB.make or CM.stabilize and then attaching some compile servers.
221		The result was that all of a sudden the previously interrupted
222		compilation would continue on its own.  This was because of
223		an over-optimization: CM did not bother to clean out certain queues
224		when no servers were attached "anyway", resulting in the contents
225		of these queues to grab control when new servers did get attached.)
226
227		There is also another minor update to the CM manual.
228
229		----------------------------------------------------------------------
230		Name: Matthias Blume
231		Date: 2001/06/26 16:15:00 EDT
232		Tag: blume-20010626-cmdoc
233		Description:
234
235		Minor typo fixed in CM manual (syntax diagram for libraries).
236
237		----------------------------------------------------------------------
238		Name: Matthias Blume
239		Date: 2001/06/25 22:55:00 EDT
240		Tag: blume-20010625-x86pc
241		Description:
242
243		Fixed a nasty bug in the X86 assembly code that caused signal
244		handlers to fail (crash) randomly.
245
246		----------------------------------------------------------------------
247		Name: Matthias Blume
248		Date: 2001/06/25 12:05:00 EDT
249		Tag: blume-20010625-nlffigen
250		Description:
251
252		This update fixes a number of minor bugs in ml-nlffigen as reported by
253		Nick Carter <nbc@andrew.cmu.edu>.
254
255		1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
256		2. Default names for generated files are now derived from the name of
257		the C file *without its directory*.  In particular, this causes generated
258		files to be placed locally even if the C file is in some system directory.
259		3. Default names for generated signatures and structures are also derived
260		from the C file name without its directory.  This avoids silly things
261		like "structure GL/GL".
262		(Other silly names are still possible because ml-nlffigen does not do
263		a thorough check of whether generated names are legal ML identifiers.
264		When in doubt, use command line arguments to force particular names.)
265
266		----------------------------------------------------------------------
267		Name: Matthias Blume
268		Date: 2001/06/21 12:25:00 EDT
269		Tag: blume-20010621-eXene
270		Description:
271
272		eXene now compiles and (sort of) works again.
273
274		The library name (for version > 110.33) is $/eXene.cm.
275
276		I also added an new example in src/eXene/examples/nbody.  See the
277		README file there for details.
278
279		----------------------------------------------------------------------
280		Name: Matthias Blume
281		Date: 2001/06/20 16:40:00 EDT
282		Tag: blume-20010620-cml
283		Description:
284
285		CML now compiles and works again.
286
287		Libraries (for version > 110.33):
288
289		$cml/cml.cm            Main CML library.
290		$cml/basis.cm          CML's version of $/basis.cm.
291		$cml/cml-internal.cm   Internal helper library.
292		$cml/core-cml.cm       Internal helper library.
293		$cml-lib/trace-cml.cm  Tracing facility.
294		$cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
295
296		The installer (config/install.sh) has been taught how to properly
297		install this stuff.
298
299		----------------------------------------------------------------------
300		Name: Matthias Blume
301		Date: 2001/06/19 17:55:00 EDT
302		Tag: blume-20010619-instantiate
303		Description:
304
305		This un-breaks the fix for bug 1432.
306		(The bug was originally fixed in 110.9 but I broke it again some
307		time after that.)
308
309		----------------------------------------------------------------------
310		Name: Matthias Blume
311		Date: 2001/06/19 17:25:00 EDT
312		Tag: blume-20010619-signals
313		Description:
314
315		This should (hopefully) fix the long-standing signal handling bug.
316		(The runtime system was constructing a continuation record with an
317		incorrect descriptor which would cause the GC to drop data on the floor...)
318
319		----------------------------------------------------------------------
320		Name: Matthias Blume
321		Date: 2001/06/15 15:05:00 EDT
322		Tag: blume-20010615-moresparc
323		Description:
324
325		Here is a short late-hour update related to Sparc c-calls:
326
327		-- made handling of double-word arguments a bit smarter
328
329		-- instruction selection phase tries to collapse certain clumsily
330		constructed ML-Trees; typical example:
331
332		ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
333
334		This currently has no further impact on SML/NJ since mlriscGen does
335		not seem to generate such patterns in the first place, and c-calls
336		(which did generate them in the beginning) has meanwhile been fixed
337		so as to avoid them as well.
