Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 874, Mon Jul 9 20:01:40 2001 UTC	revision 1062, Wed Feb 13 21:15:14 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: 2002/02/13 16:15:00 EST
18		Tag: blume-20020213-various
19		Description:
20
21		1. More tweaks to ml-nlffigen:
22
23		- better internal datastructures (resulting in slight speedup)
24		- "-match" option requires exact match
25		- "localized" gensym counters (untagged structs/unions nested within
26		other structs/unions or within typedefs get a fresh counter; their
27		tag will be prefixed by a concatenation of their parents' tags)
28		- bug fixes (related to calculation of transitive closure of types
29		to be included in the output)
30
31		2. Minor Basis updates:
32
33		- added implementations for List.collate and Option.app
34
35		----------------------------------------------------------------------
36		Name: Matthias Blume
37		Date: 2002/02/11 15:55:00 EST
38		Tag: blume-20020211-gensym
39		Description:
40
41		Added a "-gensym" option to command line of ml-nlffigen.  This can be
42		used to specify a "stem" -- a string that is inserted in all "gensym'd"
43		names (ML structure names that correspond to unnamed C structs, unions,
44		and enums), so that separate runs of ml-nlffigen do not clash.
45
46		----------------------------------------------------------------------
47		Name: Matthias Blume
48		Date: 2002/02/11 12:05:00 EST
49		Tag: blume-20020211-gensml
50		Description:
51
52		A quick fix for a problem with GenSML (in the pgraph-util library):
53		Make generation of toplevel "local" optional.  (Strictly speaking,
54		signature definitions within "local" are not legal SML.)
55
56		Other than that: updates to INSTALL and cm/TODO.
57
58		----------------------------------------------------------------------
59		Name: Matthias Blume
60		Date: 2002/02/08 15:00:00 EST
61		Tag: blume-20020208-uniquepid
62		Description:
63
64		0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
65
66		1. The installer (config/install.sh) has gotten smarter:
67
68		- Configuration options are a bit easier to specify now
69		(in config/targets).
70		- Bug in recognizing .tar.bz2 files fixed.
71		- Installer automatically resolves dependencies between
72		configuration options (e.g., if you ask for eXene, you will
73		also get cml -- regardless whether you asked for it or not).
74		- Installer can run in "quieter mode" by setting the environment
75		variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
76		"completely silent", though.
77		- Build HashCons library as part of smlnj-lib.
78
79		2. A new scheme for assigning persistent identifiers to compilation
80		units (and, by extension, to types etc.) has been put into place.
81		This fixes a long-standing bug where types and even dynamic values
82		can get internally confused, thereby compromising type safety
83		(abstraction) and dynamic correctness.  See
84
85		http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
86
87		for an example of how things could go wrong until now.
88
89		The downside of the new scheme is that pids are not quite as
90		persistent as they used to be: CM will generate a fresh pid
91		for every compilation unit that it thinks it sees for the first
92		time.  That means that if you compile starting from a clean, fresh
93		source tree at two different times, you end up with different
94		binaries.
95
96		Cutoff recompilation, however, has not been compromised because
97		CM keeps pid information in special caches between runs.
98
99		----------------------------------------------------------------------
100		Name: Lal George
101		Date: 2002/02/07 15:34:13 EST 2002
102		Tag: <none>
103		Description:
104
105		Compilers that generate assembly code may produce  global labels
106		whose value is resolved at link time. The various peephole optimization
107		modules did not take this in account.
108
109		TODO. The Labels.addrOf function should really return an option
110		type so that clients are forced to deal with this issue, rather
111		than an exception being raised.
112
113		----------------------------------------------------------------------
114		Name: Lal George
115		Date: 2002/02/06 13:55:02 EST
116		Tag: george-20020206-ra-breakup
117		Description:
118
119		1. A bug fix from Allen.
120
121		A typo causes extra fstp %st(0)'s to be generated at compensation
122		edges, which might cause stack underflow traps at runtime.  This
123		occurs in fft where there are extraneous fstps right before the 'into'
124		trap instruction (in this case they are harmless since none of the
125		integers overflow.)
126
127		2. Pulled out various utility modules that were embedded in the modules
128		of the register allocator. I need these modules for other purposes, but
129		they are not complete enough to put into a library (just yet).
130		----------------------------------------------------------------------
131		Name: Matthias Blume
132		Date: 2002/01/31 16:05:00 EST
133		Tag: blume-20020131-sparc-ccalls
134		Description:
135
136		1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
137		of extra stack space by mistake.  Fixed.
138
139		2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
140
141		----------------------------------------------------------------------
142		Name: Allen Leung
143		Date: 2002/01/30
144		Tag: leunga-20020130-nowhere-bug-fix
145		Description:
146
147		MLRISC bug fixes:
148		1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
149		2. MachineInt.fromString was negating its value.
150
151		----------------------------------------------------------------------
152		Name: Matthias Blume
153		Date: 2002/01/29
154		Tag: blume-20020129-INSTALL
155		Description:
156
157		- Added somewhat detailed installation instructions (file INSTALL).
158		- Fixed curl-detection bug in config/install.sh.
