Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 800, Fri Mar 16 17:22:47 2001 UTC	revision 902, Wed Aug 15 21:17:05 2001 UTC
#	Line 14	Line 14
14
15	----------------------------------------------------------------------	----------------------------------------------------------------------
16	Name: Matthias Blume	Name: Matthias Blume
17		Date: 2001/08/15 17:15:00 EDT
18		Tag: blume-20010815-compreorg
19		Description:
20
21		This is a first cut at reorganizing the CM libraries that make up the
22		core of the compiler.  The idea is to separate out pieces that could
23		be used independently by tools, e.g., the parser, the typechecker, etc.
24
25		The current status is a step in this direction, but it is not quite
26		satisfactory yet.  Expect more changes in the future.
27
28		Here is the current (new) organization...
29
30		What used to be $smlnj/viscomp/core.cm is now divided into
31		six CM libraries:
32
33		$smlnj/viscomp/basics.cm
34		/parser.cm
35		/elabdata.cm
36		/elaborate.cm
37		/execute.cm
38		/core.cm
39
40		The CM files for these libraries live under src/system/smlnj/viscomp.
41		All these libraries are proxy libraries that contain precisely
42		one CM library component.  Here are the locations of the components
43		(all within the src/compiler tree):
44
45		Basics/basics.cm
46		Parse/parser.cm
47		ElabData/elabdata.cm
48		Elaborator/elaborate.cm
49		Execution/execute.cm
50		core.cm
51
52		[This organization is the same that has been used already
53		for a while for the architecture-specific parts of the visible
54		compiler and for the old version of core.cm.]
55
56		As you will notice, many source files have been moved from their
57		respective original locations to a new home in one of the above
58		subtrees.
59
60		The division of labor between the new libraries is the following:
61
62		basics.cm:
63		- Simple, basic definitions that pertain to many (or all) of
64		the other libraries.
65		parser.cm:
66		- The SML parser, producing output of type Ast.dec.
67		- The type family for Ast is also defined and exported here.
68		elabdata.cm:
69		- The datatypes that describe input and output of the elaborator.
70		This includes types, absyn, and static environments.
71		elaborator.cm:
72		- The SML/NJ type checker and elaborator.
73		This maps an Ast.dec (with a given static environment) to
74		an Absyn.dec (with a new static environment).
75		- This libraries implements certain modules that used to be
76		structures as functors (to remove dependencies on FLINT).
77		execute.cm:
78		- Everything having to do with executing binary code objects.
79		- Dynamic environments.
80		core.cm:
81		- SML/NJ-specific instantiations of the elaborator and MLRISC.
82		- Top-level modules.
83		- FLINT (this should eventually become its own library)
84
85		Notes:
86
87		I am not 100% happy with the way I separated the elaborator (and its
88		data structures) from FLINT.  Two instances of the same problem:
89
90		1. Data structures contain certain fields that carry FLINT-specific
91		information.  I hacked around this using exn and the property list
92		module from smlnj-lib.  But the fact that there are middle-end
93		specific fields around at all is a bit annoying.
94
95		2. The elaborator calculates certain FLINT-related information.  I tried
96		to make this as abstract as I could using functorization, but, again,
97		the fact that the elaborator has to perform calculations on behalf
98		of the middle-end at all is not nice.
99
100		3. Having to used exn and property lists is unfortunate because it
101		weakens type checking.  The other alternative (parameterizing
102		nearly *everything*) is not appealing, though.
103
104		I removed the "rebinding =" warning hack because due to the new organization
105		it was awkward to maintain it.  As a result, the compiler now issues some of
106		these warnings when compiling init.cmi during bootstrap compilation. On
107		the plus side, you also get a warning when you do, for example:
108		val op = = Int32.+
109		which was not the case up to now.
110
111		I placed "assign" and "deref" into the _Core structure so that the
112		code that deals with the "lazy" keyword can find them there.  This
113		removes the need for having access to the primitive environment
114		during elaboration.
