Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 812, Fri Apr 20 15:07:30 2001 UTC	revision 976, Wed Nov 14 14:54:03 2001 UTC
#	Line 13	Line 13
13	Description:	Description:
14
15	----------------------------------------------------------------------	----------------------------------------------------------------------
16		Name: Lal George
17		Date: 2001/11/14 09:44:04 EST
18		Tag:
19		Description:
20
21		Fix for a backpatching bug reported by Allen.
22
23		Because the boundary between short and long span-dependent
24		instructions is +/- 128, there are an astounding number of
25		span-dependent instructions whose size is over estimated.
26
27		Allen came up with the idea of letting the size of span
28		dependent instructions be non-monotonic, for a maxIter
29		number of times, after which the size must be monotonically
30		increasing.
31
32		This table shows the number of span-dependent instructions
33		whose size was over-estimated as a function of maxIter, for the
34		file Parse/parse/ml.grm.sml:
35
36		maxIter            # of instructions:
37		10                      687
38		20                      438
39		30                      198
40		40                        0
41
42		In compiling the compiler, there is no significant difference in
43		compilation speed between maxIter=10 and maxIter=40. Actually,
44		my measurements showed that maxIter=40 was a tad faster than
45		maxIter=10! Also 96% of the  files in the compiler reach a fix
46		point within 13 iterations, so fixing maxIter at 40, while high,
47		is okay.
48
49		----------------------------------------------------------------------
50		Name: Matthias Blume
51		Date: 2001/10/31 15:25:00 EST
52		Tag: blume-20011031-pgraph
53		Description:
54
55		CKIT:
56		* Changed the "Function" constructor of type Ast.ctype to carry optional
57		argument identifiers.
58		* Changed the return type of TypeUtil.getFunction accordingly.
59		* Type equality ignores the argument names.
60		* TypeUtil.composite tries to preserve argument names but gives up quickly
61		if there is a mismatch.
62
63		installation script:
64		* attempts to use "curl" if available (unless "wget" is available as well)
65
66		CM:
67		* has an experimental implementation of "portable graphs" which I will
68		soon propose as an implementation-independent library format
69		* there are also new libraries $/pgraph.cm and $/pgraph-util.cm
70
71		NLFFI-LIB:
72		* some cleanup (all cosmetic)
73
74		NLFFIGEN:
75		* temporarily disabled the mechanism that suppresses ML output for
76		C definitions whose identifiers start with an underscore character
77		* generate val bindings for enum constants
78		* user can request that only one style (light or heavy) is being used;
79		default is to use both (command-line arguments: -heavy and -light)
80		* fixed bug in handling of function types involving incomplete pointers
81		* generate ML entry points that take record arguments (i.e., using
82		named arguments) for C functions that have a prototype with named
83		arguments
84		(see changes to CKIT)
85
86		----------------------------------------------------------------------
87		Name: Allen Leung
88		Date: 2001/10/27 20:34:00 EDT
89		Tag: leunga-20011027-x86-fast-fp-call
90		Description:
91
92		Fixed the bug described in blume-20010920-slowfp.
93
94		The fix involves
95		1. generating FCOPYs in FSTP in ia32-svid
96		2. marking a CALL with the appropriate annotation
97
98		----------------------------------------------------------------------
99		Name: Matthias Blume
100		Date: 2001/10/16 11:32:00 EDT
101		Tag: blume-20011016-netbsd
102		Description:
103
104		Underscore patch from Chris Richards (fixing problem with compiling
105		runtime system under recent NetBSD).
106
107		----------------------------------------------------------------------
108		Name: Allen Leung
109		Date: 2001/10/12 17:18:32 EDT 2001
110		Tag: leung-20011012-x86-printflowgraph
111		Description:
112
113		X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
114
115		----------------------------------------------------------------------
116		Name: Lal George
117		Date: 2001/10/11 23:51:34 EDT
118		Tag: george-20011011-too-many-instrs
119		Description:
120
121		The representation of a program point never expected to see more
122		than 65536 instructions in a basic block!
123
124		----------------------------------------------------------------------
125		Name: Lal George
126		Date: 2001/10/09 09:41:37 EDT
127		Tag: george-20011008-mlrisc-labels
128		Description:
129
130		Changed the machine description files to support printing of
131		local and global labels in assembly code, based on host assembler
132		conventions.
133
134		----------------------------------------------------------------------
135		Name: Matthias Blume
136		Date: 2001/09/25 15:25:00 EDT
137		Tag: blume-20010925-exninfo
138		Description:
139
140		I provided a non-hook implementation of exnName (at the toplevel) and
141		made the "dummy" implementation of exnMessage (at the toplevel) more
142		useful: if nothing gets "hooked in", then at least you are going to
143		see the exception name and a message indicating why you don't see more.
144
145		[For the time being, programs that need exnMessage and want to use
146		ml-build should either use General.exnMessage (strongly recommended) or
147		refer to structure General at some other point so that CM sees a
148		static dependency.]
