Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 716, Sat Nov 4 05:42:35 2000 UTC	revision 810, Fri Apr 20 13:26:46 2001 UTC
#	Line 13	Line 13
13	Description:	Description:
14
15	----------------------------------------------------------------------	----------------------------------------------------------------------
16		Name: Lal George
17		Date: 2001/04/20 09:15:28 EDT
18		Tag: george-20010420-macosX
19		Description:
20
21		- Changes to port to Mac OS X; Darwin.
22
23		- In the process I found that sqrt was broken on the PPC, because the
24		fsqrt instruction is not implemented.
25
26		----------------------------------------------------------------------
27		Name: Matthias Blume
28		Date: 2001/04/18 12:45:00 EDT
29		Tag: blume-20010418-ccalls
30		Description:
31
32		- fixed two off-by-4 errors in the x86-specific c-calls implementation
33		(this bug prevented structure arguments containing pointers from being
34		passed correctly)
35		- changed the raw-C-call code in mlriscGen.sml in such a way that
36		structure arguments are represented as a pointer to the beginning
37		of the structure (instead of having a series of synthesized arguments,
38		one for each structure member)
39
40		- made makeml script's verbosity level configurable via environment
41		variable (MAKEML_VERBOSITY)
42
43		- eliminteded placeholder implementations for f32l, w16s, i16s, and f32s
44		in rawmem-x86.sml; we are now using the real thing
45
46		----------------------------------------------------------------------
47		Name: Matthias Blume
48		Date: 2001/03/22 16:25:00 EST
49		Tag: blume-20010322-bootfiles
50		Description:
51
52		Created a new set of bootfiles (for your automatic installation convenience).
53
54		----------------------------------------------------------------------
55		Name: Matthias Blume
56		Date: 2001/03/22 15:10:00 EST
57		Tag: blume-20010322-rawmem-parcm
58		Description:
59
60		1. All "raw memory access" primitives for the new FFI are implemented now
61		(at least on the x86).
62		2. Some further cleanup of CM's parallel make mechanism.
63
64		----------------------------------------------------------------------
65		Name: Matthias Blume
66		Date: 2001/03/19 17:53:00 EST
67		Tag: blume-20010319-parallel
68		Description:
69
70		Parallel make (using compile servers) now works again.
71
72		To this end, CM.stabilize and CMB.make have been modified to work in
73		two passes when compile servers are attached:
74		1. Compile everything, do not perform stabilization; this pass
75		uses compile servers
76		2. Stabilize everything; this pass does not use compile servers
77		If there are no compile servers, the two passes are combined into one
78		(as before).  Splitting the passes increases the inherent parallelism
79		in the dependency graph because the entire graph including all
80		libraries is available at the same time.  This, in turn, improves
81		server utilization.  The downside is that the master process will
82		have to do some extra work after compilation is done (because for
83		technical reasons it must re-read all the binfiles during stabilization).
84
85		----------------------------------------------------------------------
86		Name: Matthias Blume
87		Date: 2001/03/16 12:22:00 EST
88		Tag: blume-20010316-bootfiles
89		Description:
90
91		Created a new set of bootfiles (for your automatic installation convenience).
92
93		----------------------------------------------------------------------
94		Name: Matthias Blume
95		Date: 2001/03/16 11:00:00 EST
96		Tag: blume-20010316-MLTREE-fixup
97		Description:
98
99		This is a minor fixup for an (untagged) earlier commit by Allen.
100		(A file was missing).
101
102		----------------------------------------------------------------------
103		Name: Allen Leung
104		Date: Mon Mar  5 18:54:57 EST 2001
105		Tag: leunga-20010305-cut-support
106
107		1. New support for alternative control-flow in MLTREE.
108		Currently we support
109
110		FLOW_TO(CALL ...., [k1,...,kn])
111
112		This is needed for 'cuts to' in C-- and try/handle-like constructs
113		in Moby
114
115		New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
116
117		2. Register Allocator
118
119		Changes in interface [from Fermin, John]
120
121		3. Alpha 8-bit SLL support [Fermin]
122
123		4. All architectures
124
125		A new module (ClusterExpandCopies) for expanding parallel copies.
126
127		----------------------------------------------------------------------
128		Name: Allen Leung
129		Date: 2001/02/27 23:07:00 EST
130		Tag: leunga-20010227-minor-stuff
131
132		1. Alpha bug fix for CMOVNE
133		2. Handle mltree COND(..,FCMP ...,...)
134		3. Bug fix in simplifier
135
136		----------------------------------------------------------------------
137		Name: Matthias Blume
138		Date: 2001/01/30 17:50:00 EST
139		Tag: blume-20010130-sync
140		Description:
141
142		This is just a minor update to sync my devel branch with the main brach.
143		The only visible change is the addition of some README files.
144
145		----------------------------------------------------------------------
146		Name: Matthias Blume
147		Date: 2001/01/12 23:30:00 JST
148		Tag: blume-20010112-bootfiles
149		Description:
150
151		Made a new set of bootfiles that goes with the current state of the
152		repository.