159		- It is now possible to select the URL getter using the URLGETTER
160		environment variable:
161
162		not set / "unknown"      --> automatic detection (script tries wget,
163		curl, and lynx)
164		"wget" / "curl" / "lynx" --> use the specified program (script "knows"
165		how to properly invoke them)
166		other                    --> use $URLGETTER directly, it must take
167		precisely two command-line arguments
168		(source URL and destination file name)
169
170		----------------------------------------------------------------------
171		Name: Matthias Blume
172		Date: 2002/01/28
173		Tag: blume-20020128-sparc-ccalls
174		Description:
175
176		- Fixed problem with calculation of "used" registers in sparc-c-calls.
177		- Make use of the allocParam argument in sparc-c-calls.
178
179		----------------------------------------------------------------------
180		Name: Matthias Blume
181		Date: 2002/01/28
182		Tag: blume-20020128-allocParam
183		Description:
184
185		John Reppy:  Changes c-calls API to accept client-callback for
186		allocating extra stack space.
187		me: Corresponding changes to mlriscGen (using a dummy argument that
188		does not change the current behavior).
189
190		----------------------------------------------------------------------
191		Name: Matthias Blume
192		Date: 2002/01/28 12:00:00
193		Tag: Release_110_38
194		Description:
195
196		This time for real!!!
197
198		----------------------------------------------------------------------
199		Name: Matthias Blume
200		Date: 2002/01/28 10:56:00 EST
201		Tag: blume-20020128-retraction
202		Description:
203
204		0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
205		with blume-Release_110_38-retracted.)
206
207		1. Fixed a problem with incorrect rounding modes in real64.sml.
208		(Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
209
210		2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
211		and enums fixed.  The naming of corresponding ML identifiers should
212		now be consistent again.
213
214		----------------------------------------------------------------------
215		Name: Allen Leung
216		Date: 2002/01/27
217		Tag: leunga-20020127-nowhere
218		Description:
219
220		Added a target called nowhere in the configuration scripts.
221		Enabling this will build the MLRISC 'nowhere' tool (for translating
222		programs with where-clauses into legal SML code) during installation.
223
224		----------------------------------------------------------------------
225		Name: Matthias Blume
226		Date: 2002/01/25 21:27:00 EST
227		Tag: blume-Release_110_38-retracted
228		Description:
229
230		Call it a (working) release!  Version is 110.38. Bootfiles are ready.
231
232		README will be added later.
233
234		!!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
235		(Release_110_38) removed.  Reason: Last-minute bug fixes.
236
237		----------------------------------------------------------------------
238		Name: Matthias Blume
239		Date: 2002/01/25
240		Tag: blume-20020125-ffi
241		Description:
242
243		A large number of tweaks and improvements to ml-nlffi-lib and
244		ml-nlffigen:
245
246		- ML represenation types have been streamlined
247		- getter and setter functions work with concrete values, not abstract
248		ones where possible
249		- ml-nlffigen command line more flexible (see README file there)
250		- some bugs have been fixed (hopefully)
251
252		----------------------------------------------------------------------
253		Name: Lal George
254		Date: 2002/01/24
255		Tag: george-20020124-risc-ra-interface
256		Description:
257
258		There is a dramatic simplification in the interface to the
259		register allocator for RISC architectures as a result of making
260		parallel copy instructions explicit.
261
262		----------------------------------------------------------------------
263		Name: Matthias Blume
264		Date: 2002/01/22
265		Tag: blume-20020122-x86-ccalls
266		Description:
267
268		Bug fix for c-calls on x86 (having to do with how char- and
269		short-arguments are being handled).
270
271		----------------------------------------------------------------------
272		Name: Matthias Blume
273		Date: 2002/01/21
274		Tag: blume-20020121-ff
275		Description:
276
277		Another day of fiddling with the FFI...
278
279		1. Bug fix/workaround:  CKIT does not complain about negative array
280		dimensions, so ml-nlffigen has to guard itself against this possibility.
281		(Otherwise a negative dimension would send it into an infinite loop.)
282
283		2. Some of the abstract types (light objects, light pointers, most "base"
284		types) in structure C are now eqtypes.
285
286		3. Added constructors and test functions for NULL function pointers.
287
288		----------------------------------------------------------------------
289		Name: Matthias Blume
290		Date: 2002/01/18
291		Tag: blume-20020118-ready-for-new-release
292		Description:
293
294		Made config/srcarchiveurl point to a new place.  (Will provide boot
295		files shortly.)
296
297		Maybe we christen this to be 110.38?
298
299		----------------------------------------------------------------------
300		Name: Matthias Blume
301		Date: 2002/01/18
302		Tag: blume-20020118-more-ffifiddle
303		Description:
304
305		Today's FFI fiddling:
306
307		- Provided a structure CGetSet with "convenient" versions of C.Get.* and
308		C.Set.* that use concrete (MLRep.*) arguments and results instead
309		of abstract ones.
310
311		- Provided word-style bit operations etc. for "int" representation
312		types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
313		and Long.
314
315		----------------------------------------------------------------------
316		Name: Matthias Blume
317		Date: 2002/01/18
318		Tag: blume-20020118-use-x86-fp
319		Description:
320
321		Now that x86-fast-fp seems to be working, I turned it back on again
322		by default.  (Seems to work fine now, even with the FFI.)
