Annotation of /sml/trunk/HISTORY

Parent Directory | Revision Log

---

Revision 672 - (view) (download)

1 :	dbm	570	This is the HISTORY file for the Yale SML/NJ CVS repository.
2 :
3 :			An entry should be made for _every_ commit to the repository.
4 :			The entries in this file will be used when creating the README
5 :			for new versions, so keep that in mind when writing the
6 :			description.
7 :
8 :			The form of an entry should be:
9 :
10 :			Name:
11 :			Date:
12 :			Tag: <post-commit CVS tag>
13 :			Description:
14 :	leunga	591	----------------------------------------------------------------------
15 :	blume	666	Name: Matthias Blume
16 :	blume	672	Date: 2000/06/19 13:40:00
17 :			Tag: blume-20000619-manual
18 :			Description:
19 :
20 :			CM manual and system/README update. This only covers the fact that
21 :			there are no more implicit anchors. (Path environments and the "bind"
22 :			option to "cm" have yet to be documented.)
23 :
24 :			----------------------------------------------------------------------
25 :			Name: Matthias Blume
26 :	blume	671	Date: 2000/06/19 11:05:00 JST
27 :			Tag: blume-20000619-chdir-bugfix
28 :			Description:
29 :
30 :			Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
31 :
32 :			----------------------------------------------------------------------
33 :			Name: Matthias Blume
34 :	blume	670	Date: 2000/06/18 22:00:10 JST
35 :			Tag: blume-20000618-implicit-anchors-really-gone
36 :			Description:
37 :
38 :			I updates the previous HISTORY entry where I forgot to mention that
39 :			implicit anchors are no longer with us.
40 :
41 :			The current update also gets rid of the (now useless) controller
42 :			CM.Control.implicit_anchors.
43 :
44 :			----------------------------------------------------------------------
45 :			Name: Matthias Blume
46 :	blume	666	Date: 2000/06/16 17:30:00 JST
47 :			Tag: blume-20000616-anchorenv
48 :			Description:
49 :
50 :			This patch implements the long anticipated (just kidding :) "anchor
51 :			environment" mechanism. In the course of doing this, I also
52 :			re-implemented CM's internal "SrcPath" module from scratch. The new
53 :			one should be more robust in certain boundary cases. In any case, it
54 :			is a lot cleaner than its predecessor (IMHO).
55 :
56 :			This time, although there is yet another boot file format change, I
57 :			kept the unpickler backward-compatible. As a result, no new bootfiles
58 :			are necessary and bootstrapping is straightforward. (You cannot read
59 :			new bootfiles into an old system, but the other way around is no
60 :			problem.)
61 :
62 :			Visible changes:
63 :
64 :	blume	670	** 0. Implicit path anchors (without the leading $-symbol) are no
65 :			longer recognized at all. This means that such path names are not
66 :			illegal either. For example, the name basis.cm simply refers to a
67 :			local file called "basis.cm" (i.e, the name is an ordinary path
68 :			relative to .cm-files directory). Or, to put it differently, only
69 :			names that start with $ are anchored paths.
70 :	blume	666
71 :	blume	670	** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
72 :			vanished.
73 :
74 :			John (Reppy) had critizised this as soon as I originally proposed and
75 :	blume	666	implemented it, but at that time I did not really deeply believe
76 :			him. :) Now I came full-circle because I need the $<singlearc> syntax
77 :			in another place where it cannot be seen as an abbreviation for
78 :			$/<singlearc>. To avoid the confusion, $<singlearc> now means what it
79 :			seems to mean (i.e., it "expands" into the corresponding anchor
80 :			value).
81 :
82 :			However, when paths are used as members in CM description files, it
83 :			continues to be true that there must be at least another arc after the
84 :			anchor. This is now enforced separately during semantic analysis
85 :	blume	670	(i.e., from a lexical/syntactical point of view, the notation is ok.)
86 :	blume	666
87 :			** 2. The "cm" class now accepts an option "bind". The option's value
88 :			is a sub-option list of precisely two items -- one labeled "anchor"
89 :			and the other one labeled "value". As you might expect, "anchor" is
90 :			used to specify an anchor name to be bound, and "value" specifies what
91 :			the anchor is being bound to.
92 :
93 :			The value must be a directory name and can be given in either standard
94 :			syntax (including the possibility that it is itself an anchored path)
95 :			or native syntax.
