Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 844, Wed Jun 20 20:39:15 2001 UTC	revision 1045, Mon Jan 28 22:13:13 2002 UTC
#	Line 8	Line 8
8	The form of an entry should be:	The form of an entry should be:
9
10	Name:	Name:
11	Date:	Date: yyyy/mm/dd
12	Tag: <post-commit CVS tag>	Tag: <post-commit CVS tag>
13	Description:	Description:
14
15	----------------------------------------------------------------------	----------------------------------------------------------------------
16	Name: Matthias Blume	Name: Matthias Blume
17		Date: 2002/01/28
18		Tag: blume-20020128-sparc-ccalls
19		Description:
20
21		- Fixed problem with calculation of "used" registers in sparc-c-calls.
22		- Make use of the allocParam argument in sparc-c-calls.
23
24		----------------------------------------------------------------------
25		Name: Matthias Blume
26		Date: 2002/01/28
27		Tag: blume-20020128-allocParam
28		Description:
29
30		John Reppy:  Changes c-calls API to accept client-callback for
31		allocating extra stack space.
32		me: Corresponding changes to mlriscGen (using a dummy argument that
33		does not change the current behavior).
34
35		----------------------------------------------------------------------
36		Name: Matthias Blume
37		Date: 2002/01/28 12:00:00
38		Tag: Release_110_38
39		Description:
40
41		This time for real!!!
42
43		----------------------------------------------------------------------
44		Name: Matthias Blume
45		Date: 2002/01/28 10:56:00 EST
46		Tag: blume-20020128-retraction
47		Description:
48
49		0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
50		with blume-Release_110_38-retracted.)
51
52		1. Fixed a problem with incorrect rounding modes in real64.sml.
53		(Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
54
55		2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
56		and enums fixed.  The naming of corresponding ML identifiers should
57		now be consistent again.
58
59		----------------------------------------------------------------------
60		Name: Allen Leung
61		Date: 2002/01/27
62		Tag: leunga-20020127-nowhere
63		Description:
64
65		Added a target called nowhere in the configuration scripts.
66		Enabling this will build the MLRISC 'nowhere' tool (for translating
67		programs with where-clauses into legal SML code) during installation.
68
69		----------------------------------------------------------------------
70		Name: Matthias Blume
71		Date: 2002/01/25 21:27:00 EST
72		Tag: blume-Release_110_38-retracted
73		Description:
74
75		Call it a (working) release!  Version is 110.38. Bootfiles are ready.
76
77		README will be added later.
78
79		!!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
80		(Release_110_38) removed.  Reason: Last-minute bug fixes.
81
82		----------------------------------------------------------------------
83		Name: Matthias Blume
84		Date: 2002/01/25
85		Tag: blume-20020125-ffi
86		Description:
87
88		A large number of tweaks and improvements to ml-nlffi-lib and
89		ml-nlffigen:
90
91		- ML represenation types have been streamlined
92		- getter and setter functions work with concrete values, not abstract
93		ones where possible
94		- ml-nlffigen command line more flexible (see README file there)
95		- some bugs have been fixed (hopefully)
96
97		----------------------------------------------------------------------
98		Name: Lal George
99		Date: 2002/01/24
100		Tag: george-20020124-risc-ra-interface
101		Description:
102
103		There is a dramatic simplification in the interface to the
104		register allocator for RISC architectures as a result of making
105		parallel copy instructions explicit.
106
107		----------------------------------------------------------------------
108		Name: Matthias Blume
109		Date: 2002/01/22
110		Tag: blume-20020122-x86-ccalls
111		Description:
112
113		Bug fix for c-calls on x86 (having to do with how char- and
114		short-arguments are being handled).
115
116		----------------------------------------------------------------------
117		Name: Matthias Blume
118		Date: 2002/01/21
119		Tag: blume-20020121-ff
120		Description:
121
122		Another day of fiddling with the FFI...
123
124		1. Bug fix/workaround:  CKIT does not complain about negative array
125		dimensions, so ml-nlffigen has to guard itself against this possibility.
126		(Otherwise a negative dimension would send it into an infinite loop.)
127
128		2. Some of the abstract types (light objects, light pointers, most "base"
129		types) in structure C are now eqtypes.
130
131		3. Added constructors and test functions for NULL function pointers.
132
133		----------------------------------------------------------------------
134		Name: Matthias Blume
135		Date: 2002/01/18
136		Tag: blume-20020118-ready-for-new-release
137		Description:
138
139		Made config/srcarchiveurl point to a new place.  (Will provide boot
140		files shortly.)
141
142		Maybe we christen this to be 110.38?
143
144		----------------------------------------------------------------------
145		Name: Matthias Blume
146		Date: 2002/01/18
147		Tag: blume-20020118-more-ffifiddle
148		Description:
149
150		Today's FFI fiddling:
151
152		- Provided a structure CGetSet with "convenient" versions of C.Get.* and
153		C.Set.* that use concrete (MLRep.*) arguments and results instead
154		of abstract ones.