338
339		----------------------------------------------------------------------
340		Name: Matthias Blume
341		Date: 2001/06/15 15:05:00 EDT
342		Tag: blume-20010615-sparc
343		Description:
344
345		The purpose of this update is to provide an implementation of NLFFI
346		on Sparc machines.
347
348		Here are the changes in detail:
349
350		* src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
351		the Sparc implementation of the c-calls API.
352		* The Sparc backend of SML/NJ has been modified to uniformely use %fp
353		for accessing the ML frame.  Thus, we have a real frame pointer and
354		can freely modify %sp without need for an omit-frame-ptr phase.
355		The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
356		this case.
357		* ml-nlffigen has been taught to produce code for different architectures
358		and calling conventions.
359		* In a way similar to what was done in the x86 case, the Sparc
360		backend uses its own specific extension to mltree.  (For example,
361		it needs to be able to generate UNIMP instructions which are part
362		of the calling convention.)
363		* ml-nlffi-lib was reorganized to make it more modular (in particular,
364		to make it easier to plug in new machine- and os-dependent parts).
365
366		There are some other fairly unrelated bug fixes and cleanups as well:
367
368		* I further hacked the .cm files for MLRISC tools (like MDLGen) so
369		that they properly share their libraries with existing SML/NJ libraries.
370		* I fixed a minor cosmetic bug in CM, supressing certain spurious
371		follow-up error messages.
372		* Updates to CM/CMB documentation.
373
374		TODO items:
375
376		* MLRISC should use a different register as its asmTemp on the Sparc.
377		(The current %o2 is a really bad choice because it is part of the
378		calling conventions, so things might interfere in unexpected ways.)
379
380		----------------------------------------------------------------------
381		Name: Matthias Blume
382		Date: 2001/06/07
383		Tag: blume-20010607-calls
384		Description:
385
386		A number of internal changes related to C calls and calling conventions:
387
388		1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
389		number of bytes popped implicitly (by the callee).  In most cases
390		this field is 0 but on x86/win32 it is some non-zero value.  This
391		is information provided for the benefit of the "omit-frameptr" pass.
392		2. The CALL instruction on the x86 carries a similar "pops" field.
393		The instruction selection phase copies its value from the ML-Tree
394		CALL statement.
395		3. On all other architectures, the instruction selection phase checks
396		whether "pops=0" and complains if not.
397		4. The c-calls implementation for x86 now accepts two calling conventions:
398		"ccall" and "stdcall".  When "ccall" is selected, the caller cleans
399		up after the call and pops is set to 0.  For "stdcall", the caller
400		does nothing, leaving the cleanup to the callee; pops is set to
401		the number of bytes that were pushed onto the stack.
402		5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
403		distinguish between "ccall" and "stdcall".
404		6. The UNIMP instruction has been added to the supported Sparc instruction
405		set. (This is needed for implementing the official C calling convention
406		on this architecture.)
407		7. I fixed some of the .cm files under src/MLRISC/Tools to make them
408		work with the latest CM.
409
410		----------------------------------------------------------------------
411		Name: Matthias Blume
412		Date: 2001/06/05 15:10:00 EDT
413		Tag: blume-20010605-cm-index
414		Description:
415
416		0. The "lambdasplit" parameter for class "sml" in CM has been documented.
417
418		1. CM can now generate "index files".  These are human-readable files
419		that list on a per-.cm-file basis each toplevel symbol defined or
420		imported.  The location of the index file for
421		<p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
422		To enable index-file generation, set CM.Control.generate_index to true
423		or export an environment-symbol: export CM_GENERATE_INDEX=true.
424
425		The CM manual has been updated accordingly.
426
427		2. I made some slight modifications to the c-calls API in MLRISC.
428
429		a) There is now a callback to support saving/restoring of
430		dedicated but caller-save registers around the actual call
431		instruction.