96 :
97 :			Examples:
98 :
99 :			foo.cm (bind:(anchor:bar value:$mystuff/bar))
100 :			lib.cm (bind:(anchor:a value:"H:\\x\\y\\z")) (* only works under windows *)
101 :
102 :			and so on.
103 :
104 :			The meaning of this is that the .cm-file will be processed with an
105 :			augmented anchor environment where the given anchor(s) is/are bound to
106 :			the given values(s).
107 :
108 :			The rationale for having this feature is this: Suppose you are trying
109 :			to use two different (already stable) libraries a.cm and b.cm (that
110 :			you perhaps didn't write yourself). Further, suppose each of these
111 :			two libraries internally uses its own auxiliary library $aux/lib.cm.
112 :			Normally you would now have a problem because the anchor "lib" can not
113 :			be bound to more than one value globally. Therefore, the project that
114 :			uses both a.cm and b.cm must locally redirect the anchor to some other
115 :			place:
116 :
117 :			a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
118 :			b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
119 :
120 :			This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
121 :			/usr/lib/smlnj/b-stuff/aux.cm, respectively.
122 :
123 :			Hard-wiring path names is a bit inflexible (and CM will verbosely warn
124 :			you when you do so at the time of CM.stabilize). Therefore, you can
125 :			also use an anchored path as the value:
126 :
127 :			a.cm (bind:(anchor:lib value:$a-lib))
128 :			b.cm (bind:(anchor:lib value:$b-lib))
129 :
130 :			Now you can globally configure (using the usual CM.Anchor.anchor or
131 :			pathconfig machinery) bindings for "a-lib" and "b-lib". Since "lib"
132 :			itself is always locally bound, setting it globally is no longer
133 :			meaningful or necessary (but it does not hurt either). In fact, "lib"
134 :			can still be used as a global anchor for separate purposes. As a
135 :			matter of fact, one can locally define "lib" in terms of a global
136 :			"lib":
137 :
138 :			a.cm (bind:(anchor:lib value:$lib/a))
139 :			b.cm (bind:(anchor:lib value:$lib/b))
140 :
141 :			** 3: The encoding of path names has changed. This affects the way
142 :			path names are shown in CM's progress report and also the internal
143 :			protocol encoding used for parallel make.
144 :
145 :			The encoding now uses one or more ':'-separated segments. Each
146 :			segments corresponds to a file that has been specified relative to the
147 :			file given by its preceding segment. The first segment is either
148 :			relative to the CWD, absolute, or anchored. Each segment itself is
149 :			basically a Unix pathname; all segments but the first are relative.
150 :
151 :			Example:
152 :
153 :			$foo/bar/baz.cm:a/b/c.sml
154 :
155 :			This path denotes the file bar/a/b/c.sml relative to the directory
156 :			denoted by anchor "foo". Notice that the encoding also includes
157 :			baz.cm which is the .cm-file that listed a/b/c.sml. As usual, such
158 :			paths are resolved relative to the .cm-files directory, so baz.cm must
159 :			be ignored to get the "real" pathname.
160 :
161 :			To make this fact more obvious, CM puts the names of such "virtual
162 :			arcs" into parentheses when they appear in progress reports. (No
163 :			parentheses will appear in the internal protocol encoding.) Thus,
164 :			what you really see is:
165 :
166 :			$foo/bar/(baz.cm):a/b/c.sml
167 :
168 :			I find this notation to be much more informative than before.
169 :
170 :			Another new feature of the encoding is that special characters
171 :			including parentheses, colons, (back)slashes, and white space are
172 :			written as \ddd (where ddd is the decimal encoding of the character).
173 :
174 :	blume	670	*** The CM manual still needs to be updated.
175 :
176 :	blume	666	----------------------------------------------------------------------
177 :	leunga	660	Name: Allen Leung
178 :			Date: 2000/06/15 00:38:00
179 :			Tag: leunga-20000615-x86-peephole
180 :
181 :			x86 Peephole fix by Fermin. Affects c-- and moby only.
182 :
183 :			----------------------------------------------------------------------
184 :	blume	658	Name: Matthias Blume
185 :			Date: 2000/06/12 11:40:00
186 :			Tag: blume-20000612-parmakefix
187 :			Description:
188 :
189 :			More cleanup after changing the file naming scheme: This time I
190 :			repaired the parallel make mechanism for CMB.make which I broke earlier.