149
150		[Similar remarks go for "print" and "use":  If you want to use their
151		functionality in stand-alone programs generated by ml-build, then use
152		TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
153
154		----------------------------------------------------------------------
155		Name: Matthias Blume
156		Date: 2001/09/20 17:28:00 EDT
157		Tag: blume-20010920-slowfp
158		Description:
159
160		Allen says that x86-fast-fp is not safe yet, so I turned it off again...
161
162		----------------------------------------------------------------------
163		Name: Matthias Blume
164		Date: 2001/09/20 17:20:00 EDT
165		Tag: blume-20010920-canonicalpaths
166		Description:
167
168		0. Updated the BOOT file (something that I forgot to do earlier).
169
170		1. Small internal change to CM so that it avoids "/../" in filenames
171		as much as possible (but only where it is safe).
172
173		2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
174		that arguments that contain delimiters are passed through correctly.
175		This change also means that all "special" arguments of the form
176		@SMLxxx... must come first.
177
178		3. Changed install script to put relative anchor names for tool commands
179		into pathconfig.
180
181		----------------------------------------------------------------------
182		Name: Matthias Blume
183		>>>>>>> 1.169
184		Date: 2001/09/18 15:35:00 EDT
185		Tag: blume-20010918-readme11036
186		Description:
187
188		Added README files.
189
190		----------------------------------------------------------------------
191		Name: Matthias Blume
192		Date: 2001/09/18 11:45:00 EDT
193		Tag: Release_110_36 (retag)
194		Description:
195
196		Fixed mistake in config/preloads. Retagged as 110.36.
197
198		----------------------------------------------------------------------
199		Name: Matthias Blume
200		Date: 2001/09/18 09:40:00 EDT
201		Tag: Release_110_36_orig (tag changed)
202		Description:
203
204		New version (110.36).  New bootfiles.
205
206		----------------------------------------------------------------------
207		Name: Matthias Blume
208		Date: 2001/09/14 16:15:00 EDT
209		Tag: blume-20010914-x86fastfp
210		Description:
211
212		John committed some changes that Allen made, in particular a (hopefully)
213		correctly working version of the x86-fp module.
214
215		I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
216		flag to "true".  Everything seems to compile to a fixpoint ok, and
217		"mandelbrot" speeds up by about 15%.
218
219		----------------------------------------------------------------------
220		Name: Matthias Blume
221		Date: 2001/09/13 11:20:00 EDT
222		Tag: blume-20010913-minimal
223		Description:
224
225		1. Stefan Monnier's patch to fix a miscompilation problem that
226		was brought to light by John Reppy's work on Moby.
227
228		2. Implemented a minimal "structure Compiler" that contains just
229		"version" and "architecture".  The minimal version will be
230		available when the full version is not.  This is for backward-
231		compatibility with code that wants to test Compiler.version.
232
233		----------------------------------------------------------------------
234		Name: Matthias Blume
235		Date: 2001/08/28 14:03:00 EDT
236		Tag: blume-20010828-ml-lex
237		Description:
238
239		Fix for bug 1581, received from Neophytos Michael.
240
241		----------------------------------------------------------------------
242		Name: Matthias Blume
243		Date: 2001/08/27 11:20:00 EDT
244		Tag: blume-20010827-readme11035
245		Description:
246
247		Fleshed out the README file for 110.35.
248
249		----------------------------------------------------------------------
250		Name: Matthias Blume
251		Date: 2001/08/24 17:10:00 EDT
252		Tag: Release_110_35
253		Description:
254
255		New version number (110.35).  New bootfiles.
256
257		----------------------------------------------------------------------
258		Name: Lal George
259		Date: 2001/08/24 13:47:18 EDT 2001
260		Tag: george-20010824-MLRISC-graphs
261		Description:
262
263		removed clusters from MLRISC completely and replaced with graphs.
264
265		----------------------------------------------------------------------
266		Name: Matthias Blume
267		Date: 2001/08/23 17:50:00 EDT
268		Tag: blume-20010823-toplevel
269		Description:
270
271		- some reorganization of the code that implements various kinds of
272		environments in the compiler (static, dynamic, symbolic, combined)
273		- re-implemented the EnvRef module so that evalStream works properly
274		(if the stream contains references to "use", "CM.make", etc.)
275		- cleaned up evalloop.sml and interact.sml (but they need more cleaning)
276
277		----------------------------------------------------------------------
278		Name: Matthias Blume
279		Date: 2001/08/20 15:50 EDT
280		Tag: blume20010820-slipup
281		Description:
282
283		I forgot to commit a few files.  Here they are...
284
285		----------------------------------------------------------------------
286		Name: Matthias Blume
287		Date: 2001/08/20 15:35:00 EDT
288		Tag: blume-20010820-debugprof
289		Description:
290
291		!!!! NEW BOOTFILES !!!!
292
293		This is another round of reorganizing the compiler sources.  This
294		time the main goal was to factor out all the "instrumentation"
295		passes (for profiling and backtracing) into their own library.
296		The difficulty was to do it in such a way that it does not depend
297		on elaborate.cm but only on elabdata.cm.