115
116		----------------------------------------------------------------------
117		Name: Matthias Blume
118		Date: 2001/08/13
119		Tag: blume-20010813-closures
120		Description:
121
122		This fix was sent to us by Zhong Shao.  It is supposed to improve the
123		performance of certain loops by avoiding needless closure allocation.
124
125		----------------------------------------------------------------------
126		Name: Lal George
127		Date: 2001/07/31 10:03:23 EDT 2001
128		Tag: george-20010731-x86-fmalloc
129		Description: Fixed bug in x86 calls
130
131		There was a bug where call instructions would mysteriously
132		vanish. The call instruction had to be one that returned
133		a floating point value.
134
135		----------------------------------------------------------------------
136		Name: Lal George
137		Date: 2001/07/19 16:36:29 EDT 2001
138		Tag: george-20010719-simple-cells
139		Description:
140
141		I have dramatically simplified the interface for CELLS in MLRISC.
142
143		In summary, the cells interface is broken up into three parts:
144
145		1. CellsBasis : CELLS_BASIS
146
147		CellsBasis is a top level structure and common for all
148		architectures.  it contains the definitions of basic datatypes
149		and utility  functions over these types.
150
151		2. functor Cells() : CELLS
152
153		Cells generates an interface for CELLS that incorporates the
154		specific resources on the target architecture, such as the
155		presence of special register classes, their number and size,
156		and various useful substructures.
157
158		3. <ARCH>CELLS
159
160		e.g. SparcCells: SPARCCELLS
161
162		<ARCH>CELLS usually contains additional bindings for special
163		registers  on the architecture, such as:
164
165		val r0 : cell           (* register zero *)
166		val y : cell            (* Y register *)
167		val psr : cell          (* processor status register *)
168		...
169
170		The structure returned by applying the Cells functor is opened
171		in this interface.
172
173		The main implication of all this is that the datatypes for cells is
174		split between CellsBasis and CELLS -- a fairly simple change for user
175		code.
176
177		In the old scheme the CELLS interface had a definitional binding of
178		the form:
179
180		signature CELLS = sig
181
182		structure CellsBasis = CellsBasis
183
184		...
185
186		end
187
188		With all the sharing constraints that goes on in MLRISC, this old
189		design  quickly leads to errors such as:
190
191		"structure definition spec inside of sharing ... "
192
193
194		and appears to require an unacceptable amount of sharing and where
195		constraint hackery.
196
197		I think this error message (the interaction of definitional specs and
198		sharing) requires more explanation on our web page.
199
200		----------------------------------------------------------------------
201		Name: Matthias Blume
202		Date: 2001/07/19 15:00:00 EDT
203		Tag: blume-20010719-libreorg
204		Description:
205
206		This update puts together a fairly extensive but straightforward change
207		to the way the libraries that implement the interactive system are
208		organized:
209
210		The biggest change is the elimination of structure Compiler.  As a
211		replacement for this structure, there is now a CM library
212		(known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
213		that exports all the substructures of the original structure Compiler
214		directly.  So instead of saying Compiler.Foo.bar one now simply
215		says Foo.bar.  (The CM libraries actually export a collection of
216		structures that is richer than the collection of substructures of
217		structure Compiler.)
218
219		To make the transition smooth, there is a separate library called
220		$smlnj/compiler/compiler.cm which puts together and exports the
221		original structure Compiler (or at least something very close to it).
222
223		There are five members of the original structure Compiler
224		that are not exported directly but which instead became members
225		of a new structure Backend (described by signature BACKEND).  These are:
226		structure Profile (: PROFILE), structure Compile (: COMPILE), structure
227		Interact (: INTERACT), structure Machine (: MACHINE), and val
228		architecture (: string).
229
230		Structure Compiler.Version has become structure CompilerVersion.
231
232		Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
233		by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
234		or x86, respectively.