155
156		- Provided word-style bit operations etc. for "int" representation
157		types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
158		and Long.
159
160		----------------------------------------------------------------------
161		Name: Matthias Blume
162		Date: 2002/01/18
163		Tag: blume-20020118-use-x86-fp
164		Description:
165
166		Now that x86-fast-fp seems to be working, I turned it back on again
167		by default.  (Seems to work fine now, even with the FFI.)
168
169		Other than that, I added some documentation about the FFI to
170		src/ml-nlffigen/README and updated the FFI test examples in
171		src/ml-nlffi-lib/Tests/*.
172
173		----------------------------------------------------------------------
174		Name: Allen Leung
175		Date: 2002/01/17
176		Tag: leunga-20020117-x86-fast-fp-call
177		Description:
178
179		1. Fixed a problem with handling return fp values when x86's fast fp
180		mode is turned on.
181
182		2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
183		of %f32.
184
185		3. Added a constructor INT32lit to the ast of MLRISC tools.
186
187		----------------------------------------------------------------------
188		Name: Matthias Blume
189		Date: 2002/01/16
190		Tag: blume-20020116-ffifiddle
191		Description:
192
193		More fiddling with the FFI interface:
194
195		- Make constness 'c instead of rw wherever possible.  This eliminates
196		the need for certain explicit coercions.  (However, due to ML's
197		value polymorphism, there will still be many cases where explicit
198		coercions are necessary.  Phantom types are not the whole answer
199		to modeling a subtyping relationship in ML.)
200
201		- ro/rw coersions for pointers added.  (Avoids the detour through */&.)
202
203		- "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
204		clumsy workaround for varargs problem.)
205
206		----------------------------------------------------------------------
207		Name: Lal George
208		Date: 2002/01/15
209		Tag: <none>
210		Description:
211
212		1. Since COPY instructions are no longer native to the architecture,
213		a generic functor can be used to implement the expandCopies function.
214
215		2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
216		TEXT segment.
217
218		----------------------------------------------------------------------
219		Name: Matthias Blume
220		Date: 2002/01/15
221		Tag: blume-20020115-ffiupdates
222		Description:
223
224		1. Fix for bug resulting in single-precision float values being returned
225		incorrectly from FFI calls.
226
227		2. Small modifications to C FFI API:
228
229		- memory-allocation routines return straight objects (no options)
230		and raise an exception in out-of-memory situations
231		- unsafe extensions to cast between function pointers and pointers
232		from/to ints
233		- added structure C_Debug as an alternative to structure C where
234		pointer-dereferencing (|*| and |*!) always check for null-pointers
235		- added open_lib' to DynLinkage;  open_lib' works like open_lib
236		but also takes a (possibly empty) list of existing library handles
237		that the current library depends on
238
239		----------------------------------------------------------------------
240		Name: Matthias Blume
241		Date: 2002/01/10
242		Tag: blume-20020110-newffigen
243		Description:
244
245		1. Updates to portable graph code.
246
247		2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
248		more scalable now so that even huge interfaces such as the one
249		for GTK compile in finite time and space. :-)
250		See src/ml-nlffigen/README for details on what's new.
251
252		----------------------------------------------------------------------
253		Name: Lal George
254		Date: 2001/01/09 14:31:35 EST 2002
255		Tag: george-20011206-rm-native-copy
256		Description:
257
258		Removed the native COPY and FCOPY instructions
259		from all the architectures and replaced it with the
260		explicit COPY instruction from the previous commit.
261
262		It is now possible to simplify many of the optimizations
263		modules that manipulate copies. This has not been
264		done in this change.
265
266		----------------------------------------------------------------------
267		Name: Lal George
268		Date: 2001/12/06 16:50:13 EST 2001
269		Tag: george-20011206-mlrisc-instruction
270		Description:
271
272		Changed the representation of instructions from being fully abstract
273		to being partially concrete. That is to say:
274
275		from
276		type instruction
277
278		to
279		type instr                              (* machine instruction *)
280
281		datatype instruction =
282		LIVE of {regs: C.cellset, spilled: C.cellset}
283		| KILL of {regs: C.cellset, spilled: C.cellset}
284		| COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
285		| ANNOTATION of {i: instruction, a: Annotations.annotation}
286		| INSTR of instr
287
288		This makes the handling of certain special instructions that appear on
289		all architectures easier and uniform.
290
291		LIVE and KILL say that a list of registers are live or killed at the
292		program point where they appear. No spill code is generated when an
293		element of the 'regs' field is spilled, but the register is moved to
294		the 'spilled' (which is present, more for debugging than anything else).
295
296		LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
297		We used to generate:
298
299		DEFFREG f1
300		f1 := f2 + f3
301		trapb
302
303		but now generate:
304
305		f1 := f2 + f3
306		trapb
307		LIVE {regs=[f1,f2,f3], spilled=[]}
308
309		Furthermore, the DEFFREG (hack) required that all floating point instruction
310		use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
311		defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
312		in a cleaner alpha implementation. (Hopefully, intel will not get rid of
313		this architecture).