298
299		Therefore there have been further changes to both elaborate.cm and
300		elabdata.cm -- more "generic" things have been moved from the former
301		to the latter.  As a result, I was forced to split the assignment
302		of numbers indicating "primtyc"s into two portions: SML-generic and
303		SML/NJ-specific.  Since it would have been awkward to maintain,
304		I bit the bullet and actually _changed_ the mapping between these
305		numbers and primtycs.  The bottom line of this is that you need
306		a new set of bin- and bootfiles.
307
308		I have built new bootfiles for all architectures, so doing a fresh
309		checkout and config/install.sh should be all you need.
310
311		The newly created library's name is
312
313		$smlnj/viscomp/debugprof.cm
314
315		and its sources live under
316
317		src/compiler/DebugProf
318
319		----------------------------------------------------------------------
320		Name: Matthias Blume
321		Date: 2001/08/15 17:15:00 EDT
322		Tag: blume-20010815-compreorg
323		Description:
324
325		This is a first cut at reorganizing the CM libraries that make up the
326		core of the compiler.  The idea is to separate out pieces that could
327		be used independently by tools, e.g., the parser, the typechecker, etc.
328
329		The current status is a step in this direction, but it is not quite
330		satisfactory yet.  Expect more changes in the future.
331
332		Here is the current (new) organization...
333
334		What used to be $smlnj/viscomp/core.cm is now divided into
335		six CM libraries:
336
337		$smlnj/viscomp/basics.cm
338		/parser.cm
339		/elabdata.cm
340		/elaborate.cm
341		/execute.cm
342		/core.cm
343
344		The CM files for these libraries live under src/system/smlnj/viscomp.
345		All these libraries are proxy libraries that contain precisely
346		one CM library component.  Here are the locations of the components
347		(all within the src/compiler tree):
348
349		Basics/basics.cm
350		Parse/parser.cm
351		ElabData/elabdata.cm
352		Elaborator/elaborate.cm
353		Execution/execute.cm
354		core.cm
355
356		[This organization is the same that has been used already
357		for a while for the architecture-specific parts of the visible
358		compiler and for the old version of core.cm.]
359
360		As you will notice, many source files have been moved from their
361		respective original locations to a new home in one of the above
362		subtrees.
363
364		The division of labor between the new libraries is the following:
365
366		basics.cm:
367		- Simple, basic definitions that pertain to many (or all) of
368		the other libraries.
369		parser.cm:
370		- The SML parser, producing output of type Ast.dec.
371		- The type family for Ast is also defined and exported here.
372		elabdata.cm:
373		- The datatypes that describe input and output of the elaborator.
374		This includes types, absyn, and static environments.
375		elaborator.cm:
376		- The SML/NJ type checker and elaborator.
377		This maps an Ast.dec (with a given static environment) to
378		an Absyn.dec (with a new static environment).
379		- This libraries implements certain modules that used to be
380		structures as functors (to remove dependencies on FLINT).
381		execute.cm:
382		- Everything having to do with executing binary code objects.
383		- Dynamic environments.
384		core.cm:
385		- SML/NJ-specific instantiations of the elaborator and MLRISC.
386		- Top-level modules.
387		- FLINT (this should eventually become its own library)
388
389		Notes:
390
391		I am not 100% happy with the way I separated the elaborator (and its
392		data structures) from FLINT.  Two instances of the same problem:
393
394		1. Data structures contain certain fields that carry FLINT-specific
395		information.  I hacked around this using exn and the property list
396		module from smlnj-lib.  But the fact that there are middle-end
397		specific fields around at all is a bit annoying.
398
399		2. The elaborator calculates certain FLINT-related information.  I tried
400		to make this as abstract as I could using functorization, but, again,
401		the fact that the elaborator has to perform calculations on behalf
402		of the middle-end at all is not nice.
403
404		3. Having to used exn and property lists is unfortunate because it
405		weakens type checking.  The other alternative (parameterizing
406		nearly *everything*) is not appealing, though.
407
408		I removed the "rebinding =" warning hack because due to the new organization
409		it was awkward to maintain it.  As a result, the compiler now issues some of
410		these warnings when compiling init.cmi during bootstrap compilation. On
411		the plus side, you also get a warning when you do, for example:
412		val op = = Int32.+
413		which was not the case up to now.
414
415		I placed "assign" and "deref" into the _Core structure so that the
416		code that deals with the "lazy" keyword can find them there.  This
417		removes the need for having access to the primitive environment
418		during elaboration.
419
420		----------------------------------------------------------------------
421		Name: Matthias Blume
422		Date: 2001/08/13
423		Tag: blume-20010813-closures
424		Description:
425
426		This fix was sent to us by Zhong Shao.  It is supposed to improve the
427		performance of certain loops by avoiding needless closure allocation.
428
429		----------------------------------------------------------------------
430		Name: Lal George
431		Date: 2001/07/31 10:03:23 EDT 2001
432		Tag: george-20010731-x86-fmalloc
433		Description: Fixed bug in x86 calls
434
435		There was a bug where call instructions would mysteriously
436		vanish. The call instruction had to be one that returned
437		a floating point value.