235		Each of these exports the same frontend structures that
236		$smlnj/compiler.cm exports.  But they do not have a structure Backend
237		and instead export some structure <Arch>Backend where <Arch> is Alpha32,
238		Hppa, PPC, Sparc, or X86, respectively.
239
240		Library $smlnj/compiler/all.cm exports the union of the exports of
241		$smlnj/compiler/<arch>.cm
242
243		There are no structures <Arch>Compiler anymore, use
244		$smlnj/compiler/<arch>.cm instead.
245
246		Library host-compiler-0.cm is gone.  Instead, the internal library
247		that instantiates CM is now called cm0.cm.  Selection of the host
248		compiler (backend) is no longer done here but. (Responsibility for it
249		now lies with $smlnj/compiler/current.cm.  This seems to be more
250		logical.)
251
252		Many individual files have been moved or renamed.  Some files have
253		been split into multiple files, and some "dead" files have been deleted.
254
255		Aside from these changes to library organization, there are also changes
256		to the way the code itself is organized:
257
258		Structure Binfile has been re-implemented in such a way that it no
259		longer needs any knowledge of the compiler.  It exclusively deals
260		with the details of binfile layout.  It no longer invokes the
261		compiler (for the purpose of creating new prospective binfile
262		content), and it no longer has any knowledge of how to interpret
263		pickles.
264
265		Structure Compile (: COMPILE) has been stripped down to the bare
266		essentials of compilation.  It no longer deals with linking/execution.
267		The interface has been cleaned up considerably.
268
269		Utility routines for dealing with linking and execution have been
270		moved into their own substructures.
271
272		(The ultimate goal of these changes is to provide a light-weight
273		binfile loader/linker (at least for, e.g., stable libraries) that
274		does not require CM or the compiler to be present.)
275
276		CM documentation has been updated to reflect the changes to library
277		organization.
278
279		----------------------------------------------------------------------
280		Name: Matthias Blume
281		Date: 2001/07/10 17:30:00 EDT
282		Tag: Release_110_34
283		Description:
284
285		Minor tweak to 110.34 (re-tagged):
286
287		- README.html file added to CVS repository
288		- runtime compiles properly under FreeBSD 3.X and 4.X
289
290		----------------------------------------------------------------------
291		Name: Matthias Blume
292		Date: 2001/07/10 17:30:00 EDT
293		Tag: Release_110_34
294		Description:
295
296		New version number (110.34). New bootfiles.
297
298		----------------------------------------------------------------------
299		Name: Matthias Blume
300		Date: 2001/07/09 16:00:00 EDT
301		Tag: blume-20010709-more-varargs
302		Description:
303
304		I changed the handling of varargs in ml-nlffigen again:
305		The ellipsis ... will now simply be ignored (with an accompanying warning).
306
307		The immediate effect is that you can actually call a varargs function
308		from ML -- but you can't actually supply any arguments beyond the ones
309		specified explicitly.  (For example, you can call printf with its format
310		string, but you cannot pass additional arguments.)
311
312		This behavior is only marginally more useful than the one before, but
313		it has the advantage that a function or, more importantly, a function
314		type never gets dropped on the floor, thus avoiding follow-up problems with
315		other types that refer to the offending one.
316
317		----------------------------------------------------------------------
318		Name: Matthias Blume
319		Date: 2001/07/09 11:25:00 EDT
320		Tag: blume-20010709-varargs
321		Description:
322
323		1. ckit-lib.cm now exports structure Error
324		2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
325		with a warning accompanied by a source location.  Moreover, it
326		merely skips the offending function or type and proceeds with the
327		rest of its work.u  As a result, one can safely feed C code containing
328		"..." to ml-nlffigen.
329		3. There are some internal improvements to CM, providing slightly
330		more general string substitutions in the tools subsystem.
331
332		----------------------------------------------------------------------
333		Name: Matthias Blume
334		Date: 2001/06/27 15:10:00 EDT
335		Tag: blume-20010627-concur
336		Description:
337
338		Fixed a small bug in CM's handling of parallel compilation.
339		(You could observe the bug by Control-C-interrupting an ordinary
340		CMB.make or CM.stabilize and then attaching some compile servers.
341		The result was that all of a sudden the previously interrupted
342		compilation would continue on its own.  This was because of
343		an over-optimization: CM did not bother to clean out certain queues
344		when no servers were attached "anyway", resulting in the contents
345		of these queues to grab control when new servers did get attached.)