323
324		Other than that, I added some documentation about the FFI to
325		src/ml-nlffigen/README and updated the FFI test examples in
326		src/ml-nlffi-lib/Tests/*.
327
328		----------------------------------------------------------------------
329		Name: Allen Leung
330		Date: 2002/01/17
331		Tag: leunga-20020117-x86-fast-fp-call
332		Description:
333
334		1. Fixed a problem with handling return fp values when x86's fast fp
335		mode is turned on.
336
337		2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
338		of %f32.
339
340		3. Added a constructor INT32lit to the ast of MLRISC tools.
341
342		----------------------------------------------------------------------
343		Name: Matthias Blume
344		Date: 2002/01/16
345		Tag: blume-20020116-ffifiddle
346		Description:
347
348		More fiddling with the FFI interface:
349
350		- Make constness 'c instead of rw wherever possible.  This eliminates
351		the need for certain explicit coercions.  (However, due to ML's
352		value polymorphism, there will still be many cases where explicit
353		coercions are necessary.  Phantom types are not the whole answer
354		to modeling a subtyping relationship in ML.)
355
356		- ro/rw coersions for pointers added.  (Avoids the detour through */&.)
357
358		- "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
359		clumsy workaround for varargs problem.)
360
361		----------------------------------------------------------------------
362		Name: Lal George
363		Date: 2002/01/15
364		Tag: <none>
365		Description:
366
367		1. Since COPY instructions are no longer native to the architecture,
368		a generic functor can be used to implement the expandCopies function.
369
370		2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
371		TEXT segment.
372
373		----------------------------------------------------------------------
374		Name: Matthias Blume
375		Date: 2002/01/15
376		Tag: blume-20020115-ffiupdates
377		Description:
378
379		1. Fix for bug resulting in single-precision float values being returned
380		incorrectly from FFI calls.
381
382		2. Small modifications to C FFI API:
383
384		- memory-allocation routines return straight objects (no options)
385		and raise an exception in out-of-memory situations
386		- unsafe extensions to cast between function pointers and pointers
387		from/to ints
388		- added structure C_Debug as an alternative to structure C where
389		pointer-dereferencing (|*| and |*!) always check for null-pointers
390		- added open_lib' to DynLinkage;  open_lib' works like open_lib
391		but also takes a (possibly empty) list of existing library handles
392		that the current library depends on
393
394		----------------------------------------------------------------------
395		Name: Matthias Blume
396		Date: 2002/01/10
397		Tag: blume-20020110-newffigen
398		Description:
399
400		1. Updates to portable graph code.
401
402		2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
403		more scalable now so that even huge interfaces such as the one
404		for GTK compile in finite time and space. :-)
405		See src/ml-nlffigen/README for details on what's new.
406
407		----------------------------------------------------------------------
408		Name: Lal George
409		Date: 2001/01/09 14:31:35 EST 2002
410		Tag: george-20011206-rm-native-copy
411		Description:
412
413		Removed the native COPY and FCOPY instructions
414		from all the architectures and replaced it with the
415		explicit COPY instruction from the previous commit.
416
417		It is now possible to simplify many of the optimizations
418		modules that manipulate copies. This has not been
419		done in this change.
420
421		----------------------------------------------------------------------
422		Name: Lal George
423		Date: 2001/12/06 16:50:13 EST 2001
424		Tag: george-20011206-mlrisc-instruction
425		Description:
426
427		Changed the representation of instructions from being fully abstract
428		to being partially concrete. That is to say:
429
430		from
431		type instruction
432
433		to
434		type instr                              (* machine instruction *)
435
436		datatype instruction =
437		LIVE of {regs: C.cellset, spilled: C.cellset}
438		| KILL of {regs: C.cellset, spilled: C.cellset}
439		| COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
440		| ANNOTATION of {i: instruction, a: Annotations.annotation}
441		| INSTR of instr
442
443		This makes the handling of certain special instructions that appear on
444		all architectures easier and uniform.
445
446		LIVE and KILL say that a list of registers are live or killed at the
447		program point where they appear. No spill code is generated when an
448		element of the 'regs' field is spilled, but the register is moved to
449		the 'spilled' (which is present, more for debugging than anything else).
450
451		LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
452		We used to generate:
453
454		DEFFREG f1
455		f1 := f2 + f3
456		trapb
457
458		but now generate:
459
460		f1 := f2 + f3
461		trapb
462		LIVE {regs=[f1,f2,f3], spilled=[]}
463
464		Furthermore, the DEFFREG (hack) required that all floating point instruction
465		use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
466		defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
467		in a cleaner alpha implementation. (Hopefully, intel will not get rid of
468		this architecture).
469
470		COPYXXX is intended to replace the parallel COPY and FCOPY  available on
471		all the architectures. This will result in further simplification of the
472		register allocator that must be aware of them for coalescing purposes, and
473		will also simplify certain aspects of the machine description that provides
474		callbacks related to parallel copies.
475
476		ANNOTATION should be obvious, and now INSTR represents the honest to God
477		machine instruction set!