153
154		----------------------------------------------------------------------
155		Name: Matthias Blume
156		Date: 2001/01/12 21:20:00 JST
157		Tag: blume-20010112-sync
158		Description:
159
160		I am just flushing out some minor changes that had accumulated in
161		my private branch in order to sync with the main tree.  (This is
162		mainly because I had CVS trouble when trying to merge _into_ my
163		private branch.)
164
165		Most people should be completely unaffected by this.
166
167		----------------------------------------------------------------------
168		Name: Allen Leung
169		Date: Thu Jan 11 21:03:00 EST 2001
170		Tag: leunga-20010111-labexp=mltree
171		Description:
172
173		1.  Removed the type LabelExp and replace it by MLTree.
174		2.  Rewritten mltree-simplify with the pattern matcher tool.
175		3.  There were some bugs in alpha code generator which would break
176		64-bit code generation.
177		4.  Redo the tools to generate code with the
178		5.  The CM files in MLRISC (and in src/system/smlnj/MLRISC)
179		are now generated by perl scripts.
180
181		----------------------------------------------------------------------
182		Name: Matthias Blume
183		Date: 2001/01/10 21:55:00 JST
184		Tag: blume-20010110-rcc
185		Description:
186
187		The RCC stuff now seems to work (but only on the x86).
188		This required hacking of the c-calls interface (and -implementation) in
189		MLRISC.
190
191		Normal compiler users should be unaffected.
192
193		----------------------------------------------------------------------
194		Name: Matthias Blume
195		Date: 2001/01/09 01:20:00 JST
196		Tag: blume-20010109-rcc
197		Description:
198
199		This is a fairly big patch, flushing out a large number of pending
200		changes that I made to my development copy over the last couple of days.
201
202		Of practical relevance at this moment is a workaround for a pickling
203		bug that Allen ran into the other day.  The cause of the bug itself is
204		still unknown and it might be hard to fix it properly, but the
205		workaround has some merits of its own (namely somewhat reducing pickling
206		overhead for certain libraries).  Therefore, I think this solution should
207		be satisfactory at this time.
208
209		The rest of the changes (i.e., the vast majority) has to do with my
210		ongoing efforts of providing direct support for C function calls from
211		ML.  At the moment there is a new primop "RAW_CCALL", typing magic
212		in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
213		case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
214		translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
215		handle RCC.
216
217		The last part (the changes to mlriscGen.sml) are still known to be
218		wrong on the x86 and not implemented on all other architectures.  But
219		the infrastructure is in place. I had to change a few functor
220		signatures in the backend to be able to route the CCalls interface
221		from MLRISC there, and I had to specialize the mltree type (on the
222		x86) to include the necessary extensions. (The extensions themselves
223		were already there and redy to go in MLRISC/x86).
224
225		Everything should be very happy as soon as someone helps me with
226		mlriscGen.sml...
227
228		In any case, nothing of this should matter to anyone as long as the
229		new primop is not being used (which is going to be the case unless you
230		find it where I hid it :). The rest of the compiler is completely
231		unaffected.
232
233		----------------------------------------------------------------------
234		Name: Matthias Blume
235		Date: 2001/01/05 00:30:00 JST
236		Tag: blume-20010105-primops
237		Description:
238
239		Added some experimental support for work that I am doing right now.
240		These changes mostly concern added primops, but there is also a new
241		experimental C library in the runtime system (but currently not enabled
242		anywhere except on Linux/X86).
243
244		In the course of adding primops (and playing with them), I discovered that
245		Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
246		broken.  (Zhong was very right he labeled this stuff as "major gross hack".)
247		To recover, I made type information in INL_PRIM mandatory and changed
248		prim.sml as well as built-in.sml accordingly.  The InLine structure now
249		has complete, correct type information (i.e., no bottom types).
250
251		Since all these changes mean that we need new binfiles, I also bumped the
252		version number to 110.32.1.
253
254		----------------------------------------------------------------------
255		Name: Matthias Blume
256		Date: 2000/12/30 22:10:00 JST
257		Tag: blume-20001230-various
258		Description:
259
260		Added proxy libraries for MLRISC and let MLRISC libraries refer
261		to each other using path anchors.  (See CM manual for explanation.)