432		b) One can optionally specify a comment-annotation for the
433		call instruction.
434
435		3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
436		(For example, the comment annotation shows the C prototype of
437		the function being called.)
438
439		----------------------------------------------------------------------
440		Name: Matthias Blume
441		Date: 2001/06/01 13:30:00 EDT
442		Tag: blume-20010601-nlffi-cleanup
443		Description:
444
445		This is mostly a cleanup of MLFFI stuff:
446
447		- some signature files have been put into a more exposed place
448		- the ugly 'f type parameter is gone (simplifies types tremendously!)
449		- ml-nlffigen changed accordingly
450		- tutorial updated
451
452		Other changes:
453
454		- author's affiliation in CM manual(s) updated
455		- some more recognized keywords added to Allen's sml.sty
456
457		----------------------------------------------------------------------
458		Name: Matthias Blume
459		Date: 2001/05/25 15:30:00 EDT
460		Tag: blume-20010525-iptr
461		Description:
462
463		- put the official 110.33-README (as it appears on the ftp server) under
464		CVS
465		- fixed a small bug related to incomplete pointer types in
466		ml-nlffigen
467		- small cosmetic change to the ml-nlffi-lib's "arr" type constructor
468		(it does not need the 'f type parameter)
469
470		----------------------------------------------------------------------
471		Name: Matthias Blume
472		Date: 2001/05/23 14:30:00 EDT
473		Tag: Release_110_33
474		Description:
475
476		New version number (110.33).  New bootfiles.
477
478		----------------------------------------------------------------------
479		Name: Matthias Blume
480		Date: 2001/05/22 18:06:00 EDT
481		Tag: blume-20010522-targets
482		Description:
483
484		Made install.sh use file config/targets.customized if it exists, falling
485		back to config/targets if it doesn't.  This way one can have a customized
486		version of the targets file without touching the "real thing", thus
487		eliminating the constant fear of accidentally checking something bogus
488		back into the CVS repository...  (File config/targets.customized must
489		not be added to the repository!)
490
491		----------------------------------------------------------------------
492		Name: Matthias Blume
493		Date: 2001/05/22 16:30:00 EDT
494		Tag: blume-20010522-minitut
495		Description:
496
497		1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
498		struct returns.
499		2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
500		incomplete, preliminary documentation for NLFFI.
501
502		----------------------------------------------------------------------
503		Name: Matthias Blume
504		Date: 2001/05/14 11:30:00 EDT
505		Tag: blume-20010514-script
506		Description:
507
508		Some bugs in install script fixed.
509
510		In addition to that I also made a slight change to the NLFFI API:
511		Functors generated by ml-nlffigen now take the dynamic library as a
512		straight functor argument, not as a suspended one.  (The original
513		functor code used to force the suspension right away anyway, so there
514		was nothing gained by this complication of the interface.)
515
516		----------------------------------------------------------------------
517		Name: Matthias Blume
518		Date: 2001/05/11 14:35:00 EDT
519		Tag: blume-20010511-ml-nlffi
520		Description:
521
522		I finally took the plunge and added my new FFI code to the main
523		repository.  For x86-linux it is now ready for prime-time.
524
525		There are two new subdirectories of "src":
526
527		- ml-nlffi-lib:
528		The utility library for programs using the FFI interface.
529		Here is the implementation of $/c.cm and its associated low-level
530		partners $/c-int.cm and $/memory.cm.
531		- ml-nlffigen:
532		A stand-alone program for generating ML glue code from C source
533		code.
534
535		Building ml-nlffigen requires $/ckit-lib.cm.
536
537		The config/install.sh script has been updates to do the Right Thing
538		(hopefully).
539
540		Notice that the source tree for the C-Kit will not be put under "src"
541		but directly under the installation root directory.  (This is the
542		structure that currently exists on the CVS server when you check out
543		module "sml".)  Fortunately, config/install.sh knows about this oddity.