346
347		There is also another minor update to the CM manual.
348
349		----------------------------------------------------------------------
350		Name: Matthias Blume
351		Date: 2001/06/26 16:15:00 EDT
352		Tag: blume-20010626-cmdoc
353		Description:
354
355		Minor typo fixed in CM manual (syntax diagram for libraries).
356
357		----------------------------------------------------------------------
358		Name: Matthias Blume
359		Date: 2001/06/25 22:55:00 EDT
360		Tag: blume-20010625-x86pc
361		Description:
362
363		Fixed a nasty bug in the X86 assembly code that caused signal
364		handlers to fail (crash) randomly.
365
366		----------------------------------------------------------------------
367		Name: Matthias Blume
368		Date: 2001/06/25 12:05:00 EDT
369		Tag: blume-20010625-nlffigen
370		Description:
371
372		This update fixes a number of minor bugs in ml-nlffigen as reported by
373		Nick Carter <nbc@andrew.cmu.edu>.
374
375		1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
376		2. Default names for generated files are now derived from the name of
377		the C file *without its directory*.  In particular, this causes generated
378		files to be placed locally even if the C file is in some system directory.
379		3. Default names for generated signatures and structures are also derived
380		from the C file name without its directory.  This avoids silly things
381		like "structure GL/GL".
382		(Other silly names are still possible because ml-nlffigen does not do
383		a thorough check of whether generated names are legal ML identifiers.
384		When in doubt, use command line arguments to force particular names.)
385
386		----------------------------------------------------------------------
387		Name: Matthias Blume
388		Date: 2001/06/21 12:25:00 EDT
389		Tag: blume-20010621-eXene
390		Description:
391
392		eXene now compiles and (sort of) works again.
393
394		The library name (for version > 110.33) is $/eXene.cm.
395
396		I also added an new example in src/eXene/examples/nbody.  See the
397		README file there for details.
398
399		----------------------------------------------------------------------
400		Name: Matthias Blume
401		Date: 2001/06/20 16:40:00 EDT
402		Tag: blume-20010620-cml
403		Description:
404
405		CML now compiles and works again.
406
407		Libraries (for version > 110.33):
408
409		$cml/cml.cm            Main CML library.
410		$cml/basis.cm          CML's version of $/basis.cm.
411		$cml/cml-internal.cm   Internal helper library.
412		$cml/core-cml.cm       Internal helper library.
413		$cml-lib/trace-cml.cm  Tracing facility.
414		$cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
415
416		The installer (config/install.sh) has been taught how to properly
417		install this stuff.
418
419		----------------------------------------------------------------------
420		Name: Matthias Blume
421		Date: 2001/06/19 17:55:00 EDT
422		Tag: blume-20010619-instantiate
423		Description:
424
425		This un-breaks the fix for bug 1432.
426		(The bug was originally fixed in 110.9 but I broke it again some
427		time after that.)
428
429		----------------------------------------------------------------------
430		Name: Matthias Blume
431		Date: 2001/06/19 17:25:00 EDT
432		Tag: blume-20010619-signals
433		Description:
434
435		This should (hopefully) fix the long-standing signal handling bug.
436		(The runtime system was constructing a continuation record with an
437		incorrect descriptor which would cause the GC to drop data on the floor...)
438
439		----------------------------------------------------------------------
440		Name: Matthias Blume
441		Date: 2001/06/15 15:05:00 EDT
442		Tag: blume-20010615-moresparc
443		Description:
444
445		Here is a short late-hour update related to Sparc c-calls:
446
447		-- made handling of double-word arguments a bit smarter
448
449		-- instruction selection phase tries to collapse certain clumsily
450		constructed ML-Trees; typical example:
451
452		ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
453
454		This currently has no further impact on SML/NJ since mlriscGen does
455		not seem to generate such patterns in the first place, and c-calls
456		(which did generate them in the beginning) has meanwhile been fixed
457		so as to avoid them as well.
