Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 847, Mon Jun 25 16:11:22 2001 UTC	revision 1034, Thu Jan 24 05:53:53 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: Lal George
17		Date: 2002/01/24
18		Tag: george-20020124-risc-ra-interface
19		Description:
20
21		There is a dramatic simplification in the interface to the
22		register allocator for RISC architectures as a result of making
23		parallel copy instructions explicit.
24
25		----------------------------------------------------------------------
26		Name: Matthias Blume
27		Date: 2002/01/22
28		Tag: blume-20020122-x86-ccalls
29		Description:
30
31		Bug fix for c-calls on x86 (having to do with how char- and
32		short-arguments are being handled).
33
34		----------------------------------------------------------------------
35		Name: Matthias Blume
36		Date: 2002/01/21
37		Tag: blume-20020121-ff
38		Description:
39
40		Another day of fiddling with the FFI...
41
42		1. Bug fix/workaround:  CKIT does not complain about negative array
43		dimensions, so ml-nlffigen has to guard itself against this possibility.
44		(Otherwise a negative dimension would send it into an infinite loop.)
45
46		2. Some of the abstract types (light objects, light pointers, most "base"
47		types) in structure C are now eqtypes.
48
49		3. Added constructors and test functions for NULL function pointers.
50
51		----------------------------------------------------------------------
52		Name: Matthias Blume
53		Date: 2002/01/18
54		Tag: blume-20020118-ready-for-new-release
55		Description:
56
57		Made config/srcarchiveurl point to a new place.  (Will provide boot
58		files shortly.)
59
60		Maybe we christen this to be 110.38?
61
62		----------------------------------------------------------------------
63		Name: Matthias Blume
64		Date: 2002/01/18
65		Tag: blume-20020118-more-ffifiddle
66		Description:
67
68		Today's FFI fiddling:
69
70		- Provided a structure CGetSet with "convenient" versions of C.Get.* and
71		C.Set.* that use concrete (MLRep.*) arguments and results instead
72		of abstract ones.
73
74		- Provided word-style bit operations etc. for "int" representation
75		types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
76		and Long.
77
78		----------------------------------------------------------------------
79		Name: Matthias Blume
80		Date: 2002/01/18
81		Tag: blume-20020118-use-x86-fp
82		Description:
83
84		Now that x86-fast-fp seems to be working, I turned it back on again
85		by default.  (Seems to work fine now, even with the FFI.)
86
87		Other than that, I added some documentation about the FFI to
88		src/ml-nlffigen/README and updated the FFI test examples in
89		src/ml-nlffi-lib/Tests/*.
90
91		----------------------------------------------------------------------
92		Name: Allen Leung
93		Date: 2002/01/17
94		Tag: leunga-20020117-x86-fast-fp-call
95		Description:
96
97		1. Fixed a problem with handling return fp values when x86's fast fp
98		mode is turned on.
99
100		2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
101		of %f32.
102
103		3. Added a constructor INT32lit to the ast of MLRISC tools.
104
105		----------------------------------------------------------------------
106		Name: Matthias Blume
107		Date: 2002/01/16
108		Tag: blume-20020116-ffifiddle
109		Description:
110
111		More fiddling with the FFI interface:
112
113		- Make constness 'c instead of rw wherever possible.  This eliminates
114		the need for certain explicit coercions.  (However, due to ML's
115		value polymorphism, there will still be many cases where explicit
116		coercions are necessary.  Phantom types are not the whole answer
117		to modeling a subtyping relationship in ML.)
118
119		- ro/rw coersions for pointers added.  (Avoids the detour through */&.)
120
121		- "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
122		clumsy workaround for varargs problem.)
123
124		----------------------------------------------------------------------
125		Name: Lal George
126		Date: 2002/01/15
127		Tag: <none>
128		Description:
129
130		1. Since COPY instructions are no longer native to the architecture,
131		a generic functor can be used to implement the expandCopies function.
132
133		2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
134		TEXT segment.
135
136		----------------------------------------------------------------------
137		Name: Matthias Blume
138		Date: 2002/01/15
139		Tag: blume-20020115-ffiupdates
140		Description:
141
142		1. Fix for bug resulting in single-precision float values being returned
143		incorrectly from FFI calls.