191 :
192 :			----------------------------------------------------------------------
193 :	leunga	657	Name: Allen Leung
194 :			Date: 2000/06/09 01:25:00
195 :			Tag: leunga-20000609-various
196 :
197 :			None of these things should affect normal SML/NJ operations
198 :
199 :			1. Peephole improvements provided by Fermin (c--)
200 :			2. New annotation DEFUSE for adding extra dependence (moby)
201 :			3. New X86 LOCK instructions (moby)
202 :			4. New machine description language for reservation tables (scheduling)
203 :			5. Fixes to various optimization/analysis modules (branch chaining, dominator
204 :			trees etc.)
205 :			6. I've changed the CM files so that they can work with versions
206 :			110.0.6, 110.25 and 110.28
207 :
208 :			----------------------------------------------------------------------
209 :	blume	652	Name: Matthias Blume
210 :	blume	656	Date: 2000/06/09 12:40:00
211 :			Tag: blume-20000609-log
212 :			Description:
213 :
214 :			- Removed all(?) remaining RCS Log entries from sources.
215 :
216 :			- Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
217 :			anchored paths).
218 :
219 :			----------------------------------------------------------------------
220 :			Name: Matthias Blume
221 :	blume	653	Date: 2000/06/07 17:00:00 JST
222 :			Tag: blume-20000607-no-implicit-anchors
223 :			Description:
224 :
225 :			1. This update changes the default setting for
226 :			CM.Control.implicit_anchors from true to false. This means that
227 :			implicit anchors are no longer permitted by default. I also tried to
228 :			make sure that nothing else still relies on implicit anchors.
229 :			(This is the next step on the schedule towards a CM that does not even
230 :			have the notion of implicit anchors anymore.)
231 :
232 :			2. More CM manual updates.
233 :
234 :			3. I managed to track down and fix the pickling bug I mentioned last
235 :			time. Because of the previously existing workaround, this entails no
236 :			immediate practical changes.
237 :
238 :			----------------------------------------------------------------------
239 :			Name: Matthias Blume
240 :	blume	652	Date: 2000/06/06 11:15:00 JST
241 :			Tag: blume-20000606-lazierpickle
242 :			Description:
243 :
244 :			!!!! NEW BOOT FILES !!!!
245 :
246 :			* The main purpose of this update is to make library pickles lazier in
247 :			order to reduce the initial space penalty for autoloading a library.
248 :			As a result, it is now possible to have $smlnj/compiler.cm
249 :			pre-registered. This should take care of the many complaints or
250 :			inquiries about missing structure Compiler. This required changes to
251 :			CM's internal data structures and small tweaks to some algorithms.
252 :
253 :			As a neat additional effect, it is no longer necessary (for the sake
254 :			of lean heap image files) to distinguish between a "minimal" CM and a
255 :			"full" CM. Now, there is only one CM (i.e., the "full" version:
256 :			$smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
257 :			interactive top level. ($smlnj/cm/minimal.cm is gone.)
258 :
259 :			To make the life of compiler-hackers easier, "makeml" now also
260 :			pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm). In other
261 :			words, after you bootstrap a new sml for the first time, you will not
262 :			have to autoload $smlnj/cmb.cm again afterwards. (The first time
263 :			around you will still have to do it, though.)
264 :
265 :			* A second change consists of major updates to the CM manual. There
266 :			are now several appendices with summary information and also a full
267 :			specification of the CM description file syntax.
268 :
269 :			* In directory src/system I added the script "allcross". This script
270 :			invokes sml and cross-compiles the compiler for all supported
271 :			architectures. (Useful when providing a new set of boot files.)
272 :
273 :			* There seems to be a latent bug in my "lazy pickles" mechanism. I
274 :			added a small tweak to pickle-util.sml to work around this problem,
275 :			but it is not a proper fix yet. I will investigate further. (The
276 :			effect of the bug was an inflation of library pickle size.)
277 :
278 :			* Version number increased to 110.28.1 (to avoid compatibility problems).
279 :
280 :			----------------------------------------------------------------------
281 :	leunga	646	Name: Allen Leung
282 :	leunga	648	Date: 2000/05/25 17:28 EDT
283 :			Tag: leunga-20000525-ra
284 :			Description:
285 :
286 :			Fixed a bug in freezing phase of the register allocator.
287 :
288 :			----------------------------------------------------------------------
289 :			Name: Allen Leung
290 :	leunga	647	Date: 2000/05/15 22:53 EDT
291 :	leunga	646	Tag: leunga-20000515-alpha-x86-ra
292 :			Description:
293 :
294 :			1. Alpha
295 :
296 :			Slight cleanup. Removed the instruction SGNXL
297 :
298 :			2. X86
299 :
300 :			Added the following instructions to the instruction set:
301 :
302 :			ROLx, RORx,
303 :			BTx, BTSx, BTLx, BTRx,
304 :			XCHGx, and variants with the LOCK prefix
305 :
306 :			3. Register Allocation
307 :
308 :			The module ra-rewrite-with-renaming has been improved.