314
315		COPYXXX is intended to replace the parallel COPY and FCOPY  available on
316		all the architectures. This will result in further simplification of the
317		register allocator that must be aware of them for coalescing purposes, and
318		will also simplify certain aspects of the machine description that provides
319		callbacks related to parallel copies.
320
321		ANNOTATION should be obvious, and now INSTR represents the honest to God
322		machine instruction set!
323
324		The <arch>/instructions/<arch>Instr.sml files define certain utility
325		functions for making porting easier -- essentially converting upper case
326		to lower case. All machine instructions (of type instr) are in upper case,
327		and the lower case form generates an MLRISC instruction. For example on
328		the alpha we have:
329
330		datatype instr =
331		LDA of {r:cell, b:cell, d:operand}
332		| ...
333
334		val lda : {r:cell, b:cell, d:operand} -> instruction
335		...
336
337		where lda is just (INSTR o LDA), etc.
338
339		----------------------------------------------------------------------
340		Name: Matthias Blume
341		Date: 2001/11/22 21:40:00 EST
342		Tag: Release_110_37
343		Description:
344
345		Release 110.37.  This time for real.
346
347		----------------------------------------------------------------------
348		Name: Matthias Blume
349		Date: 2001/11/21 16:35:00 EST
350		Tag: blume-20011121-foot-in-mouth
351		Description:
352
353		Removed the "Release_110_37" tag because of a serious bug.
354		This will be re-tagged once the bug is fixed.
355
356		----------------------------------------------------------------------
357		Name: Matthias Blume
358		Date: 2001/11/21 16:14:00 EST
359		Tag: blume-20011121-forgottenfile
360		Description:
361
362		Forgot to add a file.  (Just a .tex-file -- part of
363		the CM manual source.)
364
365		----------------------------------------------------------------------
366		Name: Matthias Blume
367		Date: 2001/11/21 16:10:00 EST
368		Tag: blume-20011121-invalid_110_37
369		Description:
370
371		Note: I removed the original tag "Release_110_37" from this commit
372		because we found a serious bug in all non-x86 backends.
373		- Matthias
374
375		1. Modifications to the SML/NJ code generator and to the runtime system
376		so that code object name strings are directly inserted into code
377		objects at code generation time.  The only business the runtime system
378		has with this is now to read the name strings on occasions.
379		(The encoding of the name string has also changed somewhat.)
380
381		2. CM now implements a simple "set calculus" for specifying export lists.
382		In particular, it is now possible to refer to the export lists of
383		other libraries/groups/sources and form unions as well as differences.
384		See the latest CM manual for details.
385
386		3. An separate notion of "proxy" libraries has again be eliminated from
387		CM's model.  (Proxy libraries are now simply a special case of using
388		the export list calculus.)
389
390		4. Some of the existing libraries now take advantage of the new set
391		calculus.
392		(Notice that not all libraries have been converted because some
393		of the existing .cm-files are supposed to be backward compatible
394		with 110.0.x.)
395
396		5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
397		"General.exnMessage"!  The former relies on a certain hook to be
398		initialized, and that often does not happen in the stand-alone case.)
399
400		----------------------------------------------------------------------
401		Name: Lal George
402		Date: 2001/11/21  13:56:18 EST
403		Tag: george-2001121-pseudo-ops
404		Description:
405
406		Implemented a complete redesign of MLRISC pseudo-ops. Now there
407		ought to never be any question of incompatabilities with
408		pseudo-op syntax expected by host assemblers.
409
410		For now, only modules supporting GAS syntax are implemented
411		but more should follow, such as MASM, and vendor assembler
412		syntax, e.g. IBM as, Sun as, etc.
413
414		----------------------------------------------------------------------
415		Name: Matthias Blume
416		Date: 2001/11/14 11:52:00 EST
417		Tag: blume-20011114-srcname
418		Description:
419
420		1. Routed the name of the current source file to mlriscgen where it
421		should be directly emitted into the code object.  (This last part
422		is yet to be done.)
423
424		2. Some cleanup of the pgraph code to make it match the proposal that
425		I put out the other day.  (The proposal notwithstanding, things are
426		still in flux here.)
427
428		----------------------------------------------------------------------
429		Name: Lal George
430		Date: 2001/11/14 09:44:04 EST
431		Tag:
432		Description:
433
434		Fix for a backpatching bug reported by Allen.
435
436		Because the boundary between short and long span-dependent
437		instructions is +/- 128, there are an astounding number of
438		span-dependent instructions whose size is over estimated.
439
440		Allen came up with the idea of letting the size of span
441		dependent instructions be non-monotonic, for a maxIter
442		number of times, after which the size must be monotonically
443		increasing.