144
145		2. Small modifications to C FFI API:
146
147		- memory-allocation routines return straight objects (no options)
148		and raise an exception in out-of-memory situations
149		- unsafe extensions to cast between function pointers and pointers
150		from/to ints
151		- added structure C_Debug as an alternative to structure C where
152		pointer-dereferencing (|*| and |*!) always check for null-pointers
153		- added open_lib' to DynLinkage;  open_lib' works like open_lib
154		but also takes a (possibly empty) list of existing library handles
155		that the current library depends on
156
157		----------------------------------------------------------------------
158		Name: Matthias Blume
159		Date: 2002/01/10
160		Tag: blume-20020110-newffigen
161		Description:
162
163		1. Updates to portable graph code.
164
165		2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
166		more scalable now so that even huge interfaces such as the one
167		for GTK compile in finite time and space. :-)
168		See src/ml-nlffigen/README for details on what's new.
169
170		----------------------------------------------------------------------
171		Name: Lal George
172		Date: 2001/01/09 14:31:35 EST 2002
173		Tag: george-20011206-rm-native-copy
174		Description:
175
176		Removed the native COPY and FCOPY instructions
177		from all the architectures and replaced it with the
178		explicit COPY instruction from the previous commit.
179
180		It is now possible to simplify many of the optimizations
181		modules that manipulate copies. This has not been
182		done in this change.
183
184		----------------------------------------------------------------------
185		Name: Lal George
186		Date: 2001/12/06 16:50:13 EST 2001
187		Tag: george-20011206-mlrisc-instruction
188		Description:
189
190		Changed the representation of instructions from being fully abstract
191		to being partially concrete. That is to say:
192
193		from
194		type instruction
195
196		to
197		type instr                              (* machine instruction *)
198
199		datatype instruction =
200		LIVE of {regs: C.cellset, spilled: C.cellset}
201		| KILL of {regs: C.cellset, spilled: C.cellset}
202		| COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
203		| ANNOTATION of {i: instruction, a: Annotations.annotation}
204		| INSTR of instr
205
206		This makes the handling of certain special instructions that appear on
207		all architectures easier and uniform.
208
209		LIVE and KILL say that a list of registers are live or killed at the
210		program point where they appear. No spill code is generated when an
211		element of the 'regs' field is spilled, but the register is moved to
212		the 'spilled' (which is present, more for debugging than anything else).
213
214		LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
215		We used to generate:
216
217		DEFFREG f1
218		f1 := f2 + f3
219		trapb
220
221		but now generate:
222
223		f1 := f2 + f3
224		trapb
225		LIVE {regs=[f1,f2,f3], spilled=[]}
226
227		Furthermore, the DEFFREG (hack) required that all floating point instruction
228		use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
229		defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
230		in a cleaner alpha implementation. (Hopefully, intel will not get rid of
231		this architecture).
232
233		COPYXXX is intended to replace the parallel COPY and FCOPY  available on
234		all the architectures. This will result in further simplification of the
235		register allocator that must be aware of them for coalescing purposes, and
236		will also simplify certain aspects of the machine description that provides
237		callbacks related to parallel copies.
238
239		ANNOTATION should be obvious, and now INSTR represents the honest to God
240		machine instruction set!
241
242		The <arch>/instructions/<arch>Instr.sml files define certain utility
243		functions for making porting easier -- essentially converting upper case
244		to lower case. All machine instructions (of type instr) are in upper case,
245		and the lower case form generates an MLRISC instruction. For example on
246		the alpha we have:
247
248		datatype instr =
249		LDA of {r:cell, b:cell, d:operand}
250		| ...
251
252		val lda : {r:cell, b:cell, d:operand} -> instruction
253		...
254
255		where lda is just (INSTR o LDA), etc.
256
257		----------------------------------------------------------------------
258		Name: Matthias Blume
259		Date: 2001/11/22 21:40:00 EST
260		Tag: Release_110_37
261		Description:
262
263		Release 110.37.  This time for real.
264
265		----------------------------------------------------------------------
266		Name: Matthias Blume
267		Date: 2001/11/21 16:35:00 EST
268		Tag: blume-20011121-foot-in-mouth
269		Description:
270
271		Removed the "Release_110_37" tag because of a serious bug.
272		This will be re-tagged once the bug is fixed.
273
274		----------------------------------------------------------------------
275		Name: Matthias Blume
276		Date: 2001/11/21 16:14:00 EST
277		Tag: blume-20011121-forgottenfile
278		Description:
279
280		Forgot to add a file.  (Just a .tex-file -- part of
281		the CM manual source.)