438
439		----------------------------------------------------------------------
440		Name: Lal George
441		Date: 2001/07/19 16:36:29 EDT 2001
442		Tag: george-20010719-simple-cells
443		Description:
444
445		I have dramatically simplified the interface for CELLS in MLRISC.
446
447		In summary, the cells interface is broken up into three parts:
448
449		1. CellsBasis : CELLS_BASIS
450
451		CellsBasis is a top level structure and common for all
452		architectures.  it contains the definitions of basic datatypes
453		and utility  functions over these types.
454
455		2. functor Cells() : CELLS
456
457		Cells generates an interface for CELLS that incorporates the
458		specific resources on the target architecture, such as the
459		presence of special register classes, their number and size,
460		and various useful substructures.
461
462		3. <ARCH>CELLS
463
464		e.g. SparcCells: SPARCCELLS
465
466		<ARCH>CELLS usually contains additional bindings for special
467		registers  on the architecture, such as:
468
469		val r0 : cell           (* register zero *)
470		val y : cell            (* Y register *)
471		val psr : cell          (* processor status register *)
472		...
473
474		The structure returned by applying the Cells functor is opened
475		in this interface.
476
477		The main implication of all this is that the datatypes for cells is
478		split between CellsBasis and CELLS -- a fairly simple change for user
479		code.
480
481		In the old scheme the CELLS interface had a definitional binding of
482		the form:
483
484		signature CELLS = sig
485
486		structure CellsBasis = CellsBasis
487
488		...
489
490		end
491
492		With all the sharing constraints that goes on in MLRISC, this old
493		design  quickly leads to errors such as:
494
495		"structure definition spec inside of sharing ... "
496
497
498		and appears to require an unacceptable amount of sharing and where
499		constraint hackery.
500
501		I think this error message (the interaction of definitional specs and
502		sharing) requires more explanation on our web page.
503
504		----------------------------------------------------------------------
505		Name: Matthias Blume
506		Date: 2001/07/19 15:00:00 EDT
507		Tag: blume-20010719-libreorg
508		Description:
509
510		This update puts together a fairly extensive but straightforward change
511		to the way the libraries that implement the interactive system are
512		organized:
513
514		The biggest change is the elimination of structure Compiler.  As a
515		replacement for this structure, there is now a CM library
516		(known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
517		that exports all the substructures of the original structure Compiler
518		directly.  So instead of saying Compiler.Foo.bar one now simply
519		says Foo.bar.  (The CM libraries actually export a collection of
520		structures that is richer than the collection of substructures of
521		structure Compiler.)
522
523		To make the transition smooth, there is a separate library called
524		$smlnj/compiler/compiler.cm which puts together and exports the
525		original structure Compiler (or at least something very close to it).
526
527		There are five members of the original structure Compiler
528		that are not exported directly but which instead became members
529		of a new structure Backend (described by signature BACKEND).  These are:
530		structure Profile (: PROFILE), structure Compile (: COMPILE), structure
531		Interact (: INTERACT), structure Machine (: MACHINE), and val
532		architecture (: string).
533
534		Structure Compiler.Version has become structure CompilerVersion.
535
536		Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
537		by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
538		or x86, respectively.
539		Each of these exports the same frontend structures that
540		$smlnj/compiler.cm exports.  But they do not have a structure Backend
541		and instead export some structure <Arch>Backend where <Arch> is Alpha32,
542		Hppa, PPC, Sparc, or X86, respectively.
543
544		Library $smlnj/compiler/all.cm exports the union of the exports of
545		$smlnj/compiler/<arch>.cm
546
547		There are no structures <Arch>Compiler anymore, use
548		$smlnj/compiler/<arch>.cm instead.
549
550		Library host-compiler-0.cm is gone.  Instead, the internal library
551		that instantiates CM is now called cm0.cm.  Selection of the host
552		compiler (backend) is no longer done here but. (Responsibility for it
553		now lies with $smlnj/compiler/current.cm.  This seems to be more
554		logical.)
555
556		Many individual files have been moved or renamed.  Some files have
557		been split into multiple files, and some "dead" files have been deleted.
558
559		Aside from these changes to library organization, there are also changes
560		to the way the code itself is organized:
561
562		Structure Binfile has been re-implemented in such a way that it no
563		longer needs any knowledge of the compiler.  It exclusively deals
564		with the details of binfile layout.  It no longer invokes the
565		compiler (for the purpose of creating new prospective binfile
566		content), and it no longer has any knowledge of how to interpret
567		pickles.
568
569		Structure Compile (: COMPILE) has been stripped down to the bare
570		essentials of compilation.  It no longer deals with linking/execution.
571		The interface has been cleaned up considerably.
572
573		Utility routines for dealing with linking and execution have been
574		moved into their own substructures.
575
576		(The ultimate goal of these changes is to provide a light-weight
577		binfile loader/linker (at least for, e.g., stable libraries) that
578		does not require CM or the compiler to be present.)
579
580		CM documentation has been updated to reflect the changes to library
581		organization.