458
459		----------------------------------------------------------------------
460		Name: Matthias Blume
461		Date: 2001/06/15 15:05:00 EDT
462		Tag: blume-20010615-sparc
463		Description:
464
465		The purpose of this update is to provide an implementation of NLFFI
466		on Sparc machines.
467
468		Here are the changes in detail:
469
470		* src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
471		the Sparc implementation of the c-calls API.
472		* The Sparc backend of SML/NJ has been modified to uniformely use %fp
473		for accessing the ML frame.  Thus, we have a real frame pointer and
474		can freely modify %sp without need for an omit-frame-ptr phase.
475		The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
476		this case.
477		* ml-nlffigen has been taught to produce code for different architectures
478		and calling conventions.
479		* In a way similar to what was done in the x86 case, the Sparc
480		backend uses its own specific extension to mltree.  (For example,
481		it needs to be able to generate UNIMP instructions which are part
482		of the calling convention.)
483		* ml-nlffi-lib was reorganized to make it more modular (in particular,
484		to make it easier to plug in new machine- and os-dependent parts).
485
486		There are some other fairly unrelated bug fixes and cleanups as well:
487
488		* I further hacked the .cm files for MLRISC tools (like MDLGen) so
489		that they properly share their libraries with existing SML/NJ libraries.
490		* I fixed a minor cosmetic bug in CM, supressing certain spurious
491		follow-up error messages.
492		* Updates to CM/CMB documentation.
493
494		TODO items:
495
496		* MLRISC should use a different register as its asmTemp on the Sparc.
497		(The current %o2 is a really bad choice because it is part of the
498		calling conventions, so things might interfere in unexpected ways.)
499
500		----------------------------------------------------------------------
501		Name: Matthias Blume
502		Date: 2001/06/07
503		Tag: blume-20010607-calls
504		Description:
505
506		A number of internal changes related to C calls and calling conventions:
507
508		1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
509		number of bytes popped implicitly (by the callee).  In most cases
510		this field is 0 but on x86/win32 it is some non-zero value.  This
511		is information provided for the benefit of the "omit-frameptr" pass.
512		2. The CALL instruction on the x86 carries a similar "pops" field.
513		The instruction selection phase copies its value from the ML-Tree
514		CALL statement.
515		3. On all other architectures, the instruction selection phase checks
516		whether "pops=0" and complains if not.
517		4. The c-calls implementation for x86 now accepts two calling conventions:
518		"ccall" and "stdcall".  When "ccall" is selected, the caller cleans
519		up after the call and pops is set to 0.  For "stdcall", the caller
520		does nothing, leaving the cleanup to the callee; pops is set to
521		the number of bytes that were pushed onto the stack.
522		5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
523		distinguish between "ccall" and "stdcall".
524		6. The UNIMP instruction has been added to the supported Sparc instruction
525		set. (This is needed for implementing the official C calling convention
526		on this architecture.)
527		7. I fixed some of the .cm files under src/MLRISC/Tools to make them
528		work with the latest CM.
529
530		----------------------------------------------------------------------
531		Name: Matthias Blume
532		Date: 2001/06/05 15:10:00 EDT
533		Tag: blume-20010605-cm-index
534		Description:
535
536		0. The "lambdasplit" parameter for class "sml" in CM has been documented.
537
538		1. CM can now generate "index files".  These are human-readable files
539		that list on a per-.cm-file basis each toplevel symbol defined or
540		imported.  The location of the index file for
541		<p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
542		To enable index-file generation, set CM.Control.generate_index to true
543		or export an environment-symbol: export CM_GENERATE_INDEX=true.
544
545		The CM manual has been updated accordingly.
546
547		2. I made some slight modifications to the c-calls API in MLRISC.
548
549		a) There is now a callback to support saving/restoring of
550		dedicated but caller-save registers around the actual call
551		instruction.
552		b) One can optionally specify a comment-annotation for the
553		call instruction.