282
283		----------------------------------------------------------------------
284		Name: Matthias Blume
285		Date: 2001/11/21 16:10:00 EST
286		Tag: blume-20011121-invalid_110_37
287		Description:
288
289		Note: I removed the original tag "Release_110_37" from this commit
290		because we found a serious bug in all non-x86 backends.
291		- Matthias
292
293		1. Modifications to the SML/NJ code generator and to the runtime system
294		so that code object name strings are directly inserted into code
295		objects at code generation time.  The only business the runtime system
296		has with this is now to read the name strings on occasions.
297		(The encoding of the name string has also changed somewhat.)
298
299		2. CM now implements a simple "set calculus" for specifying export lists.
300		In particular, it is now possible to refer to the export lists of
301		other libraries/groups/sources and form unions as well as differences.
302		See the latest CM manual for details.
303
304		3. An separate notion of "proxy" libraries has again be eliminated from
305		CM's model.  (Proxy libraries are now simply a special case of using
306		the export list calculus.)
307
308		4. Some of the existing libraries now take advantage of the new set
309		calculus.
310		(Notice that not all libraries have been converted because some
311		of the existing .cm-files are supposed to be backward compatible
312		with 110.0.x.)
313
314		5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
315		"General.exnMessage"!  The former relies on a certain hook to be
316		initialized, and that often does not happen in the stand-alone case.)
317
318		----------------------------------------------------------------------
319		Name: Lal George
320		Date: 2001/11/21  13:56:18 EST
321		Tag: george-2001121-pseudo-ops
322		Description:
323
324		Implemented a complete redesign of MLRISC pseudo-ops. Now there
325		ought to never be any question of incompatabilities with
326		pseudo-op syntax expected by host assemblers.
327
328		For now, only modules supporting GAS syntax are implemented
329		but more should follow, such as MASM, and vendor assembler
330		syntax, e.g. IBM as, Sun as, etc.
331
332		----------------------------------------------------------------------
333		Name: Matthias Blume
334		Date: 2001/11/14 11:52:00 EST
335		Tag: blume-20011114-srcname
336		Description:
337
338		1. Routed the name of the current source file to mlriscgen where it
339		should be directly emitted into the code object.  (This last part
340		is yet to be done.)
341
342		2. Some cleanup of the pgraph code to make it match the proposal that
343		I put out the other day.  (The proposal notwithstanding, things are
344		still in flux here.)
345
346		----------------------------------------------------------------------
347		Name: Lal George
348		Date: 2001/11/14 09:44:04 EST
349		Tag:
350		Description:
351
352		Fix for a backpatching bug reported by Allen.
353
354		Because the boundary between short and long span-dependent
355		instructions is +/- 128, there are an astounding number of
356		span-dependent instructions whose size is over estimated.
357
358		Allen came up with the idea of letting the size of span
359		dependent instructions be non-monotonic, for a maxIter
360		number of times, after which the size must be monotonically
361		increasing.
362
363		This table shows the number of span-dependent instructions
364		whose size was over-estimated as a function of maxIter, for the
365		file Parse/parse/ml.grm.sml:
366
367		maxIter            # of instructions:
368		10                      687
369		20                      438
370		30                      198
371		40                        0
372
373		In compiling the compiler, there is no significant difference in
374		compilation speed between maxIter=10 and maxIter=40. Actually,
375		my measurements showed that maxIter=40 was a tad faster than
376		maxIter=10! Also 96% of the  files in the compiler reach a fix
377		point within 13 iterations, so fixing maxIter at 40, while high,
378		is okay.
379
380		----------------------------------------------------------------------
381		Name: Matthias Blume
382		Date: 2001/10/31 15:25:00 EST
383		Tag: blume-20011031-pgraph
384		Description:
385
386		CKIT:
387		* Changed the "Function" constructor of type Ast.ctype to carry optional
388		argument identifiers.
389		* Changed the return type of TypeUtil.getFunction accordingly.
390		* Type equality ignores the argument names.
391		* TypeUtil.composite tries to preserve argument names but gives up quickly
392		if there is a mismatch.