582
583		----------------------------------------------------------------------
584		Name: Matthias Blume
585		Date: 2001/07/10 17:30:00 EDT
586		Tag: Release_110_34
587		Description:
588
589		Minor tweak to 110.34 (re-tagged):
590
591		- README.html file added to CVS repository
592		- runtime compiles properly under FreeBSD 3.X and 4.X
593
594		----------------------------------------------------------------------
595		Name: Matthias Blume
596		Date: 2001/07/10 17:30:00 EDT
597		Tag: Release_110_34
598		Description:
599
600		New version number (110.34). New bootfiles.
601
602		----------------------------------------------------------------------
603		Name: Matthias Blume
604		Date: 2001/07/09 16:00:00 EDT
605		Tag: blume-20010709-more-varargs
606		Description:
607
608		I changed the handling of varargs in ml-nlffigen again:
609		The ellipsis ... will now simply be ignored (with an accompanying warning).
610
611		The immediate effect is that you can actually call a varargs function
612		from ML -- but you can't actually supply any arguments beyond the ones
613		specified explicitly.  (For example, you can call printf with its format
614		string, but you cannot pass additional arguments.)
615
616		This behavior is only marginally more useful than the one before, but
617		it has the advantage that a function or, more importantly, a function
618		type never gets dropped on the floor, thus avoiding follow-up problems with
619		other types that refer to the offending one.
620
621		----------------------------------------------------------------------
622		Name: Matthias Blume
623		Date: 2001/07/09 11:25:00 EDT
624		Tag: blume-20010709-varargs
625		Description:
626
627		1. ckit-lib.cm now exports structure Error
628		2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
629		with a warning accompanied by a source location.  Moreover, it
630		merely skips the offending function or type and proceeds with the
631		rest of its work.u  As a result, one can safely feed C code containing
632		"..." to ml-nlffigen.
633		3. There are some internal improvements to CM, providing slightly
634		more general string substitutions in the tools subsystem.
635
636		----------------------------------------------------------------------
637		Name: Matthias Blume
638		Date: 2001/06/27 15:10:00 EDT
639		Tag: blume-20010627-concur
640		Description:
641
642		Fixed a small bug in CM's handling of parallel compilation.
643		(You could observe the bug by Control-C-interrupting an ordinary
644		CMB.make or CM.stabilize and then attaching some compile servers.
645		The result was that all of a sudden the previously interrupted
646		compilation would continue on its own.  This was because of
647		an over-optimization: CM did not bother to clean out certain queues
648		when no servers were attached "anyway", resulting in the contents
649		of these queues to grab control when new servers did get attached.)
650
651		There is also another minor update to the CM manual.
652
653		----------------------------------------------------------------------
654		Name: Matthias Blume
655		Date: 2001/06/26 16:15:00 EDT
656		Tag: blume-20010626-cmdoc
657		Description:
658
659		Minor typo fixed in CM manual (syntax diagram for libraries).
660
661		----------------------------------------------------------------------
662		Name: Matthias Blume
663		Date: 2001/06/25 22:55:00 EDT
664		Tag: blume-20010625-x86pc
665		Description:
666
667		Fixed a nasty bug in the X86 assembly code that caused signal
668		handlers to fail (crash) randomly.
669
670		----------------------------------------------------------------------
671		Name: Matthias Blume
672		Date: 2001/06/25 12:05:00 EDT
673		Tag: blume-20010625-nlffigen
674		Description:
675
676		This update fixes a number of minor bugs in ml-nlffigen as reported by
677		Nick Carter <nbc@andrew.cmu.edu>.
678
679		1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
680		2. Default names for generated files are now derived from the name of
681		the C file *without its directory*.  In particular, this causes generated
682		files to be placed locally even if the C file is in some system directory.
683		3. Default names for generated signatures and structures are also derived
684		from the C file name without its directory.  This avoids silly things
685		like "structure GL/GL".
686		(Other silly names are still possible because ml-nlffigen does not do
687		a thorough check of whether generated names are legal ML identifiers.
688		When in doubt, use command line arguments to force particular names.)
689
690		----------------------------------------------------------------------
691		Name: Matthias Blume
692		Date: 2001/06/21 12:25:00 EDT
693		Tag: blume-20010621-eXene
694		Description:
695
696		eXene now compiles and (sort of) works again.
697
698		The library name (for version > 110.33) is $/eXene.cm.
699
700		I also added an new example in src/eXene/examples/nbody.  See the
701		README file there for details.
702
703		----------------------------------------------------------------------
704		Name: Matthias Blume
705		Date: 2001/06/20 16:40:00 EDT
706		Tag: blume-20010620-cml
707		Description:
708
709		CML now compiles and works again.
710
711		Libraries (for version > 110.33):
712
713		$cml/cml.cm            Main CML library.
714		$cml/basis.cm          CML's version of $/basis.cm.
715		$cml/cml-internal.cm   Internal helper library.
716		$cml/core-cml.cm       Internal helper library.
717		$cml-lib/trace-cml.cm  Tracing facility.
718		$cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
719
720		The installer (config/install.sh) has been taught how to properly
721		install this stuff.
722
723		----------------------------------------------------------------------
724		Name: Matthias Blume
725		Date: 2001/06/19 17:55:00 EDT
726		Tag: blume-20010619-instantiate
727		Description:
728
729		This un-breaks the fix for bug 1432.
730		(The bug was originally fixed in 110.9 but I broke it again some
731		time after that.)