554
555		3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
556		(For example, the comment annotation shows the C prototype of
557		the function being called.)
558
559		----------------------------------------------------------------------
560		Name: Matthias Blume
561		Date: 2001/06/01 13:30:00 EDT
562		Tag: blume-20010601-nlffi-cleanup
563		Description:
564
565		This is mostly a cleanup of MLFFI stuff:
566
567		- some signature files have been put into a more exposed place
568		- the ugly 'f type parameter is gone (simplifies types tremendously!)
569		- ml-nlffigen changed accordingly
570		- tutorial updated
571
572		Other changes:
573
574		- author's affiliation in CM manual(s) updated
575		- some more recognized keywords added to Allen's sml.sty
576
577		----------------------------------------------------------------------
578		Name: Matthias Blume
579		Date: 2001/05/25 15:30:00 EDT
580		Tag: blume-20010525-iptr
581		Description:
582
583		- put the official 110.33-README (as it appears on the ftp server) under
584		CVS
585		- fixed a small bug related to incomplete pointer types in
586		ml-nlffigen
587		- small cosmetic change to the ml-nlffi-lib's "arr" type constructor
588		(it does not need the 'f type parameter)
589
590		----------------------------------------------------------------------
591		Name: Matthias Blume
592		Date: 2001/05/23 14:30:00 EDT
593		Tag: Release_110_33
594		Description:
595
596		New version number (110.33).  New bootfiles.
597
598		----------------------------------------------------------------------
599		Name: Matthias Blume
600		Date: 2001/05/22 18:06:00 EDT
601		Tag: blume-20010522-targets
602		Description:
603
604		Made install.sh use file config/targets.customized if it exists, falling
605		back to config/targets if it doesn't.  This way one can have a customized
606		version of the targets file without touching the "real thing", thus
607		eliminating the constant fear of accidentally checking something bogus
608		back into the CVS repository...  (File config/targets.customized must
609		not be added to the repository!)
610
611		----------------------------------------------------------------------
612		Name: Matthias Blume
613		Date: 2001/05/22 16:30:00 EDT
614		Tag: blume-20010522-minitut
615		Description:
616
617		1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
618		struct returns.
619		2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
620		incomplete, preliminary documentation for NLFFI.
621
622		----------------------------------------------------------------------
623		Name: Matthias Blume
624		Date: 2001/05/14 11:30:00 EDT
625		Tag: blume-20010514-script
626		Description:
627
628		Some bugs in install script fixed.
629
630		In addition to that I also made a slight change to the NLFFI API:
631		Functors generated by ml-nlffigen now take the dynamic library as a
632		straight functor argument, not as a suspended one.  (The original
633		functor code used to force the suspension right away anyway, so there
634		was nothing gained by this complication of the interface.)
635
636		----------------------------------------------------------------------
637		Name: Matthias Blume
638		Date: 2001/05/11 14:35:00 EDT
639		Tag: blume-20010511-ml-nlffi
640		Description:
641
642		I finally took the plunge and added my new FFI code to the main
643		repository.  For x86-linux it is now ready for prime-time.
644
645		There are two new subdirectories of "src":
646
647		- ml-nlffi-lib:
648		The utility library for programs using the FFI interface.
649		Here is the implementation of $/c.cm and its associated low-level
650		partners $/c-int.cm and $/memory.cm.
651		- ml-nlffigen:
652		A stand-alone program for generating ML glue code from C source
653		code.
654
655		Building ml-nlffigen requires $/ckit-lib.cm.
656
657		The config/install.sh script has been updates to do the Right Thing
658		(hopefully).
659
660		Notice that the source tree for the C-Kit will not be put under "src"
661		but directly under the installation root directory.  (This is the
662		structure that currently exists on the CVS server when you check out
663		module "sml".)  Fortunately, config/install.sh knows about this oddity.