544
545		Bugs: No documentation yet.
546
547		----------------------------------------------------------------------
548		Name: Matthias Blume
549		Date: 2001/05/09 16:35:00 EDT
550		Tag: blume-20010509-cpscontract
551		Description:
552
553		Fixed a bug in the accounting code in cpsopt/contract.sml.  (The
554		wrapper/unwrapper elimination did not decrement usage counts and some
555		dead variables got overlooked by the dead-up logic.)
556
557		----------------------------------------------------------------------
558		Name: Lal George
559		Date: 2001/05/08  17:26:09 EDT
560		Tag: george-20010508-omit-frameptr
561		Description:
562
563		Changes to implement the omit-frame-pointer optimization to support
564		raw C calls. For now, there is only support on the Intel x86, but
565		other architectures will follow as more experience is gained with this.
566
567
568		----------------------------------------------------------------------
569		Name: Matthias Blume
570		Date: 2001/05/07 14:40:00 EDT
571		Tag: blume-20010507-proxies
572		Description:
573
574		I made into "proxy libraries" all libraries that qualify for such a
575		change.  (A qualifying library is a library that has another library or
576		groups as its sole member and repeats that member's export list
577		verbatim.  A proxy library avoids this repetition by omitting its export
578		list, effectively inheriting the list that its (only) member exports.
579		See the CM manual for more explanation.)
580		The main effect is that explicit export lists for these libraries
581		do not have to be kepts in sync, making maintenance a bit easier.
582
583		I also added copyright notices to many .cm-files.
584
585		Last but not least, I made a new set of bootfiles.
586
587		----------------------------------------------------------------------
588		Name: Matthias Blume
589		Date: 2001/05/04 17:00:00 EDT
590		Tag: blume-20010504-cm-lsplit
591		Description:
592
593		0. John merged pending changes to $/smlnj-lib.cm
594
595		1. Allen's previous change accidentally backed out of one of Lal's
596		earlier changes.  I undid this mistake (re-introducing Lal's change).
597
598		2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
599		within the compiler where applicable.  There is some code simplification
600		because of that.
601
602		3. The "split" phase (in FLINT) is now part of the default list of phases.
603		Compiler.Control.LambdaSplitting.* can be used to globally control the
604		lambda-splitting (cross-module-inlining) engine.  In addition to that,
605		it can now also be controlled on a per-source basis: CM has been taught
606		a new tool parameter applicable to ML source files.
607
608		- To turn lambda-splitting off completely:
609		local open Compiler.Control.LambdaSplitting in
610		val _ = set Off
611		end
612		- To make "no lambda-splitting" the global default (but allow per-source
613		overriding); this is the initial setting:
614		local open Compiler.Control.LambdaSplitting in
615		val _ = set (Default NONE)
616		end
617		- To make "lambda-splitting with aggressiveness a" the global default
618		(and allow per-source overriding):
619		local open Compiler.Control.LambdaSplitting in
620		val _ = set (Default (SOME a))
621		end
622
623		- To turn lambda-splitting off for a given ML souce file (say: a.sml)
624		write (in the respective .cm-file):
625		a.sml (lambdasplitting:off)
626		- To turn lambda-splitting for a.sml on with minimal aggressiveness:
627		a.sml (lambdasplitting:on)
628		- To turn lambda-splitting for a.sml on with aggressiveness <a> (where
629		<a> is a decimal non-negative integer):
630		a.sml (lambdasplitting:<a>)
631		- To turn lambda-splitting for a.sml on with maximal aggressiveness:
632		a.sml (lambdasplitting:infinity)
633		- To use the global default for a.sml:
634		a.sml (lambdasplitting:default)
635		or simply
636		a.sml
637
638		----------------------------------------------------------------------
639		Name: Allen Leung
640		Date: 2001/05/04 01:57:00 EDT
641		Tag: leunga-20010504-sync
642		Description:
643
644		MLRISC features.
645
646		1. Fix to CMPXCHG instructions.
647		2. Changed RA interface to allow annotations in callbacks.
648		3. Added a new method to the stream interface to allow annotations updates.