393
394		installation script:
395		* attempts to use "curl" if available (unless "wget" is available as well)
396
397		CM:
398		* has an experimental implementation of "portable graphs" which I will
399		soon propose as an implementation-independent library format
400		* there are also new libraries $/pgraph.cm and $/pgraph-util.cm
401
402		NLFFI-LIB:
403		* some cleanup (all cosmetic)
404
405		NLFFIGEN:
406		* temporarily disabled the mechanism that suppresses ML output for
407		C definitions whose identifiers start with an underscore character
408		* generate val bindings for enum constants
409		* user can request that only one style (light or heavy) is being used;
410		default is to use both (command-line arguments: -heavy and -light)
411		* fixed bug in handling of function types involving incomplete pointers
412		* generate ML entry points that take record arguments (i.e., using
413		named arguments) for C functions that have a prototype with named
414		arguments
415		(see changes to CKIT)
416
417		----------------------------------------------------------------------
418		Name: Allen Leung
419		Date: 2001/10/27 20:34:00 EDT
420		Tag: leunga-20011027-x86-fast-fp-call
421		Description:
422
423		Fixed the bug described in blume-20010920-slowfp.
424
425		The fix involves
426		1. generating FCOPYs in FSTP in ia32-svid
427		2. marking a CALL with the appropriate annotation
428
429		----------------------------------------------------------------------
430		Name: Matthias Blume
431		Date: 2001/10/16 11:32:00 EDT
432		Tag: blume-20011016-netbsd
433		Description:
434
435		Underscore patch from Chris Richards (fixing problem with compiling
436		runtime system under recent NetBSD).
437
438		----------------------------------------------------------------------
439		Name: Allen Leung
440		Date: 2001/10/12 17:18:32 EDT 2001
441		Tag: leung-20011012-x86-printflowgraph
442		Description:
443
444		X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
445
446		----------------------------------------------------------------------
447		Name: Lal George
448		Date: 2001/10/11 23:51:34 EDT
449		Tag: george-20011011-too-many-instrs
450		Description:
451
452		The representation of a program point never expected to see more
453		than 65536 instructions in a basic block!
454
455		----------------------------------------------------------------------
456		Name: Lal George
457		Date: 2001/10/09 09:41:37 EDT
458		Tag: george-20011008-mlrisc-labels
459		Description:
460
461		Changed the machine description files to support printing of
462		local and global labels in assembly code, based on host assembler
463		conventions.
464
465		----------------------------------------------------------------------
466		Name: Matthias Blume
467		Date: 2001/09/25 15:25:00 EDT
468		Tag: blume-20010925-exninfo
469		Description:
470
471		I provided a non-hook implementation of exnName (at the toplevel) and
472		made the "dummy" implementation of exnMessage (at the toplevel) more
473		useful: if nothing gets "hooked in", then at least you are going to
474		see the exception name and a message indicating why you don't see more.
475
476		[For the time being, programs that need exnMessage and want to use
477		ml-build should either use General.exnMessage (strongly recommended) or
478		refer to structure General at some other point so that CM sees a
479		static dependency.]
480
481		[Similar remarks go for "print" and "use":  If you want to use their
482		functionality in stand-alone programs generated by ml-build, then use
483		TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
484
485		----------------------------------------------------------------------
486		Name: Matthias Blume
487		Date: 2001/09/20 17:28:00 EDT
488		Tag: blume-20010920-slowfp
489		Description:
490
491		Allen says that x86-fast-fp is not safe yet, so I turned it off again...
492
493		----------------------------------------------------------------------
494		Name: Matthias Blume
495		Date: 2001/09/20 17:20:00 EDT
496		Tag: blume-20010920-canonicalpaths
497		Description:
498
499		0. Updated the BOOT file (something that I forgot to do earlier).
500
501		1. Small internal change to CM so that it avoids "/../" in filenames
502		as much as possible (but only where it is safe).
503
504		2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
505		that arguments that contain delimiters are passed through correctly.
506		This change also means that all "special" arguments of the form
507		@SMLxxx... must come first.
508
509		3. Changed install script to put relative anchor names for tool commands
510		into pathconfig.
511
512		----------------------------------------------------------------------
513		Name: Matthias Blume
514		>>>>>>> 1.169
515		Date: 2001/09/18 15:35:00 EDT
516		Tag: blume-20010918-readme11036
517		Description:
518
519		Added README files.
520
521		----------------------------------------------------------------------
522		Name: Matthias Blume
523		Date: 2001/09/18 11:45:00 EDT
524		Tag: Release_110_36 (retag)
525		Description:
526
527		Fixed mistake in config/preloads. Retagged as 110.36.
528
529		----------------------------------------------------------------------
530		Name: Matthias Blume
531		Date: 2001/09/18 09:40:00 EDT
532		Tag: Release_110_36_orig (tag changed)
533		Description:
534
535		New version (110.36).  New bootfiles.
536
537		----------------------------------------------------------------------
538		Name: Matthias Blume
539		Date: 2001/09/14 16:15:00 EDT
540		Tag: blume-20010914-x86fastfp
541		Description:
542
543		John committed some changes that Allen made, in particular a (hopefully)
544		correctly working version of the x86-fp module.