262
263		Updated CM documentation.
264
265		Fixed some bugs in CM.
266
267		Implemented "proxy" libraries (= syntactic sugar for CM).
268
269		Added "-quiet" option to makeml and changed runtime system accordingly.
270
271		Added cleanup handler for exportML to reset timers and compiler stats.
272
273		----------------------------------------------------------------------
274		Name: Lal George
275		Date: 2000/12/22 22:22:58 EST 2000
276		Tag: Release_110_32
277		Description:
278
279		Infinite precision used throughout MLRISC.
280		see MLRISC/mltree/machine-int.sig
281
282		----------------------------------------------------------------------
283		Name: Matthias Blume
284		Date: 2000/12/22 23:16:00 JST
285		Tag: blume-20001222-warn
286		Description:
287
288		Corrected wording and formatting of some CM warning message which I
289		broke in my previous patch.
290
291		----------------------------------------------------------------------
292		Name: Matthias Blume
293		Date: 2000/12/22 21:20:00 JST
294		Tag: blume-20001222-anchorenv
295		Description:
296
297		Fixed CM's handling of anchor environments in connection with CMB.make.
298
299		----------------------------------------------------------------------
300		Name: Matthias Blume
301		Date: 2000/12/22 13:15:00 JST
302		Tag: blume-20001222-cleanup
303		Description:
304
305		Removed src/cm/ffi which does not (and did not) belong here.
306
307		----------------------------------------------------------------------
308		Name: Matthias Blume
309		Date: 2000/12/21 23:55:00 JST
310		Tag: blume-20001221-exn
311		Description:
312
313		Probably most important: CM no longer silently swallows all exceptions
314		in the compiler.
315		Plus: some other minor CM changes.  For example, CM now reports some
316		sizes for generated binfiles (code, data, envpickle, lambdapickle).
317
318		----------------------------------------------------------------------
319		Name: Matthias Blume
320		Date: 2000/12/15 00:01:05 JST
321		Tag: blume-20001215-dirtool
322		Description:
323
324		- "dir" tool added.
325		- improvements and cleanup to Tools structure
326		- documentation updates
327
328		----------------------------------------------------------------------
329		Name: Allen Leung
330		Date: Thu Dec 14 03:45:24 EST 2000
331		Description:
332		Tag:  leunga-20001214-int-inf
333		Description:
334
335		In IntInf, added these standard functions, which are missing from our
336		implementation:
337
338		andb : int * int -> int
339		xorb : int * int -> int
340		orb  : int * int -> int
341		notb : int -> int
342		<<   : int * word -> int
343		~>>  : int * word -> int
344
345		Not tested, I hope they are correct.
346
347		----------------------------------------------------------------------
348		Name: Allen Leung
349		Date: Fri Dec  8 19:23:26 EST 2000
350		Description:
351		Tag:  leunga-20001208-nowhere
352		Description:
353
354		Slight improvements to the 'nowhere' tool to handle OR-patterns,
355		to generate better error messages etc.  Plus a brief manual.
356
357		----------------------------------------------------------------------
358		Name: Lal George
359		Date: 2000/12/08 09:54:02 EST 2000
360		Tag: Release_110_31
361		Description:
362
363		- Version 110.31
364		----------------------------------------------------------------------
365		Name: Allen Leung
366		Date: Thu Dec  7 22:01:04 EST 2000
367		Tag:  leunga-20001207-cell-monster-hack
368		Description:
369
370		Major MLRISC internal changes.  Affect all clients.
371		Summary:
372
373		1.  Type CELLS.cell = int is now replaced by a datatype.
374		As a result, the old regmap is now gone.  Almost all interfaces
375		in MLRISC change as a consequence.
376
377		2.  A new brand version of machine description tool (v3.0) that generates
378		modules expecting the new interface.  The old version is removed.
379
380		3.  The RA interface has been further abstracted into two new functors.
381		RISC_RA and X86RA.  These functors have much simpler interfaces.
382		[See also directory MLRISC/demo.]
383
384		4.  Some other new source->source code generation tools are available:
385
386		a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
387		b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
388		I use this tool to generate the peephole optimizers---with the new
389		cell type changes, peephole rules are becoming difficult to write
390		without conditional pattern matching.
391
392		5.  More Intmap -> IntHashTable change.  Previous changes by Matthias didn't
393		cover the entire MLRISC source tree so many things broke.
394
395		6.  CM files have been moved to the subdirectory MLRISC/cm.
396		They are moved because there are a lot of them and they clutter up the
397		root dir.
398
399		7.  More detailed documentation to come...
400
401		NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
402		a makeml -rebuild first.  This is because of other other
403		changes that Matthias has made (see below).
404
405
406		----------------------------------------------------------------------
407		Name: Matthias Blume
408		Date: 2000/11/30 23:12:00 JST
409		Tag: blume-20001130-filereorg
410		Description:
411
412		Some manual updates and some file reorganizations in CM.
413
414		----------------------------------------------------------------------
415		Name: Matthias Blume
416		Date: 2000/11/24 17:45:00 JST
417		Tag: blume-20001124-link
418		Description:
419
420		Drastically improved link traversal code for the case that the dynamic
421		value was already loaded at bootstrap time.  As a result, CM and CMB
422		now both load blazingly fast -- even on a very slow machine.  Also,
423		memory consumption has been further reduced by this.