649
650		----------------------------------------------------------------------
651		Name: Matthias Blume
652		Date: 2001/05/01 11:45:00 EDT
653		Tag: blume-20010501-pcedittmp
654		Description:
655
656		Changed install.sh to use the current working directory instead of
657		/usr/tmp for a temporary file (pcedittmp).  The previous choice
658		of /usr/tmp caused trouble with MacOS X because of file premission
659		problems.
660
661		----------------------------------------------------------------------
662		Name: Matthias Blume
663		Date: 2001/04/20 11:10:00 EDT
664		Tag: blume-20010420-inMLflag
665		Description:
666
667		- added vp_limitPtrMask to vproc-state.h
668		(for use by the raw-C-calls mechanism to implement proper interrupt
669		handling)
670		- made the ML compiler aware of various data-structure offsets so it
671		can generate code for accessing the vp_inML flag and vp_limitPtrMask
672		- tweaked mlriscGen.sml to have it emit interrupt-handling code for
673		raw C-calls
674
675		----------------------------------------------------------------------
676		Name: Lal George
677		Date: 2001/04/20 09:15:28 EDT
678		Tag: george-20010420-macosX
679		Description:
680
681		- Changes to port to Mac OS X; Darwin.
682
683		- In the process I found that sqrt was broken on the PPC, because the
684		fsqrt instruction is not implemented.
685
686		----------------------------------------------------------------------
687		Name: Matthias Blume
688		Date: 2001/04/18 12:45:00 EDT
689		Tag: blume-20010418-ccalls
690		Description:
691
692		- fixed two off-by-4 errors in the x86-specific c-calls implementation
693		(this bug prevented structure arguments containing pointers from being
694		passed correctly)
695		- changed the raw-C-call code in mlriscGen.sml in such a way that
696		structure arguments are represented as a pointer to the beginning
697		of the structure (instead of having a series of synthesized arguments,
698		one for each structure member)
699
700		- made makeml script's verbosity level configurable via environment
701		variable (MAKEML_VERBOSITY)
702
703		- eliminated placeholder implementations for f32l, w16s, i16s, and f32s
704		in rawmem-x86.sml; we are now using the real thing
705
706		----------------------------------------------------------------------
707		Name: Matthias Blume
708		Date: 2001/03/22 16:25:00 EST
709		Tag: blume-20010322-bootfiles
710		Description:
711
712		Created a new set of bootfiles (for your automatic installation convenience).
713
714		----------------------------------------------------------------------
715		Name: Matthias Blume
716		Date: 2001/03/22 15:10:00 EST
717		Tag: blume-20010322-rawmem-parcm
718		Description:
719
720		1. All "raw memory access" primitives for the new FFI are implemented now
721		(at least on the x86).
722		2. Some further cleanup of CM's parallel make mechanism.
723
724		----------------------------------------------------------------------
725		Name: Matthias Blume
726		Date: 2001/03/19 17:53:00 EST
727		Tag: blume-20010319-parallel
728		Description:
729
730		Parallel make (using compile servers) now works again.
731
732		To this end, CM.stabilize and CMB.make have been modified to work in
733		two passes when compile servers are attached:
734		1. Compile everything, do not perform stabilization; this pass
735		uses compile servers
736		2. Stabilize everything; this pass does not use compile servers
737		If there are no compile servers, the two passes are combined into one
738		(as before).  Splitting the passes increases the inherent parallelism
739		in the dependency graph because the entire graph including all
740		libraries is available at the same time.  This, in turn, improves
741		server utilization.  The downside is that the master process will
742		have to do some extra work after compilation is done (because for
743		technical reasons it must re-read all the binfiles during stabilization).
744
745		----------------------------------------------------------------------
746		Name: Matthias Blume
747		Date: 2001/03/16 12:22:00 EST
748		Tag: blume-20010316-bootfiles
749		Description:
750
751		Created a new set of bootfiles (for your automatic installation convenience).
752
753		----------------------------------------------------------------------
754		Name: Matthias Blume
755		Date: 2001/03/16 11:00:00 EST
756		Tag: blume-20010316-MLTREE-fixup
757		Description:
758
759		This is a minor fixup for an (untagged) earlier commit by Allen.
760		(A file was missing).