732
733		----------------------------------------------------------------------
734		Name: Matthias Blume
735		Date: 2001/06/19 17:25:00 EDT
736		Tag: blume-20010619-signals
737		Description:
738
739		This should (hopefully) fix the long-standing signal handling bug.
740		(The runtime system was constructing a continuation record with an
741		incorrect descriptor which would cause the GC to drop data on the floor...)
742
743		----------------------------------------------------------------------
744		Name: Matthias Blume
745		Date: 2001/06/15 15:05:00 EDT
746		Tag: blume-20010615-moresparc
747		Description:
748
749		Here is a short late-hour update related to Sparc c-calls:
750
751		-- made handling of double-word arguments a bit smarter
752
753		-- instruction selection phase tries to collapse certain clumsily
754		constructed ML-Trees; typical example:
755
756		ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
757
758		This currently has no further impact on SML/NJ since mlriscGen does
759		not seem to generate such patterns in the first place, and c-calls
760		(which did generate them in the beginning) has meanwhile been fixed
761		so as to avoid them as well.
762
763		----------------------------------------------------------------------
764		Name: Matthias Blume
765		Date: 2001/06/15 15:05:00 EDT
766		Tag: blume-20010615-sparc
767		Description:
768
769		The purpose of this update is to provide an implementation of NLFFI
770		on Sparc machines.
771
772		Here are the changes in detail:
773
774		* src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
775		the Sparc implementation of the c-calls API.
776		* The Sparc backend of SML/NJ has been modified to uniformely use %fp
777		for accessing the ML frame.  Thus, we have a real frame pointer and
778		can freely modify %sp without need for an omit-frame-ptr phase.
779		The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
780		this case.
781		* ml-nlffigen has been taught to produce code for different architectures
782		and calling conventions.
783		* In a way similar to what was done in the x86 case, the Sparc
784		backend uses its own specific extension to mltree.  (For example,
785		it needs to be able to generate UNIMP instructions which are part
786		of the calling convention.)
787		* ml-nlffi-lib was reorganized to make it more modular (in particular,
788		to make it easier to plug in new machine- and os-dependent parts).
789
790		There are some other fairly unrelated bug fixes and cleanups as well:
791
792		* I further hacked the .cm files for MLRISC tools (like MDLGen) so
793		that they properly share their libraries with existing SML/NJ libraries.
794		* I fixed a minor cosmetic bug in CM, supressing certain spurious
795		follow-up error messages.
796		* Updates to CM/CMB documentation.
797
798		TODO items:
799
800		* MLRISC should use a different register as its asmTemp on the Sparc.
801		(The current %o2 is a really bad choice because it is part of the
802		calling conventions, so things might interfere in unexpected ways.)
803
804		----------------------------------------------------------------------
805		Name: Matthias Blume
806		Date: 2001/06/07
807		Tag: blume-20010607-calls
808		Description:
809
810		A number of internal changes related to C calls and calling conventions:
811
812		1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
813		number of bytes popped implicitly (by the callee).  In most cases
814		this field is 0 but on x86/win32 it is some non-zero value.  This
815		is information provided for the benefit of the "omit-frameptr" pass.
816		2. The CALL instruction on the x86 carries a similar "pops" field.
817		The instruction selection phase copies its value from the ML-Tree
818		CALL statement.
819		3. On all other architectures, the instruction selection phase checks
820		whether "pops=0" and complains if not.
821		4. The c-calls implementation for x86 now accepts two calling conventions:
822		"ccall" and "stdcall".  When "ccall" is selected, the caller cleans
823		up after the call and pops is set to 0.  For "stdcall", the caller
824		does nothing, leaving the cleanup to the callee; pops is set to
825		the number of bytes that were pushed onto the stack.
826		5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
827		distinguish between "ccall" and "stdcall".
828		6. The UNIMP instruction has been added to the supported Sparc instruction
829		set. (This is needed for implementing the official C calling convention
830		on this architecture.)
831		7. I fixed some of the .cm files under src/MLRISC/Tools to make them
832		work with the latest CM.
833
834		----------------------------------------------------------------------
835		Name: Matthias Blume
836		Date: 2001/06/05 15:10:00 EDT
837		Tag: blume-20010605-cm-index
838		Description:
839
840		0. The "lambdasplit" parameter for class "sml" in CM has been documented.
841
842		1. CM can now generate "index files".  These are human-readable files
843		that list on a per-.cm-file basis each toplevel symbol defined or
844		imported.  The location of the index file for
845		<p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
846		To enable index-file generation, set CM.Control.generate_index to true
847		or export an environment-symbol: export CM_GENERATE_INDEX=true.
848
849		The CM manual has been updated accordingly.
850
851		2. I made some slight modifications to the c-calls API in MLRISC.
852
853		a) There is now a callback to support saving/restoring of
854		dedicated but caller-save registers around the actual call
855		instruction.
856		b) One can optionally specify a comment-annotation for the
857		call instruction.