478
479		The <arch>/instructions/<arch>Instr.sml files define certain utility
480		functions for making porting easier -- essentially converting upper case
481		to lower case. All machine instructions (of type instr) are in upper case,
482		and the lower case form generates an MLRISC instruction. For example on
483		the alpha we have:
484
485		datatype instr =
486		LDA of {r:cell, b:cell, d:operand}
487		| ...
488
489		val lda : {r:cell, b:cell, d:operand} -> instruction
490		...
491
492		where lda is just (INSTR o LDA), etc.
493
494		----------------------------------------------------------------------
495		Name: Matthias Blume
496		Date: 2001/11/22 21:40:00 EST
497		Tag: Release_110_37
498		Description:
499
500		Release 110.37.  This time for real.
501
502		----------------------------------------------------------------------
503		Name: Matthias Blume
504		Date: 2001/11/21 16:35:00 EST
505		Tag: blume-20011121-foot-in-mouth
506		Description:
507
508		Removed the "Release_110_37" tag because of a serious bug.
509		This will be re-tagged once the bug is fixed.
510
511		----------------------------------------------------------------------
512		Name: Matthias Blume
513		Date: 2001/11/21 16:14:00 EST
514		Tag: blume-20011121-forgottenfile
515		Description:
516
517		Forgot to add a file.  (Just a .tex-file -- part of
518		the CM manual source.)
519
520		----------------------------------------------------------------------
521		Name: Matthias Blume
522		Date: 2001/11/21 16:10:00 EST
523		Tag: blume-20011121-invalid_110_37
524		Description:
525
526		Note: I removed the original tag "Release_110_37" from this commit
527		because we found a serious bug in all non-x86 backends.
528		- Matthias
529
530		1. Modifications to the SML/NJ code generator and to the runtime system
531		so that code object name strings are directly inserted into code
532		objects at code generation time.  The only business the runtime system
533		has with this is now to read the name strings on occasions.
534		(The encoding of the name string has also changed somewhat.)
535
536		2. CM now implements a simple "set calculus" for specifying export lists.
537		In particular, it is now possible to refer to the export lists of
538		other libraries/groups/sources and form unions as well as differences.
539		See the latest CM manual for details.
540
541		3. An separate notion of "proxy" libraries has again be eliminated from
542		CM's model.  (Proxy libraries are now simply a special case of using
543		the export list calculus.)
544
545		4. Some of the existing libraries now take advantage of the new set
546		calculus.
547		(Notice that not all libraries have been converted because some
548		of the existing .cm-files are supposed to be backward compatible
549		with 110.0.x.)
550
551		5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
552		"General.exnMessage"!  The former relies on a certain hook to be
553		initialized, and that often does not happen in the stand-alone case.)
554
555		----------------------------------------------------------------------
556		Name: Lal George
557		Date: 2001/11/21  13:56:18 EST
558		Tag: george-2001121-pseudo-ops
559		Description:
560
561		Implemented a complete redesign of MLRISC pseudo-ops. Now there
562		ought to never be any question of incompatabilities with
563		pseudo-op syntax expected by host assemblers.
564
565		For now, only modules supporting GAS syntax are implemented
566		but more should follow, such as MASM, and vendor assembler
567		syntax, e.g. IBM as, Sun as, etc.
568
569		----------------------------------------------------------------------
570		Name: Matthias Blume
571		Date: 2001/11/14 11:52:00 EST
572		Tag: blume-20011114-srcname
573		Description:
574
575		1. Routed the name of the current source file to mlriscgen where it
576		should be directly emitted into the code object.  (This last part
577		is yet to be done.)
578
579		2. Some cleanup of the pgraph code to make it match the proposal that
580		I put out the other day.  (The proposal notwithstanding, things are
581		still in flux here.)
582
583		----------------------------------------------------------------------
584		Name: Lal George
585		Date: 2001/11/14 09:44:04 EST
586		Tag:
587		Description:
588
589		Fix for a backpatching bug reported by Allen.
590
591		Because the boundary between short and long span-dependent
592		instructions is +/- 128, there are an astounding number of
593		span-dependent instructions whose size is over estimated.
594
595		Allen came up with the idea of letting the size of span
596		dependent instructions be non-monotonic, for a maxIter
597		number of times, after which the size must be monotonically
598		increasing.
599
600		This table shows the number of span-dependent instructions
601		whose size was over-estimated as a function of maxIter, for the
602		file Parse/parse/ml.grm.sml:
603
604		maxIter            # of instructions:
605		10                      687
606		20                      438
607		30                      198
608		40                        0
609
610		In compiling the compiler, there is no significant difference in
611		compilation speed between maxIter=10 and maxIter=40. Actually,
612		my measurements showed that maxIter=40 was a tad faster than
613		maxIter=10! Also 96% of the  files in the compiler reach a fix
614		point within 13 iterations, so fixing maxIter at 40, while high,
615		is okay.
616
617		----------------------------------------------------------------------
618		Name: Matthias Blume
619		Date: 2001/10/31 15:25:00 EST
620		Tag: blume-20011031-pgraph
621		Description:
622
623		CKIT:
624		* Changed the "Function" constructor of type Ast.ctype to carry optional
625		argument identifiers.