444
445		This table shows the number of span-dependent instructions
446		whose size was over-estimated as a function of maxIter, for the
447		file Parse/parse/ml.grm.sml:
448
449		maxIter            # of instructions:
450		10                      687
451		20                      438
452		30                      198
453		40                        0
454
455		In compiling the compiler, there is no significant difference in
456		compilation speed between maxIter=10 and maxIter=40. Actually,
457		my measurements showed that maxIter=40 was a tad faster than
458		maxIter=10! Also 96% of the  files in the compiler reach a fix
459		point within 13 iterations, so fixing maxIter at 40, while high,
460		is okay.
461
462		----------------------------------------------------------------------
463		Name: Matthias Blume
464		Date: 2001/10/31 15:25:00 EST
465		Tag: blume-20011031-pgraph
466		Description:
467
468		CKIT:
469		* Changed the "Function" constructor of type Ast.ctype to carry optional
470		argument identifiers.
471		* Changed the return type of TypeUtil.getFunction accordingly.
472		* Type equality ignores the argument names.
473		* TypeUtil.composite tries to preserve argument names but gives up quickly
474		if there is a mismatch.
475
476		installation script:
477		* attempts to use "curl" if available (unless "wget" is available as well)
478
479		CM:
480		* has an experimental implementation of "portable graphs" which I will
481		soon propose as an implementation-independent library format
482		* there are also new libraries $/pgraph.cm and $/pgraph-util.cm
483
484		NLFFI-LIB:
485		* some cleanup (all cosmetic)
486
487		NLFFIGEN:
488		* temporarily disabled the mechanism that suppresses ML output for
489		C definitions whose identifiers start with an underscore character
490		* generate val bindings for enum constants
491		* user can request that only one style (light or heavy) is being used;
492		default is to use both (command-line arguments: -heavy and -light)
493		* fixed bug in handling of function types involving incomplete pointers
494		* generate ML entry points that take record arguments (i.e., using
495		named arguments) for C functions that have a prototype with named
496		arguments
497		(see changes to CKIT)
498
499		----------------------------------------------------------------------
500		Name: Allen Leung
501		Date: 2001/10/27 20:34:00 EDT
502		Tag: leunga-20011027-x86-fast-fp-call
503		Description:
504
505		Fixed the bug described in blume-20010920-slowfp.
506
507		The fix involves
508		1. generating FCOPYs in FSTP in ia32-svid
509		2. marking a CALL with the appropriate annotation
510
511		----------------------------------------------------------------------
512		Name: Matthias Blume
513		Date: 2001/10/16 11:32:00 EDT
514		Tag: blume-20011016-netbsd
515		Description:
516
517		Underscore patch from Chris Richards (fixing problem with compiling
518		runtime system under recent NetBSD).
519
520		----------------------------------------------------------------------
521		Name: Allen Leung
522		Date: 2001/10/12 17:18:32 EDT 2001
523		Tag: leung-20011012-x86-printflowgraph
524		Description:
525
526		X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
527
528		----------------------------------------------------------------------
529		Name: Lal George
530		Date: 2001/10/11 23:51:34 EDT
531		Tag: george-20011011-too-many-instrs
532		Description:
533
534		The representation of a program point never expected to see more
535		than 65536 instructions in a basic block!
536
537		----------------------------------------------------------------------
538		Name: Lal George
539		Date: 2001/10/09 09:41:37 EDT
540		Tag: george-20011008-mlrisc-labels
541		Description:
542
543		Changed the machine description files to support printing of
544		local and global labels in assembly code, based on host assembler
545		conventions.
546
547		----------------------------------------------------------------------
548		Name: Matthias Blume
549		Date: 2001/09/25 15:25:00 EDT
550		Tag: blume-20010925-exninfo
551		Description:
552
553		I provided a non-hook implementation of exnName (at the toplevel) and
554		made the "dummy" implementation of exnMessage (at the toplevel) more
555		useful: if nothing gets "hooked in", then at least you are going to
556		see the exception name and a message indicating why you don't see more.
557
558		[For the time being, programs that need exnMessage and want to use
559		ml-build should either use General.exnMessage (strongly recommended) or
560		refer to structure General at some other point so that CM sees a
561		static dependency.]
562
563		[Similar remarks go for "print" and "use":  If you want to use their
564		functionality in stand-alone programs generated by ml-build, then use
565		TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
566
567		----------------------------------------------------------------------
568		Name: Matthias Blume
569		Date: 2001/09/20 17:28:00 EDT
570		Tag: blume-20010920-slowfp
571		Description:
572
573		Allen says that x86-fast-fp is not safe yet, so I turned it off again...
574
575		----------------------------------------------------------------------
576		Name: Matthias Blume
577		Date: 2001/09/20 17:20:00 EDT
578		Tag: blume-20010920-canonicalpaths
579		Description:
580
581		0. Updated the BOOT file (something that I forgot to do earlier).
582
583		1. Small internal change to CM so that it avoids "/../" in filenames
584		as much as possible (but only where it is safe).
585
586		2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
587		that arguments that contain delimiters are passed through correctly.
588		This change also means that all "special" arguments of the form
589		@SMLxxx... must come first.
590
591		3. Changed install script to put relative anchor names for tool commands
592		into pathconfig.