858
859		3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
860		(For example, the comment annotation shows the C prototype of
861		the function being called.)
862
863		----------------------------------------------------------------------
864		Name: Matthias Blume
865		Date: 2001/06/01 13:30:00 EDT
866		Tag: blume-20010601-nlffi-cleanup
867		Description:
868
869		This is mostly a cleanup of MLFFI stuff:
870
871		- some signature files have been put into a more exposed place
872		- the ugly 'f type parameter is gone (simplifies types tremendously!)
873		- ml-nlffigen changed accordingly
874		- tutorial updated
875
876		Other changes:
877
878		- author's affiliation in CM manual(s) updated
879		- some more recognized keywords added to Allen's sml.sty
880
881		----------------------------------------------------------------------
882		Name: Matthias Blume
883		Date: 2001/05/25 15:30:00 EDT
884		Tag: blume-20010525-iptr
885		Description:
886
887		- put the official 110.33-README (as it appears on the ftp server) under
888		CVS
889		- fixed a small bug related to incomplete pointer types in
890		ml-nlffigen
891		- small cosmetic change to the ml-nlffi-lib's "arr" type constructor
892		(it does not need the 'f type parameter)
893
894		----------------------------------------------------------------------
895		Name: Matthias Blume
896		Date: 2001/05/23 14:30:00 EDT
897		Tag: Release_110_33
898		Description:
899
900		New version number (110.33).  New bootfiles.
901
902		----------------------------------------------------------------------
903		Name: Matthias Blume
904		Date: 2001/05/22 18:06:00 EDT
905		Tag: blume-20010522-targets
906		Description:
907
908		Made install.sh use file config/targets.customized if it exists, falling
909		back to config/targets if it doesn't.  This way one can have a customized
910		version of the targets file without touching the "real thing", thus
911		eliminating the constant fear of accidentally checking something bogus
912		back into the CVS repository...  (File config/targets.customized must
913		not be added to the repository!)
914
915		----------------------------------------------------------------------
916		Name: Matthias Blume
917		Date: 2001/05/22 16:30:00 EDT
918		Tag: blume-20010522-minitut
919		Description:
920
921		1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
922		struct returns.
923		2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
924		incomplete, preliminary documentation for NLFFI.
925
926		----------------------------------------------------------------------
927		Name: Matthias Blume
928		Date: 2001/05/14 11:30:00 EDT
929		Tag: blume-20010514-script
930		Description:
931
932		Some bugs in install script fixed.
933
934		In addition to that I also made a slight change to the NLFFI API:
935		Functors generated by ml-nlffigen now take the dynamic library as a
936		straight functor argument, not as a suspended one.  (The original
937		functor code used to force the suspension right away anyway, so there
938		was nothing gained by this complication of the interface.)
939
940		----------------------------------------------------------------------
941		Name: Matthias Blume
942		Date: 2001/05/11 14:35:00 EDT
943		Tag: blume-20010511-ml-nlffi
944		Description:
945
946		I finally took the plunge and added my new FFI code to the main
947		repository.  For x86-linux it is now ready for prime-time.
948
949		There are two new subdirectories of "src":
950
951		- ml-nlffi-lib:
952		The utility library for programs using the FFI interface.
953		Here is the implementation of $/c.cm and its associated low-level
954		partners $/c-int.cm and $/memory.cm.
955		- ml-nlffigen:
956		A stand-alone program for generating ML glue code from C source
957		code.
958
959		Building ml-nlffigen requires $/ckit-lib.cm.
960
961		The config/install.sh script has been updates to do the Right Thing
962		(hopefully).
963
964		Notice that the source tree for the C-Kit will not be put under "src"
965		but directly under the installation root directory.  (This is the
966		structure that currently exists on the CVS server when you check out
967		module "sml".)  Fortunately, config/install.sh knows about this oddity.
968
969		Bugs: No documentation yet.
970
971		----------------------------------------------------------------------
972		Name: Matthias Blume
973		Date: 2001/05/09 16:35:00 EDT
974		Tag: blume-20010509-cpscontract
975		Description:
976
977		Fixed a bug in the accounting code in cpsopt/contract.sml.  (The
978		wrapper/unwrapper elimination did not decrement usage counts and some
979		dead variables got overlooked by the dead-up logic.)
980
981		----------------------------------------------------------------------
982		Name: Lal George
983		Date: 2001/05/08  17:26:09 EDT
984		Tag: george-20010508-omit-frameptr
985		Description:
986
987		Changes to implement the omit-frame-pointer optimization to support
988		raw C calls. For now, there is only support on the Intel x86, but
989		other architectures will follow as more experience is gained with this.
990
991
992		----------------------------------------------------------------------
993		Name: Matthias Blume
994		Date: 2001/05/07 14:40:00 EDT
995		Tag: blume-20010507-proxies
996		Description:
997
998		I made into "proxy libraries" all libraries that qualify for such a
999		change.  (A qualifying library is a library that has another library or
1000		groups as its sole member and repeats that member's export list
1001		verbatim.  A proxy library avoids this repetition by omitting its export
1002		list, effectively inheriting the list that its (only) member exports.