424
425		Warning: The format of the PIDMAP file has changed.  THerefore, to
426		bootstrap you have to do this:
427
428		1. Run CMB.make
429		2. Make a symbolic link for the boot directory:
430		ln -s sml.boot.ARCH-OS xxx
431		3. "Rebuild" the boot directory:
432		./makeml -boot xxx -rebuild sml ; rm xxx
433		4. Boot normally:
434		./makeml
435
436		----------------------------------------------------------------------
437		Name: Matthias Blume
438		Date: 2000/11/21 21:20:00 JST
439		Tag: blume-20001121-tools
440		Description:
441
442		Continued hacking on autoloading problem -- with success this time.
443		Also changed tool-plugin mechanism.  See new CM manual.
444
445		----------------------------------------------------------------------
446		Name: Matthias Blume
447		Date: 2000/11/19 14:30:00 JST
448		Tag:  blume-20001119-autoload
449		Description:
450
451		Some hacking to make autoloading faster.  Success for CMB, no success
452		so far for CM.  There is a reduced structure CM' that autoloads faster.
453		(This is a temporary, non-documented hack to be eliminated again when
454		the general problem is solved.)
455
456		----------------------------------------------------------------------
457		Name: Matthias Blume
458		Date: 2000/11/17 14:10:00 JST
459		Tag: blume-20001117-pickle-lib
460		Description:
461
462		1. Eliminated comp-lib.cm
463		2. Made pickle-lib.cm
464		3. Eliminated all uses of intset.sml (from comp-lib.cm)
465		4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
466		equivalent constructs from smlnj-lib.cm (INtHashTable).
467		5. Point 4. also goes for those uses of intmap.* in MLRISC.
468		Duplicated intmap modules thrown out.
469		6. Hunted down all duplicated SCC code and replaced it with
470		equivalent stuff (GraphSCCFn from smlnj-lib.cm).
471		7. Rewrote Feedback module.
472		8. Moved sortedlist.sml into viscomp-lib.cm.  Eventually it
473		should be thrown out and equivalent modules from smlnj-lib.cm
474		should be used (IntRedBlackSet, IntListSet, ...).
475
476		Confirmed that compiler compiles to fixpoint.
477
478		----------------------------------------------------------------------
479		Name: Allen Leung
480		Date: 2000/11/10 18:00:00
481		Tag: leunga-20001110-new-x86-fp
482
483		A new x86 floating point code generator has been added.
484		By default this is turned off.  To turn this on, do:
485
486		CM.autoload "$smlnj/compiler.cm";
487		Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
488
489		Changes:
490
491		1.  Changed FTAN to FPTAN so that the assembly output is correct.
492		2.  Changed the extension callback for FTANGENT to generate:
493
494		fptan
495		fstp  %st(0)
496		instead of
497		fptan
498		fstpl ftempmem
499
500		3.  Numerous assembly fixes for x86.
501
502		5.  Cleaned up the machine code output module x86/x86MC.sml and added
503		support for a whole bunch of instructions and addressing modes:
504
505		fadd/fsub/fsubr/fmul/fdiv/fdivr  %st, %st(n)
506		faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp  %st, %st(n)
507		fadd/fsub/fsubr/fmul/fdiv/fdivr  %st(n), %st
508		fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
509		fxch %st(n)
510		fld %st(n)
511		fst %st(n)
512		fst mem
513		fstp %st(n)
514		fucom %st(n)
515		fucomp %st(n)
516
517		All these are now generated when the fast fp mode is turned on.
518
519		6.  Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
520
521		----------------------------------------------------------------------
522		Name: Matthias Blume
523		Date: 2000/11/09 11:20:00 JST
524		Tag: blume-20001109-scc
525		Description:
526
527		Eliminated some code duplication:
528
529		1. Added "where" clause to GraphSCCFn in SML/NJ Library.
530		(Otherwise the functor is useless.)
531		2. Used GraphSCCFn where SCCUtilFun was used previously.
532		3. Got rid of SCCUtilFun (in comp-lib.cm).
533
534		----------------------------------------------------------------------
535		Name: Lal George
536		Date: 2000/11/06 09:02:21 EST 2000
537		Tag: Release_110_30
538		Description:
539
540		- Version 110.30
541		----------------------------------------------------------------------
542	Name: Matthias Blume	Name: Matthias Blume
543	Date: 2000/11/04 14:45:00	Date: 2000/11/04 14:45:00
544	Tag: blume-20001104-mlbuild	Tag: blume-20001104-mlbuild

---

Colored Diff Long Colored Diff Unidiff Context Diff Side by Side

Legend: Removed from v.716   changed lines   Added in v.810

---

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