593
594		----------------------------------------------------------------------
595		Name: Matthias Blume
596		>>>>>>> 1.169
597		Date: 2001/09/18 15:35:00 EDT
598		Tag: blume-20010918-readme11036
599		Description:
600
601		Added README files.
602
603		----------------------------------------------------------------------
604		Name: Matthias Blume
605		Date: 2001/09/18 11:45:00 EDT
606		Tag: Release_110_36 (retag)
607		Description:
608
609		Fixed mistake in config/preloads. Retagged as 110.36.
610
611		----------------------------------------------------------------------
612		Name: Matthias Blume
613		Date: 2001/09/18 09:40:00 EDT
614		Tag: Release_110_36_orig (tag changed)
615		Description:
616
617		New version (110.36).  New bootfiles.
618
619		----------------------------------------------------------------------
620		Name: Matthias Blume
621		Date: 2001/09/14 16:15:00 EDT
622		Tag: blume-20010914-x86fastfp
623		Description:
624
625		John committed some changes that Allen made, in particular a (hopefully)
626		correctly working version of the x86-fp module.
627
628		I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
629		flag to "true".  Everything seems to compile to a fixpoint ok, and
630		"mandelbrot" speeds up by about 15%.
631
632		----------------------------------------------------------------------
633		Name: Matthias Blume
634		Date: 2001/09/13 11:20:00 EDT
635		Tag: blume-20010913-minimal
636		Description:
637
638		1. Stefan Monnier's patch to fix a miscompilation problem that
639		was brought to light by John Reppy's work on Moby.
640
641		2. Implemented a minimal "structure Compiler" that contains just
642		"version" and "architecture".  The minimal version will be
643		available when the full version is not.  This is for backward-
644		compatibility with code that wants to test Compiler.version.
645
646		----------------------------------------------------------------------
647		Name: Matthias Blume
648		Date: 2001/08/28 14:03:00 EDT
649		Tag: blume-20010828-ml-lex
650		Description:
651
652		Fix for bug 1581, received from Neophytos Michael.
653
654		----------------------------------------------------------------------
655		Name: Matthias Blume
656		Date: 2001/08/27 11:20:00 EDT
657		Tag: blume-20010827-readme11035
658		Description:
659
660		Fleshed out the README file for 110.35.
661
662		----------------------------------------------------------------------
663		Name: Matthias Blume
664		Date: 2001/08/24 17:10:00 EDT
665		Tag: Release_110_35
666		Description:
667
668		New version number (110.35).  New bootfiles.
669
670		----------------------------------------------------------------------
671		Name: Lal George
672		Date: 2001/08/24 13:47:18 EDT 2001
673		Tag: george-20010824-MLRISC-graphs
674		Description:
675
676		removed clusters from MLRISC completely and replaced with graphs.
677
678		----------------------------------------------------------------------
679		Name: Matthias Blume
680		Date: 2001/08/23 17:50:00 EDT
681		Tag: blume-20010823-toplevel
682		Description:
683
684		- some reorganization of the code that implements various kinds of
685		environments in the compiler (static, dynamic, symbolic, combined)
686		- re-implemented the EnvRef module so that evalStream works properly
687		(if the stream contains references to "use", "CM.make", etc.)
688		- cleaned up evalloop.sml and interact.sml (but they need more cleaning)
689
690		----------------------------------------------------------------------
691		Name: Matthias Blume
692		Date: 2001/08/20 15:50 EDT
693		Tag: blume20010820-slipup
694		Description:
695
696		I forgot to commit a few files.  Here they are...
697
698		----------------------------------------------------------------------
699		Name: Matthias Blume
700		Date: 2001/08/20 15:35:00 EDT
701		Tag: blume-20010820-debugprof
702		Description:
703
704		!!!! NEW BOOTFILES !!!!
705
706		This is another round of reorganizing the compiler sources.  This
707		time the main goal was to factor out all the "instrumentation"
708		passes (for profiling and backtracing) into their own library.
709		The difficulty was to do it in such a way that it does not depend
710		on elaborate.cm but only on elabdata.cm.
711
712		Therefore there have been further changes to both elaborate.cm and
713		elabdata.cm -- more "generic" things have been moved from the former
714		to the latter.  As a result, I was forced to split the assignment
715		of numbers indicating "primtyc"s into two portions: SML-generic and
716		SML/NJ-specific.  Since it would have been awkward to maintain,
717		I bit the bullet and actually _changed_ the mapping between these
718		numbers and primtycs.  The bottom line of this is that you need
719		a new set of bin- and bootfiles.
720
721		I have built new bootfiles for all architectures, so doing a fresh
722		checkout and config/install.sh should be all you need.
723
724		The newly created library's name is
725
726		$smlnj/viscomp/debugprof.cm
727
728		and its sources live under
729
730		src/compiler/DebugProf
731
732		----------------------------------------------------------------------
733		Name: Matthias Blume
734		Date: 2001/08/15 17:15:00 EDT
735		Tag: blume-20010815-compreorg
736		Description:
737
738		This is a first cut at reorganizing the CM libraries that make up the
739		core of the compiler.  The idea is to separate out pieces that could
740		be used independently by tools, e.g., the parser, the typechecker, etc.