545
546		I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
547		flag to "true".  Everything seems to compile to a fixpoint ok, and
548		"mandelbrot" speeds up by about 15%.
549
550		----------------------------------------------------------------------
551		Name: Matthias Blume
552		Date: 2001/09/13 11:20:00 EDT
553		Tag: blume-20010913-minimal
554		Description:
555
556		1. Stefan Monnier's patch to fix a miscompilation problem that
557		was brought to light by John Reppy's work on Moby.
558
559		2. Implemented a minimal "structure Compiler" that contains just
560		"version" and "architecture".  The minimal version will be
561		available when the full version is not.  This is for backward-
562		compatibility with code that wants to test Compiler.version.
563
564		----------------------------------------------------------------------
565		Name: Matthias Blume
566		Date: 2001/08/28 14:03:00 EDT
567		Tag: blume-20010828-ml-lex
568		Description:
569
570		Fix for bug 1581, received from Neophytos Michael.
571
572		----------------------------------------------------------------------
573		Name: Matthias Blume
574		Date: 2001/08/27 11:20:00 EDT
575		Tag: blume-20010827-readme11035
576		Description:
577
578		Fleshed out the README file for 110.35.
579
580		----------------------------------------------------------------------
581		Name: Matthias Blume
582		Date: 2001/08/24 17:10:00 EDT
583		Tag: Release_110_35
584		Description:
585
586		New version number (110.35).  New bootfiles.
587
588		----------------------------------------------------------------------
589		Name: Lal George
590		Date: 2001/08/24 13:47:18 EDT 2001
591		Tag: george-20010824-MLRISC-graphs
592		Description:
593
594		removed clusters from MLRISC completely and replaced with graphs.
595
596		----------------------------------------------------------------------
597		Name: Matthias Blume
598		Date: 2001/08/23 17:50:00 EDT
599		Tag: blume-20010823-toplevel
600		Description:
601
602		- some reorganization of the code that implements various kinds of
603		environments in the compiler (static, dynamic, symbolic, combined)
604		- re-implemented the EnvRef module so that evalStream works properly
605		(if the stream contains references to "use", "CM.make", etc.)
606		- cleaned up evalloop.sml and interact.sml (but they need more cleaning)
607
608		----------------------------------------------------------------------
609		Name: Matthias Blume
610		Date: 2001/08/20 15:50 EDT
611		Tag: blume20010820-slipup
612		Description:
613
614		I forgot to commit a few files.  Here they are...
615
616		----------------------------------------------------------------------
617		Name: Matthias Blume
618		Date: 2001/08/20 15:35:00 EDT
619		Tag: blume-20010820-debugprof
620		Description:
621
622		!!!! NEW BOOTFILES !!!!
623
624		This is another round of reorganizing the compiler sources.  This
625		time the main goal was to factor out all the "instrumentation"
626		passes (for profiling and backtracing) into their own library.
627		The difficulty was to do it in such a way that it does not depend
628		on elaborate.cm but only on elabdata.cm.
629
630		Therefore there have been further changes to both elaborate.cm and
631		elabdata.cm -- more "generic" things have been moved from the former
632		to the latter.  As a result, I was forced to split the assignment
633		of numbers indicating "primtyc"s into two portions: SML-generic and
634		SML/NJ-specific.  Since it would have been awkward to maintain,
635		I bit the bullet and actually _changed_ the mapping between these
636		numbers and primtycs.  The bottom line of this is that you need
637		a new set of bin- and bootfiles.
638
639		I have built new bootfiles for all architectures, so doing a fresh
640		checkout and config/install.sh should be all you need.
641
642		The newly created library's name is
643
644		$smlnj/viscomp/debugprof.cm
645
646		and its sources live under
647
648		src/compiler/DebugProf
649
650		----------------------------------------------------------------------
651		Name: Matthias Blume
652		Date: 2001/08/15 17:15:00 EDT
653		Tag: blume-20010815-compreorg
654		Description:
655
656		This is a first cut at reorganizing the CM libraries that make up the
657		core of the compiler.  The idea is to separate out pieces that could
658		be used independently by tools, e.g., the parser, the typechecker, etc.