626		* Changed the return type of TypeUtil.getFunction accordingly.
627		* Type equality ignores the argument names.
628		* TypeUtil.composite tries to preserve argument names but gives up quickly
629		if there is a mismatch.
630
631		installation script:
632		* attempts to use "curl" if available (unless "wget" is available as well)
633
634		CM:
635		* has an experimental implementation of "portable graphs" which I will
636		soon propose as an implementation-independent library format
637		* there are also new libraries $/pgraph.cm and $/pgraph-util.cm
638
639		NLFFI-LIB:
640		* some cleanup (all cosmetic)
641
642		NLFFIGEN:
643		* temporarily disabled the mechanism that suppresses ML output for
644		C definitions whose identifiers start with an underscore character
645		* generate val bindings for enum constants
646		* user can request that only one style (light or heavy) is being used;
647		default is to use both (command-line arguments: -heavy and -light)
648		* fixed bug in handling of function types involving incomplete pointers
649		* generate ML entry points that take record arguments (i.e., using
650		named arguments) for C functions that have a prototype with named
651		arguments
652		(see changes to CKIT)
653
654		----------------------------------------------------------------------
655		Name: Allen Leung
656		Date: 2001/10/27 20:34:00 EDT
657		Tag: leunga-20011027-x86-fast-fp-call
658		Description:
659
660		Fixed the bug described in blume-20010920-slowfp.
661
662		The fix involves
663		1. generating FCOPYs in FSTP in ia32-svid
664		2. marking a CALL with the appropriate annotation
665
666		----------------------------------------------------------------------
667		Name: Matthias Blume
668		Date: 2001/10/16 11:32:00 EDT
669		Tag: blume-20011016-netbsd
670		Description:
671
672		Underscore patch from Chris Richards (fixing problem with compiling
673		runtime system under recent NetBSD).
674
675		----------------------------------------------------------------------
676		Name: Allen Leung
677		Date: 2001/10/12 17:18:32 EDT 2001
678		Tag: leung-20011012-x86-printflowgraph
679		Description:
680
681		X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
682
683		----------------------------------------------------------------------
684		Name: Lal George
685		Date: 2001/10/11 23:51:34 EDT
686		Tag: george-20011011-too-many-instrs
687		Description:
688
689		The representation of a program point never expected to see more
690		than 65536 instructions in a basic block!
691
692		----------------------------------------------------------------------
693		Name: Lal George
694		Date: 2001/10/09 09:41:37 EDT
695		Tag: george-20011008-mlrisc-labels
696		Description:
697
698		Changed the machine description files to support printing of
699		local and global labels in assembly code, based on host assembler
700		conventions.
701
702		----------------------------------------------------------------------
703		Name: Matthias Blume
704		Date: 2001/09/25 15:25:00 EDT
705		Tag: blume-20010925-exninfo
706		Description:
707
708		I provided a non-hook implementation of exnName (at the toplevel) and
709		made the "dummy" implementation of exnMessage (at the toplevel) more
710		useful: if nothing gets "hooked in", then at least you are going to
711		see the exception name and a message indicating why you don't see more.
712
713		[For the time being, programs that need exnMessage and want to use
714		ml-build should either use General.exnMessage (strongly recommended) or
715		refer to structure General at some other point so that CM sees a
716		static dependency.]
717
718		[Similar remarks go for "print" and "use":  If you want to use their
719		functionality in stand-alone programs generated by ml-build, then use
720		TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
721
722		----------------------------------------------------------------------
723		Name: Matthias Blume
724		Date: 2001/09/20 17:28:00 EDT
725		Tag: blume-20010920-slowfp
726		Description:
727
728		Allen says that x86-fast-fp is not safe yet, so I turned it off again...
729
730		----------------------------------------------------------------------
731		Name: Matthias Blume
732		Date: 2001/09/20 17:20:00 EDT
733		Tag: blume-20010920-canonicalpaths
734		Description:
735
736		0. Updated the BOOT file (something that I forgot to do earlier).
737
738		1. Small internal change to CM so that it avoids "/../" in filenames
739		as much as possible (but only where it is safe).
740
741		2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
742		that arguments that contain delimiters are passed through correctly.
743		This change also means that all "special" arguments of the form
744		@SMLxxx... must come first.
745
746		3. Changed install script to put relative anchor names for tool commands
747		into pathconfig.
748
749		----------------------------------------------------------------------
750		Name: Matthias Blume
751		Date: 2001/09/18 15:35:00 EDT
752		Tag: blume-20010918-readme11036
753		Description:
754
755		Added README files.
756
757		----------------------------------------------------------------------
758		Name: Matthias Blume
759		Date: 2001/09/18 11:45:00 EDT
760		Tag: Release_110_36 (retag)
761		Description:
762
763		Fixed mistake in config/preloads. Retagged as 110.36.
764
765		----------------------------------------------------------------------
766		Name: Matthias Blume
767		Date: 2001/09/18 09:40:00 EDT
768		Tag: Release_110_36_orig (tag changed)
769		Description:
770
771		New version (110.36).  New bootfiles.
772
773		----------------------------------------------------------------------
774		Name: Matthias Blume
775		Date: 2001/09/14 16:15:00 EDT
776		Tag: blume-20010914-x86fastfp
777		Description:
778
779		John committed some changes that Allen made, in particular a (hopefully)
780		correctly working version of the x86-fp module.