741
742		The current status is a step in this direction, but it is not quite
743		satisfactory yet.  Expect more changes in the future.
744
745		Here is the current (new) organization...
746
747		What used to be $smlnj/viscomp/core.cm is now divided into
748		six CM libraries:
749
750		$smlnj/viscomp/basics.cm
751		/parser.cm
752		/elabdata.cm
753		/elaborate.cm
754		/execute.cm
755		/core.cm
756
757		The CM files for these libraries live under src/system/smlnj/viscomp.
758		All these libraries are proxy libraries that contain precisely
759		one CM library component.  Here are the locations of the components
760		(all within the src/compiler tree):
761
762		Basics/basics.cm
763		Parse/parser.cm
764		ElabData/elabdata.cm
765		Elaborator/elaborate.cm
766		Execution/execute.cm
767		core.cm
768
769		[This organization is the same that has been used already
770		for a while for the architecture-specific parts of the visible
771		compiler and for the old version of core.cm.]
772
773		As you will notice, many source files have been moved from their
774		respective original locations to a new home in one of the above
775		subtrees.
776
777		The division of labor between the new libraries is the following:
778
779		basics.cm:
780		- Simple, basic definitions that pertain to many (or all) of
781		the other libraries.
782		parser.cm:
783		- The SML parser, producing output of type Ast.dec.
784		- The type family for Ast is also defined and exported here.
785		elabdata.cm:
786		- The datatypes that describe input and output of the elaborator.
787		This includes types, absyn, and static environments.
788		elaborator.cm:
789		- The SML/NJ type checker and elaborator.
790		This maps an Ast.dec (with a given static environment) to
791		an Absyn.dec (with a new static environment).
792		- This libraries implements certain modules that used to be
793		structures as functors (to remove dependencies on FLINT).
794		execute.cm:
795		- Everything having to do with executing binary code objects.
796		- Dynamic environments.
797		core.cm:
798		- SML/NJ-specific instantiations of the elaborator and MLRISC.
799		- Top-level modules.
800		- FLINT (this should eventually become its own library)
801
802		Notes:
803
804		I am not 100% happy with the way I separated the elaborator (and its
805		data structures) from FLINT.  Two instances of the same problem:
806
807		1. Data structures contain certain fields that carry FLINT-specific
808		information.  I hacked around this using exn and the property list
809		module from smlnj-lib.  But the fact that there are middle-end
810		specific fields around at all is a bit annoying.
811
812		2. The elaborator calculates certain FLINT-related information.  I tried
813		to make this as abstract as I could using functorization, but, again,
814		the fact that the elaborator has to perform calculations on behalf
815		of the middle-end at all is not nice.
816
817		3. Having to used exn and property lists is unfortunate because it
818		weakens type checking.  The other alternative (parameterizing
819		nearly *everything*) is not appealing, though.
820
821		I removed the "rebinding =" warning hack because due to the new organization
822		it was awkward to maintain it.  As a result, the compiler now issues some of
823		these warnings when compiling init.cmi during bootstrap compilation. On
824		the plus side, you also get a warning when you do, for example:
825		val op = = Int32.+
826		which was not the case up to now.
827
828		I placed "assign" and "deref" into the _Core structure so that the
829		code that deals with the "lazy" keyword can find them there.  This
830		removes the need for having access to the primitive environment
831		during elaboration.
832
833		----------------------------------------------------------------------
834		Name: Matthias Blume
835		Date: 2001/08/13
836		Tag: blume-20010813-closures
837		Description:
838
839		This fix was sent to us by Zhong Shao.  It is supposed to improve the
840		performance of certain loops by avoiding needless closure allocation.
841
842		----------------------------------------------------------------------
843		Name: Lal George
844		Date: 2001/07/31 10:03:23 EDT 2001
845		Tag: george-20010731-x86-fmalloc
846		Description: Fixed bug in x86 calls
847
848		There was a bug where call instructions would mysteriously
849		vanish. The call instruction had to be one that returned
850		a floating point value.
851
852		----------------------------------------------------------------------
853		Name: Lal George
854		Date: 2001/07/19 16:36:29 EDT 2001
855		Tag: george-20010719-simple-cells
856		Description:
857
858		I have dramatically simplified the interface for CELLS in MLRISC.
859
860		In summary, the cells interface is broken up into three parts:
861
862		1. CellsBasis : CELLS_BASIS
863
864		CellsBasis is a top level structure and common for all
865		architectures.  it contains the definitions of basic datatypes
866		and utility  functions over these types.
867
868		2. functor Cells() : CELLS
869
870		Cells generates an interface for CELLS that incorporates the
871		specific resources on the target architecture, such as the
872		presence of special register classes, their number and size,
873		and various useful substructures.