659
660		The current status is a step in this direction, but it is not quite
661		satisfactory yet.  Expect more changes in the future.
662
663		Here is the current (new) organization...
664
665		What used to be $smlnj/viscomp/core.cm is now divided into
666		six CM libraries:
667
668		$smlnj/viscomp/basics.cm
669		/parser.cm
670		/elabdata.cm
671		/elaborate.cm
672		/execute.cm
673		/core.cm
674
675		The CM files for these libraries live under src/system/smlnj/viscomp.
676		All these libraries are proxy libraries that contain precisely
677		one CM library component.  Here are the locations of the components
678		(all within the src/compiler tree):
679
680		Basics/basics.cm
681		Parse/parser.cm
682		ElabData/elabdata.cm
683		Elaborator/elaborate.cm
684		Execution/execute.cm
685		core.cm
686
687		[This organization is the same that has been used already
688		for a while for the architecture-specific parts of the visible
689		compiler and for the old version of core.cm.]
690
691		As you will notice, many source files have been moved from their
692		respective original locations to a new home in one of the above
693		subtrees.
694
695		The division of labor between the new libraries is the following:
696
697		basics.cm:
698		- Simple, basic definitions that pertain to many (or all) of
699		the other libraries.
700		parser.cm:
701		- The SML parser, producing output of type Ast.dec.
702		- The type family for Ast is also defined and exported here.
703		elabdata.cm:
704		- The datatypes that describe input and output of the elaborator.
705		This includes types, absyn, and static environments.
706		elaborator.cm:
707		- The SML/NJ type checker and elaborator.
708		This maps an Ast.dec (with a given static environment) to
709		an Absyn.dec (with a new static environment).
710		- This libraries implements certain modules that used to be
711		structures as functors (to remove dependencies on FLINT).
712		execute.cm:
713		- Everything having to do with executing binary code objects.
714		- Dynamic environments.
715		core.cm:
716		- SML/NJ-specific instantiations of the elaborator and MLRISC.
717		- Top-level modules.
718		- FLINT (this should eventually become its own library)
719
720		Notes:
721
722		I am not 100% happy with the way I separated the elaborator (and its
723		data structures) from FLINT.  Two instances of the same problem:
724
725		1. Data structures contain certain fields that carry FLINT-specific
726		information.  I hacked around this using exn and the property list
727		module from smlnj-lib.  But the fact that there are middle-end
728		specific fields around at all is a bit annoying.
729
730		2. The elaborator calculates certain FLINT-related information.  I tried
731		to make this as abstract as I could using functorization, but, again,
732		the fact that the elaborator has to perform calculations on behalf
733		of the middle-end at all is not nice.
734
735		3. Having to used exn and property lists is unfortunate because it
736		weakens type checking.  The other alternative (parameterizing
737		nearly *everything*) is not appealing, though.
738
739		I removed the "rebinding =" warning hack because due to the new organization
740		it was awkward to maintain it.  As a result, the compiler now issues some of
741		these warnings when compiling init.cmi during bootstrap compilation. On
742		the plus side, you also get a warning when you do, for example:
743		val op = = Int32.+
744		which was not the case up to now.
745
746		I placed "assign" and "deref" into the _Core structure so that the
747		code that deals with the "lazy" keyword can find them there.  This
748		removes the need for having access to the primitive environment
749		during elaboration.
750
751		----------------------------------------------------------------------
752		Name: Matthias Blume
753		Date: 2001/08/13
754		Tag: blume-20010813-closures
755		Description:
756
757		This fix was sent to us by Zhong Shao.  It is supposed to improve the
758		performance of certain loops by avoiding needless closure allocation.
759
760		----------------------------------------------------------------------
761		Name: Lal George
762		Date: 2001/07/31 10:03:23 EDT 2001
763		Tag: george-20010731-x86-fmalloc
764		Description: Fixed bug in x86 calls
765
766		There was a bug where call instructions would mysteriously
767		vanish. The call instruction had to be one that returned
768		a floating point value.
769
770		----------------------------------------------------------------------
771		Name: Lal George
772		Date: 2001/07/19 16:36:29 EDT 2001
773		Tag: george-20010719-simple-cells
774		Description:
775
776		I have dramatically simplified the interface for CELLS in MLRISC.
777
778		In summary, the cells interface is broken up into three parts:
779
780		1. CellsBasis : CELLS_BASIS
781
782		CellsBasis is a top level structure and common for all
783		architectures.  it contains the definitions of basic datatypes
784		and utility  functions over these types.