664
665		Bugs: No documentation yet.
666
667		----------------------------------------------------------------------
668		Name: Matthias Blume
669		Date: 2001/05/09 16:35:00 EDT
670		Tag: blume-20010509-cpscontract
671		Description:
672
673		Fixed a bug in the accounting code in cpsopt/contract.sml.  (The
674		wrapper/unwrapper elimination did not decrement usage counts and some
675		dead variables got overlooked by the dead-up logic.)
676
677		----------------------------------------------------------------------
678		Name: Lal George
679		Date: 2001/05/08  17:26:09 EDT
680		Tag: george-20010508-omit-frameptr
681		Description:
682
683		Changes to implement the omit-frame-pointer optimization to support
684		raw C calls. For now, there is only support on the Intel x86, but
685		other architectures will follow as more experience is gained with this.
686
687
688		----------------------------------------------------------------------
689		Name: Matthias Blume
690		Date: 2001/05/07 14:40:00 EDT
691		Tag: blume-20010507-proxies
692		Description:
693
694		I made into "proxy libraries" all libraries that qualify for such a
695		change.  (A qualifying library is a library that has another library or
696		groups as its sole member and repeats that member's export list
697		verbatim.  A proxy library avoids this repetition by omitting its export
698		list, effectively inheriting the list that its (only) member exports.
699		See the CM manual for more explanation.)
700		The main effect is that explicit export lists for these libraries
701		do not have to be kepts in sync, making maintenance a bit easier.
702
703		I also added copyright notices to many .cm-files.
704
705		Last but not least, I made a new set of bootfiles.
706
707		----------------------------------------------------------------------
708		Name: Matthias Blume
709		Date: 2001/05/04 17:00:00 EDT
710		Tag: blume-20010504-cm-lsplit
711		Description:
712
713		0. John merged pending changes to $/smlnj-lib.cm
714
715		1. Allen's previous change accidentally backed out of one of Lal's
716		earlier changes.  I undid this mistake (re-introducing Lal's change).
717
718		2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
719		within the compiler where applicable.  There is some code simplification
720		because of that.
721
722		3. The "split" phase (in FLINT) is now part of the default list of phases.
723		Compiler.Control.LambdaSplitting.* can be used to globally control the
724		lambda-splitting (cross-module-inlining) engine.  In addition to that,
725		it can now also be controlled on a per-source basis: CM has been taught
726		a new tool parameter applicable to ML source files.
727
728		- To turn lambda-splitting off completely:
729		local open Compiler.Control.LambdaSplitting in
730		val _ = set Off
731		end
732		- To make "no lambda-splitting" the global default (but allow per-source
733		overriding); this is the initial setting:
734		local open Compiler.Control.LambdaSplitting in
735		val _ = set (Default NONE)
736		end
737		- To make "lambda-splitting with aggressiveness a" the global default
738		(and allow per-source overriding):
739		local open Compiler.Control.LambdaSplitting in
740		val _ = set (Default (SOME a))
741		end
742
743		- To turn lambda-splitting off for a given ML souce file (say: a.sml)
744		write (in the respective .cm-file):
745		a.sml (lambdasplitting:off)
746		- To turn lambda-splitting for a.sml on with minimal aggressiveness:
747		a.sml (lambdasplitting:on)
748		- To turn lambda-splitting for a.sml on with aggressiveness <a> (where
749		<a> is a decimal non-negative integer):
750		a.sml (lambdasplitting:<a>)
751		- To turn lambda-splitting for a.sml on with maximal aggressiveness:
752		a.sml (lambdasplitting:infinity)
753		- To use the global default for a.sml:
754		a.sml (lambdasplitting:default)
755		or simply
756		a.sml
757
758		----------------------------------------------------------------------
759		Name: Allen Leung
760		Date: 2001/05/04 01:57:00 EDT
761		Tag: leunga-20010504-sync
762		Description:
763
764		MLRISC features.
765
766		1. Fix to CMPXCHG instructions.
767		2. Changed RA interface to allow annotations in callbacks.
768		3. Added a new method to the stream interface to allow annotations updates.