874
875		3. <ARCH>CELLS
876
877		e.g. SparcCells: SPARCCELLS
878
879		<ARCH>CELLS usually contains additional bindings for special
880		registers  on the architecture, such as:
881
882		val r0 : cell           (* register zero *)
883		val y : cell            (* Y register *)
884		val psr : cell          (* processor status register *)
885		...
886
887		The structure returned by applying the Cells functor is opened
888		in this interface.
889
890		The main implication of all this is that the datatypes for cells is
891		split between CellsBasis and CELLS -- a fairly simple change for user
892		code.
893
894		In the old scheme the CELLS interface had a definitional binding of
895		the form:
896
897		signature CELLS = sig
898
899		structure CellsBasis = CellsBasis
900
901		...
902
903		end
904
905		With all the sharing constraints that goes on in MLRISC, this old
906		design  quickly leads to errors such as:
907
908		"structure definition spec inside of sharing ... "
909
910
911		and appears to require an unacceptable amount of sharing and where
912		constraint hackery.
913
914		I think this error message (the interaction of definitional specs and
915		sharing) requires more explanation on our web page.
916
917		----------------------------------------------------------------------
918		Name: Matthias Blume
919		Date: 2001/07/19 15:00:00 EDT
920		Tag: blume-20010719-libreorg
921		Description:
922
923		This update puts together a fairly extensive but straightforward change
924		to the way the libraries that implement the interactive system are
925		organized:
926
927		The biggest change is the elimination of structure Compiler.  As a
928		replacement for this structure, there is now a CM library
929		(known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
930		that exports all the substructures of the original structure Compiler
931		directly.  So instead of saying Compiler.Foo.bar one now simply
932		says Foo.bar.  (The CM libraries actually export a collection of
933		structures that is richer than the collection of substructures of
934		structure Compiler.)
935
936		To make the transition smooth, there is a separate library called
937		$smlnj/compiler/compiler.cm which puts together and exports the
938		original structure Compiler (or at least something very close to it).
939
940		There are five members of the original structure Compiler
941		that are not exported directly but which instead became members
942		of a new structure Backend (described by signature BACKEND).  These are:
943		structure Profile (: PROFILE), structure Compile (: COMPILE), structure
944		Interact (: INTERACT), structure Machine (: MACHINE), and val
945		architecture (: string).
946
947		Structure Compiler.Version has become structure CompilerVersion.
948
949		Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
950		by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
951		or x86, respectively.
952		Each of these exports the same frontend structures that
953		$smlnj/compiler.cm exports.  But they do not have a structure Backend
954		and instead export some structure <Arch>Backend where <Arch> is Alpha32,
955		Hppa, PPC, Sparc, or X86, respectively.
956
957		Library $smlnj/compiler/all.cm exports the union of the exports of
958		$smlnj/compiler/<arch>.cm
959
960		There are no structures <Arch>Compiler anymore, use
961		$smlnj/compiler/<arch>.cm instead.
962
963		Library host-compiler-0.cm is gone.  Instead, the internal library
964		that instantiates CM is now called cm0.cm.  Selection of the host
965		compiler (backend) is no longer done here but. (Responsibility for it
966		now lies with $smlnj/compiler/current.cm.  This seems to be more
967		logical.)
968
969		Many individual files have been moved or renamed.  Some files have
970		been split into multiple files, and some "dead" files have been deleted.
971
972		Aside from these changes to library organization, there are also changes
973		to the way the code itself is organized:
974
975		Structure Binfile has been re-implemented in such a way that it no
976		longer needs any knowledge of the compiler.  It exclusively deals
977		with the details of binfile layout.  It no longer invokes the
978		compiler (for the purpose of creating new prospective binfile
979		content), and it no longer has any knowledge of how to interpret
980		pickles.
981
982		Structure Compile (: COMPILE) has been stripped down to the bare
983		essentials of compilation.  It no longer deals with linking/execution.
984		The interface has been cleaned up considerably.
985
986		Utility routines for dealing with linking and execution have been
987		moved into their own substructures.
988
989		(The ultimate goal of these changes is to provide a light-weight
990		binfile loader/linker (at least for, e.g., stable libraries) that
991		does not require CM or the compiler to be present.)
992
993		CM documentation has been updated to reflect the changes to library
994		organization.
995
996		----------------------------------------------------------------------
997		Name: Matthias Blume
998		Date: 2001/07/10 17:30:00 EDT
999		Tag: Release_110_34
1000		Description:
1001
1002		Minor tweak to 110.34 (re-tagged):
1003
1004		- README.html file added to CVS repository
1005		- runtime compiles properly under FreeBSD 3.X and 4.X
1006
1007		----------------------------------------------------------------------
1008		Name: Matthias Blume
1009		Date: 2001/07/10 17:30:00 EDT
1010		Tag: Release_110_34
1011		Description:
1012
1013		New version number (110.34). New bootfiles.
1014
1015		----------------------------------------------------------------------
1016		Name: Matthias Blume
1017		Date: 2001/07/09 16:00:00 EDT
1018		Tag: blume-20010709-more-varargs
1019		Description:
1020
1021		I changed the handling of varargs in ml-nlffigen again:
1022		The ellipsis ... will now simply be ignored (with an accompanying warning).
1023
1024		The immediate effect is that you can actually call a varargs function
1025		from ML -- but you can't actually supply any arguments beyond the ones
1026		specified explicitly.  (For example, you can call printf with its format
1027		string, but you cannot pass additional arguments.)