781
782		I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
783		flag to "true".  Everything seems to compile to a fixpoint ok, and
784		"mandelbrot" speeds up by about 15%.
785
786		----------------------------------------------------------------------
787		Name: Matthias Blume
788		Date: 2001/09/13 11:20:00 EDT
789		Tag: blume-20010913-minimal
790		Description:
791
792		1. Stefan Monnier's patch to fix a miscompilation problem that
793		was brought to light by John Reppy's work on Moby.
794
795		2. Implemented a minimal "structure Compiler" that contains just
796		"version" and "architecture".  The minimal version will be
797		available when the full version is not.  This is for backward-
798		compatibility with code that wants to test Compiler.version.
799
800		----------------------------------------------------------------------
801		Name: Matthias Blume
802		Date: 2001/08/28 14:03:00 EDT
803		Tag: blume-20010828-ml-lex
804		Description:
805
806		Fix for bug 1581, received from Neophytos Michael.
807
808		----------------------------------------------------------------------
809		Name: Matthias Blume
810		Date: 2001/08/27 11:20:00 EDT
811		Tag: blume-20010827-readme11035
812		Description:
813
814		Fleshed out the README file for 110.35.
815
816		----------------------------------------------------------------------
817		Name: Matthias Blume
818		Date: 2001/08/24 17:10:00 EDT
819		Tag: Release_110_35
820		Description:
821
822		New version number (110.35).  New bootfiles.
823
824		----------------------------------------------------------------------
825		Name: Lal George
826		Date: 2001/08/24 13:47:18 EDT 2001
827		Tag: george-20010824-MLRISC-graphs
828		Description:
829
830		removed clusters from MLRISC completely and replaced with graphs.
831
832		----------------------------------------------------------------------
833		Name: Matthias Blume
834		Date: 2001/08/23 17:50:00 EDT
835		Tag: blume-20010823-toplevel
836		Description:
837
838		- some reorganization of the code that implements various kinds of
839		environments in the compiler (static, dynamic, symbolic, combined)
840		- re-implemented the EnvRef module so that evalStream works properly
841		(if the stream contains references to "use", "CM.make", etc.)
842		- cleaned up evalloop.sml and interact.sml (but they need more cleaning)
843
844		----------------------------------------------------------------------
845		Name: Matthias Blume
846		Date: 2001/08/20 15:50 EDT
847		Tag: blume20010820-slipup
848		Description:
849
850		I forgot to commit a few files.  Here they are...
851
852		----------------------------------------------------------------------
853		Name: Matthias Blume
854		Date: 2001/08/20 15:35:00 EDT
855		Tag: blume-20010820-debugprof
856		Description:
857
858		!!!! NEW BOOTFILES !!!!
859
860		This is another round of reorganizing the compiler sources.  This
861		time the main goal was to factor out all the "instrumentation"
862		passes (for profiling and backtracing) into their own library.
863		The difficulty was to do it in such a way that it does not depend
864		on elaborate.cm but only on elabdata.cm.
865
866		Therefore there have been further changes to both elaborate.cm and
867		elabdata.cm -- more "generic" things have been moved from the former
868		to the latter.  As a result, I was forced to split the assignment
869		of numbers indicating "primtyc"s into two portions: SML-generic and
870		SML/NJ-specific.  Since it would have been awkward to maintain,
871		I bit the bullet and actually _changed_ the mapping between these
872		numbers and primtycs.  The bottom line of this is that you need
873		a new set of bin- and bootfiles.
874
875		I have built new bootfiles for all architectures, so doing a fresh
876		checkout and config/install.sh should be all you need.
877
878		The newly created library's name is
879
880		$smlnj/viscomp/debugprof.cm
881
882		and its sources live under
883
884		src/compiler/DebugProf
885
886		----------------------------------------------------------------------
887		Name: Matthias Blume
888		Date: 2001/08/15 17:15:00 EDT
889		Tag: blume-20010815-compreorg
890		Description:
891
892		This is a first cut at reorganizing the CM libraries that make up the
893		core of the compiler.  The idea is to separate out pieces that could
894		be used independently by tools, e.g., the parser, the typechecker, etc.
895
896		The current status is a step in this direction, but it is not quite
897		satisfactory yet.  Expect more changes in the future.
898
899		Here is the current (new) organization...
900
901		What used to be $smlnj/viscomp/core.cm is now divided into
902		six CM libraries:
903
904		$smlnj/viscomp/basics.cm
905		/parser.cm
906		/elabdata.cm
907		/elaborate.cm
908		/execute.cm
909		/core.cm
910
911		The CM files for these libraries live under src/system/smlnj/viscomp.
912		All these libraries are proxy libraries that contain precisely
913		one CM library component.  Here are the locations of the components
914		(all within the src/compiler tree):
915
916		Basics/basics.cm
917		Parse/parser.cm
918		ElabData/elabdata.cm
919		Elaborator/elaborate.cm
920		Execution/execute.cm
921		core.cm
922
923		[This organization is the same that has been used already
924		for a while for the architecture-specific parts of the visible
925		compiler and for the old version of core.cm.]