761
762		----------------------------------------------------------------------
763		Name: Allen Leung
764		Date: Mon Mar  5 18:54:57 EST 2001
765		Tag: leunga-20010305-cut-support
766
767		1. New support for alternative control-flow in MLTREE.
768		Currently we support
769
770		FLOW_TO(CALL ...., [k1,...,kn])
771
772		This is needed for 'cuts to' in C-- and try/handle-like constructs
773		in Moby
774
775		New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
776
777		2. Register Allocator
778
779		Changes in interface [from Fermin, John]
780
781		3. Alpha 8-bit SLL support [Fermin]
782
783		4. All architectures
784
785		A new module (ClusterExpandCopies) for expanding parallel copies.
786
787		----------------------------------------------------------------------
788		Name: Allen Leung
789		Date: 2001/02/27 23:07:00 EST
790		Tag: leunga-20010227-minor-stuff
791
792		1. Alpha bug fix for CMOVNE
793		2. Handle mltree COND(..,FCMP ...,...)
794		3. Bug fix in simplifier
795
796		----------------------------------------------------------------------
797		Name: Matthias Blume
798		Date: 2001/01/30 17:50:00 EST
799		Tag: blume-20010130-sync
800		Description:
801
802		This is just a minor update to sync my devel branch with the main brach.
803		The only visible change is the addition of some README files.
804
805		----------------------------------------------------------------------
806		Name: Matthias Blume
807		Date: 2001/01/12 23:30:00 JST
808		Tag: blume-20010112-bootfiles
809		Description:
810
811		Made a new set of bootfiles that goes with the current state of the
812		repository.
813
814		----------------------------------------------------------------------
815		Name: Matthias Blume
816		Date: 2001/01/12 21:20:00 JST
817		Tag: blume-20010112-sync
818		Description:
819
820		I am just flushing out some minor changes that had accumulated in
821		my private branch in order to sync with the main tree.  (This is
822		mainly because I had CVS trouble when trying to merge _into_ my
823		private branch.)
824
825		Most people should be completely unaffected by this.
826
827		----------------------------------------------------------------------
828		Name: Allen Leung
829		Date: Thu Jan 11 21:03:00 EST 2001
830		Tag: leunga-20010111-labexp=mltree
831		Description:
832
833		1.  Removed the type LabelExp and replace it by MLTree.
834		2.  Rewritten mltree-simplify with the pattern matcher tool.
835		3.  There were some bugs in alpha code generator which would break
836		64-bit code generation.
837		4.  Redo the tools to generate code with the
838		5.  The CM files in MLRISC (and in src/system/smlnj/MLRISC)
839		are now generated by perl scripts.
840
841		----------------------------------------------------------------------
842		Name: Matthias Blume
843		Date: 2001/01/10 21:55:00 JST
844		Tag: blume-20010110-rcc
845		Description:
846
847		The RCC stuff now seems to work (but only on the x86).
848		This required hacking of the c-calls interface (and -implementation) in
849		MLRISC.
850
851		Normal compiler users should be unaffected.
852
853		----------------------------------------------------------------------
854		Name: Matthias Blume
855		Date: 2001/01/09 01:20:00 JST
856		Tag: blume-20010109-rcc
857		Description:
858
859		This is a fairly big patch, flushing out a large number of pending
860		changes that I made to my development copy over the last couple of days.
861
862		Of practical relevance at this moment is a workaround for a pickling
863		bug that Allen ran into the other day.  The cause of the bug itself is
864		still unknown and it might be hard to fix it properly, but the
865		workaround has some merits of its own (namely somewhat reducing pickling
866		overhead for certain libraries).  Therefore, I think this solution should
867		be satisfactory at this time.
868
869		The rest of the changes (i.e., the vast majority) has to do with my
870		ongoing efforts of providing direct support for C function calls from
871		ML.  At the moment there is a new primop "RAW_CCALL", typing magic
872		in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
873		case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
874		translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
875		handle RCC.