1028
1029		This behavior is only marginally more useful than the one before, but
1030		it has the advantage that a function or, more importantly, a function
1031		type never gets dropped on the floor, thus avoiding follow-up problems with
1032		other types that refer to the offending one.
1033
1034		----------------------------------------------------------------------
1035		Name: Matthias Blume
1036		Date: 2001/07/09 11:25:00 EDT
1037		Tag: blume-20010709-varargs
1038		Description:
1039
1040		1. ckit-lib.cm now exports structure Error
1041		2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
1042		with a warning accompanied by a source location.  Moreover, it
1043		merely skips the offending function or type and proceeds with the
1044		rest of its work.u  As a result, one can safely feed C code containing
1045		"..." to ml-nlffigen.
1046		3. There are some internal improvements to CM, providing slightly
1047		more general string substitutions in the tools subsystem.
1048
1049		----------------------------------------------------------------------
1050		Name: Matthias Blume
1051		Date: 2001/06/27 15:10:00 EDT
1052		Tag: blume-20010627-concur
1053		Description:
1054
1055		Fixed a small bug in CM's handling of parallel compilation.
1056		(You could observe the bug by Control-C-interrupting an ordinary
1057		CMB.make or CM.stabilize and then attaching some compile servers.
1058		The result was that all of a sudden the previously interrupted
1059		compilation would continue on its own.  This was because of
1060		an over-optimization: CM did not bother to clean out certain queues
1061		when no servers were attached "anyway", resulting in the contents
1062		of these queues to grab control when new servers did get attached.)
1063
1064		There is also another minor update to the CM manual.
1065
1066		----------------------------------------------------------------------
1067		Name: Matthias Blume
1068		Date: 2001/06/26 16:15:00 EDT
1069		Tag: blume-20010626-cmdoc
1070		Description:
1071
1072		Minor typo fixed in CM manual (syntax diagram for libraries).
1073
1074		----------------------------------------------------------------------
1075		Name: Matthias Blume
1076		Date: 2001/06/25 22:55:00 EDT
1077		Tag: blume-20010625-x86pc
1078		Description:
1079
1080		Fixed a nasty bug in the X86 assembly code that caused signal
1081		handlers to fail (crash) randomly.
1082
1083		----------------------------------------------------------------------
1084		Name: Matthias Blume
1085		Date: 2001/06/25 12:05:00 EDT
1086		Tag: blume-20010625-nlffigen
1087		Description:
1088
1089		This update fixes a number of minor bugs in ml-nlffigen as reported by
1090		Nick Carter <nbc@andrew.cmu.edu>.
1091
1092		1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
1093		2. Default names for generated files are now derived from the name of
1094		the C file *without its directory*.  In particular, this causes generated
1095		files to be placed locally even if the C file is in some system directory.
1096		3. Default names for generated signatures and structures are also derived
1097		from the C file name without its directory.  This avoids silly things
1098		like "structure GL/GL".
1099		(Other silly names are still possible because ml-nlffigen does not do
1100		a thorough check of whether generated names are legal ML identifiers.
1101		When in doubt, use command line arguments to force particular names.)
1102
1103		----------------------------------------------------------------------
1104		Name: Matthias Blume
1105		Date: 2001/06/21 12:25:00 EDT
1106		Tag: blume-20010621-eXene
1107		Description:
1108
1109		eXene now compiles and (sort of) works again.
1110
1111		The library name (for version > 110.33) is $/eXene.cm.
1112
1113		I also added an new example in src/eXene/examples/nbody.  See the
1114		README file there for details.
1115
1116		----------------------------------------------------------------------
1117		Name: Matthias Blume
1118	Date: 2001/06/20 16:40:00 EDT	Date: 2001/06/20 16:40:00 EDT
1119	Tag: blume-20010620-cml	Tag: blume-20010620-cml
1120	Description:	Description:
#	Line 2469	Line 3570
3570	elaborator). There were a lot of changes during my "linkpath" trials	elaborator). There were a lot of changes during my "linkpath" trials
3571	that could have been reverted to their original state but weren't.	that could have been reverted to their original state but weren't.
3572	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
3573	than what was strictly necessary...  (I _did_ resist the tempation	than what was strictly necessary...  (I _did_ resist the temptation
3574	of doing any "global reformatting" to avoid an untimely death at	of doing any "global reformatting" to avoid an untimely death at
3575	Dave's hands. :)	Dave's hands. :)
3576

---

Colored Diff Long Colored Diff Unidiff Context Diff Side by Side

Legend: Removed from v.844   changed lines   Added in v.1045

---

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