785
786		2. functor Cells() : CELLS
787
788		Cells generates an interface for CELLS that incorporates the
789		specific resources on the target architecture, such as the
790		presence of special register classes, their number and size,
791		and various useful substructures.
792
793		3. <ARCH>CELLS
794
795		e.g. SparcCells: SPARCCELLS
796
797		<ARCH>CELLS usually contains additional bindings for special
798		registers  on the architecture, such as:
799
800		val r0 : cell           (* register zero *)
801		val y : cell            (* Y register *)
802		val psr : cell          (* processor status register *)
803		...
804
805		The structure returned by applying the Cells functor is opened
806		in this interface.
807
808		The main implication of all this is that the datatypes for cells is
809		split between CellsBasis and CELLS -- a fairly simple change for user
810		code.
811
812		In the old scheme the CELLS interface had a definitional binding of
813		the form:
814
815		signature CELLS = sig
816
817		structure CellsBasis = CellsBasis
818
819		...
820
821		end
822
823		With all the sharing constraints that goes on in MLRISC, this old
824		design  quickly leads to errors such as:
825
826		"structure definition spec inside of sharing ... "
827
828
829		and appears to require an unacceptable amount of sharing and where
830		constraint hackery.
831
832		I think this error message (the interaction of definitional specs and
833		sharing) requires more explanation on our web page.
834
835		----------------------------------------------------------------------
836		Name: Matthias Blume
837		Date: 2001/07/19 15:00:00 EDT
838		Tag: blume-20010719-libreorg
839		Description:
840
841		This update puts together a fairly extensive but straightforward change
842		to the way the libraries that implement the interactive system are
843		organized:
844
845		The biggest change is the elimination of structure Compiler.  As a
846		replacement for this structure, there is now a CM library
847		(known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
848		that exports all the substructures of the original structure Compiler
849		directly.  So instead of saying Compiler.Foo.bar one now simply
850		says Foo.bar.  (The CM libraries actually export a collection of
851		structures that is richer than the collection of substructures of
852		structure Compiler.)
853
854		To make the transition smooth, there is a separate library called
855		$smlnj/compiler/compiler.cm which puts together and exports the
856		original structure Compiler (or at least something very close to it).
857
858		There are five members of the original structure Compiler
859		that are not exported directly but which instead became members
860		of a new structure Backend (described by signature BACKEND).  These are:
861		structure Profile (: PROFILE), structure Compile (: COMPILE), structure
862		Interact (: INTERACT), structure Machine (: MACHINE), and val
863		architecture (: string).
864
865		Structure Compiler.Version has become structure CompilerVersion.
866
867		Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
868		by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
869		or x86, respectively.
870		Each of these exports the same frontend structures that
871		$smlnj/compiler.cm exports.  But they do not have a structure Backend
872		and instead export some structure <Arch>Backend where <Arch> is Alpha32,
873		Hppa, PPC, Sparc, or X86, respectively.
874
875		Library $smlnj/compiler/all.cm exports the union of the exports of
876		$smlnj/compiler/<arch>.cm
877
878		There are no structures <Arch>Compiler anymore, use
879		$smlnj/compiler/<arch>.cm instead.
880
881		Library host-compiler-0.cm is gone.  Instead, the internal library
882		that instantiates CM is now called cm0.cm.  Selection of the host
883		compiler (backend) is no longer done here but. (Responsibility for it
884		now lies with $smlnj/compiler/current.cm.  This seems to be more
885		logical.)
886
887		Many individual files have been moved or renamed.  Some files have
888		been split into multiple files, and some "dead" files have been deleted.
889
890		Aside from these changes to library organization, there are also changes
891		to the way the code itself is organized:
892
893		Structure Binfile has been re-implemented in such a way that it no
894		longer needs any knowledge of the compiler.  It exclusively deals
895		with the details of binfile layout.  It no longer invokes the
896		compiler (for the purpose of creating new prospective binfile
897		content), and it no longer has any knowledge of how to interpret
898		pickles.
899
900		Structure Compile (: COMPILE) has been stripped down to the bare
901		essentials of compilation.  It no longer deals with linking/execution.
902		The interface has been cleaned up considerably.
903
904		Utility routines for dealing with linking and execution have been
905		moved into their own substructures.
906
907		(The ultimate goal of these changes is to provide a light-weight
908		binfile loader/linker (at least for, e.g., stable libraries) that
909		does not require CM or the compiler to be present.)