876
877		The last part (the changes to mlriscGen.sml) are still known to be
878		wrong on the x86 and not implemented on all other architectures.  But
879		the infrastructure is in place. I had to change a few functor
880		signatures in the backend to be able to route the CCalls interface
881		from MLRISC there, and I had to specialize the mltree type (on the
882		x86) to include the necessary extensions. (The extensions themselves
883		were already there and redy to go in MLRISC/x86).
884
885		Everything should be very happy as soon as someone helps me with
886		mlriscGen.sml...
887
888		In any case, nothing of this should matter to anyone as long as the
889		new primop is not being used (which is going to be the case unless you
890		find it where I hid it :). The rest of the compiler is completely
891		unaffected.
892
893		----------------------------------------------------------------------
894		Name: Matthias Blume
895		Date: 2001/01/05 00:30:00 JST
896		Tag: blume-20010105-primops
897		Description:
898
899		Added some experimental support for work that I am doing right now.
900		These changes mostly concern added primops, but there is also a new
901		experimental C library in the runtime system (but currently not enabled
902		anywhere except on Linux/X86).
903
904		In the course of adding primops (and playing with them), I discovered that
905		Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
906		broken.  (Zhong was very right he labeled this stuff as "major gross hack".)
907		To recover, I made type information in INL_PRIM mandatory and changed
908		prim.sml as well as built-in.sml accordingly.  The InLine structure now
909		has complete, correct type information (i.e., no bottom types).
910
911		Since all these changes mean that we need new binfiles, I also bumped the
912		version number to 110.32.1.
913
914		----------------------------------------------------------------------
915		Name: Matthias Blume
916		Date: 2000/12/30 22:10:00 JST
917		Tag: blume-20001230-various
918		Description:
919
920		Added proxy libraries for MLRISC and let MLRISC libraries refer
921		to each other using path anchors.  (See CM manual for explanation.)
922
923		Updated CM documentation.
924
925		Fixed some bugs in CM.
926
927		Implemented "proxy" libraries (= syntactic sugar for CM).
928
929		Added "-quiet" option to makeml and changed runtime system accordingly.
930
931		Added cleanup handler for exportML to reset timers and compiler stats.
932
933		----------------------------------------------------------------------
934		Name: Lal George
935		Date: 2000/12/22 22:22:58 EST 2000
936		Tag: Release_110_32
937		Description:
938
939		Infinite precision used throughout MLRISC.
940		see MLRISC/mltree/machine-int.sig
941
942		----------------------------------------------------------------------
943		Name: Matthias Blume
944		Date: 2000/12/22 23:16:00 JST
945		Tag: blume-20001222-warn
946		Description:
947
948		Corrected wording and formatting of some CM warning message which I
949		broke in my previous patch.
950
951		----------------------------------------------------------------------
952		Name: Matthias Blume
953		Date: 2000/12/22 21:20:00 JST
954		Tag: blume-20001222-anchorenv
955		Description:
956
957		Fixed CM's handling of anchor environments in connection with CMB.make.
958
959		----------------------------------------------------------------------
960		Name: Matthias Blume
961		Date: 2000/12/22 13:15:00 JST
962		Tag: blume-20001222-cleanup
963		Description:
964
965		Removed src/cm/ffi which does not (and did not) belong here.
966
967		----------------------------------------------------------------------
968		Name: Matthias Blume
969		Date: 2000/12/21 23:55:00 JST
970		Tag: blume-20001221-exn
971		Description:
972
973		Probably most important: CM no longer silently swallows all exceptions
974		in the compiler.
975		Plus: some other minor CM changes.  For example, CM now reports some
976		sizes for generated binfiles (code, data, envpickle, lambdapickle).
977
978	----------------------------------------------------------------------	----------------------------------------------------------------------
979	Name: Matthias Blume	Name: Matthias Blume
980	Date: 2000/12/15 00:01:05 JST	Date: 2000/12/15 00:01:05 JST

---

Colored Diff Long Colored Diff Unidiff Context Diff Side by Side

Legend: Removed from v.756   changed lines   Added in v.890

---

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