926
927		As you will notice, many source files have been moved from their
928		respective original locations to a new home in one of the above
929		subtrees.
930
931		The division of labor between the new libraries is the following:
932
933		basics.cm:
934		- Simple, basic definitions that pertain to many (or all) of
935		the other libraries.
936		parser.cm:
937		- The SML parser, producing output of type Ast.dec.
938		- The type family for Ast is also defined and exported here.
939		elabdata.cm:
940		- The datatypes that describe input and output of the elaborator.
941		This includes types, absyn, and static environments.
942		elaborator.cm:
943		- The SML/NJ type checker and elaborator.
944		This maps an Ast.dec (with a given static environment) to
945		an Absyn.dec (with a new static environment).
946		- This libraries implements certain modules that used to be
947		structures as functors (to remove dependencies on FLINT).
948		execute.cm:
949		- Everything having to do with executing binary code objects.
950		- Dynamic environments.
951		core.cm:
952		- SML/NJ-specific instantiations of the elaborator and MLRISC.
953		- Top-level modules.
954		- FLINT (this should eventually become its own library)
955
956		Notes:
957
958		I am not 100% happy with the way I separated the elaborator (and its
959		data structures) from FLINT.  Two instances of the same problem:
960
961		1. Data structures contain certain fields that carry FLINT-specific
962		information.  I hacked around this using exn and the property list
963		module from smlnj-lib.  But the fact that there are middle-end
964		specific fields around at all is a bit annoying.
965
966		2. The elaborator calculates certain FLINT-related information.  I tried
967		to make this as abstract as I could using functorization, but, again,
968		the fact that the elaborator has to perform calculations on behalf
969		of the middle-end at all is not nice.
970
971		3. Having to used exn and property lists is unfortunate because it
972		weakens type checking.  The other alternative (parameterizing
973		nearly *everything*) is not appealing, though.
974
975		I removed the "rebinding =" warning hack because due to the new organization
976		it was awkward to maintain it.  As a result, the compiler now issues some of
977		these warnings when compiling init.cmi during bootstrap compilation. On
978		the plus side, you also get a warning when you do, for example:
979		val op = = Int32.+
980		which was not the case up to now.
981
982		I placed "assign" and "deref" into the _Core structure so that the
983		code that deals with the "lazy" keyword can find them there.  This
984		removes the need for having access to the primitive environment
985		during elaboration.
986
987		----------------------------------------------------------------------
988		Name: Matthias Blume
989		Date: 2001/08/13
990		Tag: blume-20010813-closures
991		Description:
992
993		This fix was sent to us by Zhong Shao.  It is supposed to improve the
994		performance of certain loops by avoiding needless closure allocation.
995
996		----------------------------------------------------------------------
997		Name: Lal George
998		Date: 2001/07/31 10:03:23 EDT 2001
999		Tag: george-20010731-x86-fmalloc
1000		Description: Fixed bug in x86 calls
1001
1002		There was a bug where call instructions would mysteriously
1003		vanish. The call instruction had to be one that returned
1004		a floating point value.
1005
1006		----------------------------------------------------------------------
1007		Name: Lal George
1008		Date: 2001/07/19 16:36:29 EDT 2001
1009		Tag: george-20010719-simple-cells
1010		Description:
1011
1012		I have dramatically simplified the interface for CELLS in MLRISC.
1013
1014		In summary, the cells interface is broken up into three parts:
1015
1016		1. CellsBasis : CELLS_BASIS
1017
1018		CellsBasis is a top level structure and common for all
1019		architectures.  it contains the definitions of basic datatypes
1020		and utility  functions over these types.
1021
1022		2. functor Cells() : CELLS
1023
1024		Cells generates an interface for CELLS that incorporates the
1025		specific resources on the target architecture, such as the
1026		presence of special register classes, their number and size,
1027		and various useful substructures.
1028
1029		3. <ARCH>CELLS
1030
1031		e.g. SparcCells: SPARCCELLS
1032
1033		<ARCH>CELLS usually contains additional bindings for special
1034		registers  on the architecture, such as:
1035
1036		val r0 : cell           (* register zero *)
1037		val y : cell            (* Y register *)
1038		val psr : cell          (* processor status register *)
1039		...
1040
1041		The structure returned by applying the Cells functor is opened
1042		in this interface.
1043
1044		The main implication of all this is that the datatypes for cells is
1045		split between CellsBasis and CELLS -- a fairly simple change for user
1046		code.
1047
1048		In the old scheme the CELLS interface had a definitional binding of
1049		the form:
1050
1051		signature CELLS = sig
1052
1053		structure CellsBasis = CellsBasis
1054
1055		...
1056
1057		end
1058
1059		With all the sharing constraints that goes on in MLRISC, this old
1060		design  quickly leads to errors such as:
1061
1062		"structure definition spec inside of sharing ... "
1063
1064
1065		and appears to require an unacceptable amount of sharing and where
1066		constraint hackery.