910
911		CM documentation has been updated to reflect the changes to library
912		organization.
913
914		----------------------------------------------------------------------
915		Name: Matthias Blume
916		Date: 2001/07/10 17:30:00 EDT
917		Tag: Release_110_34
918		Description:
919
920		Minor tweak to 110.34 (re-tagged):
921
922		- README.html file added to CVS repository
923		- runtime compiles properly under FreeBSD 3.X and 4.X
924
925		----------------------------------------------------------------------
926		Name: Matthias Blume
927		Date: 2001/07/10 17:30:00 EDT
928		Tag: Release_110_34
929		Description:
930
931		New version number (110.34). New bootfiles.
932
933		----------------------------------------------------------------------
934		Name: Matthias Blume
935		Date: 2001/07/09 16:00:00 EDT
936		Tag: blume-20010709-more-varargs
937		Description:
938
939		I changed the handling of varargs in ml-nlffigen again:
940		The ellipsis ... will now simply be ignored (with an accompanying warning).
941
942		The immediate effect is that you can actually call a varargs function
943		from ML -- but you can't actually supply any arguments beyond the ones
944		specified explicitly.  (For example, you can call printf with its format
945		string, but you cannot pass additional arguments.)
946
947		This behavior is only marginally more useful than the one before, but
948		it has the advantage that a function or, more importantly, a function
949		type never gets dropped on the floor, thus avoiding follow-up problems with
950		other types that refer to the offending one.
951
952		----------------------------------------------------------------------
953		Name: Matthias Blume
954		Date: 2001/07/09 11:25:00 EDT
955		Tag: blume-20010709-varargs
956		Description:
957
958		1. ckit-lib.cm now exports structure Error
959		2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
960		with a warning accompanied by a source location.  Moreover, it
961		merely skips the offending function or type and proceeds with the
962		rest of its work.u  As a result, one can safely feed C code containing
963		"..." to ml-nlffigen.
964		3. There are some internal improvements to CM, providing slightly
965		more general string substitutions in the tools subsystem.
966
967		----------------------------------------------------------------------
968		Name: Matthias Blume
969		Date: 2001/06/27 15:10:00 EDT
970		Tag: blume-20010627-concur
971		Description:
972
973		Fixed a small bug in CM's handling of parallel compilation.
974		(You could observe the bug by Control-C-interrupting an ordinary
975		CMB.make or CM.stabilize and then attaching some compile servers.
976		The result was that all of a sudden the previously interrupted
977		compilation would continue on its own.  This was because of
978		an over-optimization: CM did not bother to clean out certain queues
979		when no servers were attached "anyway", resulting in the contents
980		of these queues to grab control when new servers did get attached.)
981
982		There is also another minor update to the CM manual.
983
984		----------------------------------------------------------------------
985		Name: Matthias Blume
986		Date: 2001/06/26 16:15:00 EDT
987		Tag: blume-20010626-cmdoc
988		Description:
989
990		Minor typo fixed in CM manual (syntax diagram for libraries).
991
992		----------------------------------------------------------------------
993		Name: Matthias Blume
994		Date: 2001/06/25 22:55:00 EDT
995		Tag: blume-20010625-x86pc
996		Description:
997
998		Fixed a nasty bug in the X86 assembly code that caused signal
999		handlers to fail (crash) randomly.
1000
1001		----------------------------------------------------------------------
1002	Name: Matthias Blume	Name: Matthias Blume
1003	Date: 2001/06/25 12:05:00 EDT	Date: 2001/06/25 12:05:00 EDT
1004	Tag: blume-20010625-nlffigen	Tag: blume-20010625-nlffigen
#	Line 2502	Line 3488
3488	elaborator). There were a lot of changes during my "linkpath" trials	elaborator). There were a lot of changes during my "linkpath" trials
3489	that could have been reverted to their original state but weren't.	that could have been reverted to their original state but weren't.
3490	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
3491	than what was strictly necessary...  (I _did_ resist the tempation	than what was strictly necessary...  (I _did_ resist the temptation
3492	of doing any "global reformatting" to avoid an untimely death at	of doing any "global reformatting" to avoid an untimely death at
3493	Dave's hands. :)	Dave's hands. :)
3494

---

Colored Diff Long Colored Diff Unidiff Context Diff Side by Side

Legend: Removed from v.847   changed lines   Added in v.1034

---

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