1003		See the CM manual for more explanation.)
1004		The main effect is that explicit export lists for these libraries
1005		do not have to be kepts in sync, making maintenance a bit easier.
1006
1007		I also added copyright notices to many .cm-files.
1008
1009		Last but not least, I made a new set of bootfiles.
1010
1011		----------------------------------------------------------------------
1012		Name: Matthias Blume
1013		Date: 2001/05/04 17:00:00 EDT
1014		Tag: blume-20010504-cm-lsplit
1015		Description:
1016
1017		0. John merged pending changes to $/smlnj-lib.cm
1018
1019		1. Allen's previous change accidentally backed out of one of Lal's
1020		earlier changes.  I undid this mistake (re-introducing Lal's change).
1021
1022		2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
1023		within the compiler where applicable.  There is some code simplification
1024		because of that.
1025
1026		3. The "split" phase (in FLINT) is now part of the default list of phases.
1027		Compiler.Control.LambdaSplitting.* can be used to globally control the
1028		lambda-splitting (cross-module-inlining) engine.  In addition to that,
1029		it can now also be controlled on a per-source basis: CM has been taught
1030		a new tool parameter applicable to ML source files.
1031
1032		- To turn lambda-splitting off completely:
1033		local open Compiler.Control.LambdaSplitting in
1034		val _ = set Off
1035		end
1036		- To make "no lambda-splitting" the global default (but allow per-source
1037		overriding); this is the initial setting:
1038		local open Compiler.Control.LambdaSplitting in
1039		val _ = set (Default NONE)
1040		end
1041		- To make "lambda-splitting with aggressiveness a" the global default
1042		(and allow per-source overriding):
1043		local open Compiler.Control.LambdaSplitting in
1044		val _ = set (Default (SOME a))
1045		end
1046
1047		- To turn lambda-splitting off for a given ML souce file (say: a.sml)
1048		write (in the respective .cm-file):
1049		a.sml (lambdasplitting:off)
1050		- To turn lambda-splitting for a.sml on with minimal aggressiveness:
1051		a.sml (lambdasplitting:on)
1052		- To turn lambda-splitting for a.sml on with aggressiveness <a> (where
1053		<a> is a decimal non-negative integer):
1054		a.sml (lambdasplitting:<a>)
1055		- To turn lambda-splitting for a.sml on with maximal aggressiveness:
1056		a.sml (lambdasplitting:infinity)
1057		- To use the global default for a.sml:
1058		a.sml (lambdasplitting:default)
1059		or simply
1060		a.sml
1061
1062		----------------------------------------------------------------------
1063		Name: Allen Leung
1064		Date: 2001/05/04 01:57:00 EDT
1065		Tag: leunga-20010504-sync
1066		Description:
1067
1068		MLRISC features.
1069
1070		1. Fix to CMPXCHG instructions.
1071		2. Changed RA interface to allow annotations in callbacks.
1072		3. Added a new method to the stream interface to allow annotations updates.
1073
1074		----------------------------------------------------------------------
1075		Name: Matthias Blume
1076		Date: 2001/05/01 11:45:00 EDT
1077		Tag: blume-20010501-pcedittmp
1078		Description:
1079
1080		Changed install.sh to use the current working directory instead of
1081		/usr/tmp for a temporary file (pcedittmp).  The previous choice
1082		of /usr/tmp caused trouble with MacOS X because of file premission
1083		problems.
1084
1085		----------------------------------------------------------------------
1086	Name: Matthias Blume	Name: Matthias Blume
1087	Date: 2001/04/20 11:10:00 EDT	Date: 2001/04/20 11:10:00 EDT
1088	Tag: blume-20010420-inMLflag	Tag: blume-20010420-inMLflag
#	Line 54	Line 1124
1124	- made makeml script's verbosity level configurable via environment	- made makeml script's verbosity level configurable via environment
1125	variable (MAKEML_VERBOSITY)	variable (MAKEML_VERBOSITY)
1126
1127	- eliminteded placeholder implementations for f32l, w16s, i16s, and f32s	- eliminated placeholder implementations for f32l, w16s, i16s, and f32s
1128	in rawmem-x86.sml; we are now using the real thing	in rawmem-x86.sml; we are now using the real thing
1129
1130	----------------------------------------------------------------------	----------------------------------------------------------------------
#	Line 2087	Line 3157
3157	elaborator). There were a lot of changes during my "linkpath" trials	elaborator). There were a lot of changes during my "linkpath" trials
3158	that could have been reverted to their original state but weren't.	that could have been reverted to their original state but weren't.
3159	Please, don't be too harsh on me for messing with this code a bit more	Please, don't be too harsh on me for messing with this code a bit more
3160	than what was strictly necessary...  (I _did_ resist the tempation	than what was strictly necessary...  (I _did_ resist the temptation
3161	of doing any "global reformatting" to avoid an untimely death at	of doing any "global reformatting" to avoid an untimely death at
3162	Dave's hands. :)	Dave's hands. :)
3163

---

Colored Diff Long Colored Diff Unidiff Context Diff Side by Side

Legend: Removed from v.812   changed lines   Added in v.976

---

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