1067
1068		I think this error message (the interaction of definitional specs and
1069		sharing) requires more explanation on our web page.
1070
1071		----------------------------------------------------------------------
1072		Name: Matthias Blume
1073		Date: 2001/07/19 15:00:00 EDT
1074		Tag: blume-20010719-libreorg
1075		Description:
1076
1077		This update puts together a fairly extensive but straightforward change
1078		to the way the libraries that implement the interactive system are
1079		organized:
1080
1081		The biggest change is the elimination of structure Compiler.  As a
1082		replacement for this structure, there is now a CM library
1083		(known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
1084		that exports all the substructures of the original structure Compiler
1085		directly.  So instead of saying Compiler.Foo.bar one now simply
1086		says Foo.bar.  (The CM libraries actually export a collection of
1087		structures that is richer than the collection of substructures of
1088		structure Compiler.)
1089
1090		To make the transition smooth, there is a separate library called
1091		$smlnj/compiler/compiler.cm which puts together and exports the
1092		original structure Compiler (or at least something very close to it).
1093
1094		There are five members of the original structure Compiler
1095		that are not exported directly but which instead became members
1096		of a new structure Backend (described by signature BACKEND).  These are:
1097		structure Profile (: PROFILE), structure Compile (: COMPILE), structure
1098		Interact (: INTERACT), structure Machine (: MACHINE), and val
1099		architecture (: string).
1100
1101		Structure Compiler.Version has become structure CompilerVersion.
1102
1103		Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
1104		by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
1105		or x86, respectively.
1106		Each of these exports the same frontend structures that
1107		$smlnj/compiler.cm exports.  But they do not have a structure Backend
1108		and instead export some structure <Arch>Backend where <Arch> is Alpha32,
1109		Hppa, PPC, Sparc, or X86, respectively.
1110
1111		Library $smlnj/compiler/all.cm exports the union of the exports of
1112		$smlnj/compiler/<arch>.cm
1113
1114		There are no structures <Arch>Compiler anymore, use
1115		$smlnj/compiler/<arch>.cm instead.
1116
1117		Library host-compiler-0.cm is gone.  Instead, the internal library
1118		that instantiates CM is now called cm0.cm.  Selection of the host
1119		compiler (backend) is no longer done here but. (Responsibility for it
1120		now lies with $smlnj/compiler/current.cm.  This seems to be more
1121		logical.)
1122
1123		Many individual files have been moved or renamed.  Some files have
1124		been split into multiple files, and some "dead" files have been deleted.
1125
1126		Aside from these changes to library organization, there are also changes
1127		to the way the code itself is organized:
1128
1129		Structure Binfile has been re-implemented in such a way that it no
1130		longer needs any knowledge of the compiler.  It exclusively deals
1131		with the details of binfile layout.  It no longer invokes the
1132		compiler (for the purpose of creating new prospective binfile
1133		content), and it no longer has any knowledge of how to interpret
1134		pickles.
1135
1136		Structure Compile (: COMPILE) has been stripped down to the bare
1137		essentials of compilation.  It no longer deals with linking/execution.
1138		The interface has been cleaned up considerably.
1139
1140		Utility routines for dealing with linking and execution have been
1141		moved into their own substructures.
1142
1143		(The ultimate goal of these changes is to provide a light-weight
1144		binfile loader/linker (at least for, e.g., stable libraries) that
1145		does not require CM or the compiler to be present.)
1146
1147		CM documentation has been updated to reflect the changes to library
1148		organization.
1149
1150		----------------------------------------------------------------------
1151		Name: Matthias Blume
1152		Date: 2001/07/10 17:30:00 EDT
1153		Tag: Release_110_34
1154		Description:
1155
1156		Minor tweak to 110.34 (re-tagged):
1157
1158		- README.html file added to CVS repository
1159		- runtime compiles properly under FreeBSD 3.X and 4.X
1160
1161		----------------------------------------------------------------------
1162		Name: Matthias Blume
1163		Date: 2001/07/10 17:30:00 EDT
1164		Tag: Release_110_34
1165		Description:
1166
1167		New version number (110.34). New bootfiles.
1168
1169		----------------------------------------------------------------------
1170		Name: Matthias Blume
1171	Date: 2001/07/09 16:00:00 EDT	Date: 2001/07/09 16:00:00 EDT
1172	Tag: blume-20010709-more-varargs	Tag: blume-20010709-more-varargs
1173	Description:	Description:
#	Line 2570	Line 3724
3724	elaborator). There were a lot of changes during my "linkpath" trials	elaborator). There were a lot of changes during my "linkpath" trials
3725	that could have been reverted to their original state but weren't.	that could have been reverted to their original state but weren't.
3726	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
3727	than what was strictly necessary...  (I _did_ resist the tempation	than what was strictly necessary...  (I _did_ resist the temptation
3728	of doing any "global reformatting" to avoid an untimely death at	of doing any "global reformatting" to avoid an untimely death at
3729	Dave's hands. :)	Dave's hands. :)
3730

---

Colored Diff Long Colored Diff Unidiff Context Diff Side by Side

Legend: Removed from v.874   changed lines   Added in v.1062

---

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