769
770		----------------------------------------------------------------------
771		Name: Matthias Blume
772		Date: 2001/05/01 11:45:00 EDT
773		Tag: blume-20010501-pcedittmp
774		Description:
775
776		Changed install.sh to use the current working directory instead of
777		/usr/tmp for a temporary file (pcedittmp).  The previous choice
778		of /usr/tmp caused trouble with MacOS X because of file premission
779		problems.
780
781		----------------------------------------------------------------------
782		Name: Matthias Blume
783		Date: 2001/04/20 11:10:00 EDT
784		Tag: blume-20010420-inMLflag
785		Description:
786
787		- added vp_limitPtrMask to vproc-state.h
788		(for use by the raw-C-calls mechanism to implement proper interrupt
789		handling)
790		- made the ML compiler aware of various data-structure offsets so it
791		can generate code for accessing the vp_inML flag and vp_limitPtrMask
792		- tweaked mlriscGen.sml to have it emit interrupt-handling code for
793		raw C-calls
794
795		----------------------------------------------------------------------
796		Name: Lal George
797		Date: 2001/04/20 09:15:28 EDT
798		Tag: george-20010420-macosX
799		Description:
800
801		- Changes to port to Mac OS X; Darwin.
802
803		- In the process I found that sqrt was broken on the PPC, because the
804		fsqrt instruction is not implemented.
805
806		----------------------------------------------------------------------
807		Name: Matthias Blume
808		Date: 2001/04/18 12:45:00 EDT
809		Tag: blume-20010418-ccalls
810		Description:
811
812		- fixed two off-by-4 errors in the x86-specific c-calls implementation
813		(this bug prevented structure arguments containing pointers from being
814		passed correctly)
815		- changed the raw-C-call code in mlriscGen.sml in such a way that
816		structure arguments are represented as a pointer to the beginning
817		of the structure (instead of having a series of synthesized arguments,
818		one for each structure member)
819
820		- made makeml script's verbosity level configurable via environment
821		variable (MAKEML_VERBOSITY)
822
823		- eliminated placeholder implementations for f32l, w16s, i16s, and f32s
824		in rawmem-x86.sml; we are now using the real thing
825
826		----------------------------------------------------------------------
827		Name: Matthias Blume
828		Date: 2001/03/22 16:25:00 EST
829		Tag: blume-20010322-bootfiles
830		Description:
831
832		Created a new set of bootfiles (for your automatic installation convenience).
833
834		----------------------------------------------------------------------
835		Name: Matthias Blume
836		Date: 2001/03/22 15:10:00 EST
837		Tag: blume-20010322-rawmem-parcm
838		Description:
839
840		1. All "raw memory access" primitives for the new FFI are implemented now
841		(at least on the x86).
842		2. Some further cleanup of CM's parallel make mechanism.
843
844		----------------------------------------------------------------------
845		Name: Matthias Blume
846		Date: 2001/03/19 17:53:00 EST
847		Tag: blume-20010319-parallel
848		Description:
849
850		Parallel make (using compile servers) now works again.
851
852		To this end, CM.stabilize and CMB.make have been modified to work in
853		two passes when compile servers are attached:
854		1. Compile everything, do not perform stabilization; this pass
855		uses compile servers
856		2. Stabilize everything; this pass does not use compile servers
857		If there are no compile servers, the two passes are combined into one
858		(as before).  Splitting the passes increases the inherent parallelism
859		in the dependency graph because the entire graph including all
860		libraries is available at the same time.  This, in turn, improves
861		server utilization.  The downside is that the master process will
862		have to do some extra work after compilation is done (because for
863		technical reasons it must re-read all the binfiles during stabilization).
864
865		----------------------------------------------------------------------
866		Name: Matthias Blume
867	Date: 2001/03/16 12:22:00 EST	Date: 2001/03/16 12:22:00 EST
868	Tag: blume-20010316-bootfiles	Tag: blume-20010316-bootfiles
869	Description:	Description:

---

Colored Diff Long Colored Diff Unidiff Context Diff Side by Side

Legend: Removed from v.800   changed lines   Added in v.902

---

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