309 :
310 :	leunga	647	These have no effect on SML/NJ.
311 :
312 :	leunga	646	----------------------------------------------------------------------
313 :	blume	642	Name: Matthias Blume
314 :	blume	645	Date: 2000/05/15 16:20:00 JST
315 :			Tag: blume-20000515-lightrebuild
316 :			Description:
317 :
318 :			1. I added an alternative to "-rebuild" to "makeml". The difference is
319 :			that prior to calling CMB.make' the CM-variable "LIGHT" will be
320 :			defined. In effect, the command will not build any cross-compiler
321 :			backends and therefore finish more quickly.
322 :
323 :			The "fixpt" script also takes a "-light" switch to be able to use
324 :			this new facility while compiling for a fixpoint.
325 :
326 :			2. I replaced all mentions of anchored paths in group owner specifications
327 :			with simple relative paths (usually starting with "..").
328 :			The rationale is that a library's internal workings should not be
329 :			compromised by the lack of some anchor. (An anchor is necessary
330 :			for someone who wants to refer to the library by an anchored path,
331 :			but it should not be necessary to build the same library in the first
332 :			place.)
333 :
334 :			3. I changed the way CM's tool mechanism determines the shell command
335 :			string used for things like ml-yacc etc. so that it does not break
336 :			when CM.Control.implicit_anchors is turned off.
337 :
338 :			----------------------------------------------------------------------
339 :			Name: Matthias Blume
340 :	blume	644	Date: 2000/05/12 18:20:00 JST
341 :			Tag: blume-20000512-ml-build
342 :			Description:
343 :
344 :			Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
345 :			getting installed properly (by config/install.sh).
346 :
347 :			----------------------------------------------------------------------
348 :			Name: Matthias Blume
349 :	blume	643	Date: 2000/05/12 17:30:00 JST
350 :			Tag: blume-20000512-anchors
351 :			Description:
352 :
353 :			!!! NEW BOOT FILES !!!
354 :
355 :			This change is in preparation of fading out support for "implicitly
356 :			anchored path names". I went through all sources and used the
357 :			explicit (and relatively new) $-notation. See system/README and the
358 :			CM manual for more info on this.
359 :
360 :			I also modified the anchoring scheme for some things such as "smlnj",
361 :			"MLRISC", "cm", etc. to take advantage of the fact that explicit
362 :			anchors are more expressive: anchor name and first arc do not have to
363 :			coincide. This entails the following user-visible change:
364 :
365 :			You have to write $smlnj/foo/bar instead of smlnj/foo/bar. In
366 :			particular, when you fire up sml with a command-line argument, say,
367 :			e.g.:
368 :
369 :			sml '$smlnj/cmb.cm'
370 :
371 :			At the ML toplevel prompt:
372 :
373 :			CM.autoload "$smlnj/cmb.cm";
374 :
375 :			There is also a new controller in CM.Control that can be used to turn
376 :			off all remaining support for implicit anchors by saying:
377 :
378 :			CM.autoload "$smlnj/
379 :			#set CM.Control.implicit_anchors false;
380 :
381 :			This causes CM to reject implicitly anchored paths. This is (for the
382 :			time being) less permissive than the "final" version where there will
383 :			be no more such implicit anchors and relative paths will be just that:
384 :			relative.
385 :
386 :			The next step (version after next version?) will be to make the
387 :			default for CM.Control.implicit_anchors false. After the dust has
388 :			settled, I can then produce the "final" version of this...
389 :
390 :			Note: Since bootstrapping is a bit tricky, I provided new boot files.
391 :
392 :			----------------------------------------------------------------------
393 :			Name: Matthias Blume
394 :	blume	642	Date: 2000/05/11 16:30:00 JST
395 :			Tag: blume-20000511-sources
396 :			Description:
397 :
398 :			The main change is that I added function CM.sources as a generalized
399 :			version of the earlier CM.makedepend. This entails the following
400 :			additional changes:
401 :
402 :			- CM.makedepend has been dropped.
403 :
404 :			- CM manual has been updated.
405 :
406 :			- TOOLS signature and API have been changed.
407 :
408 :			----------------------------------------------------------------------
409 :	leunga	640	Name: Allen Leung
410 :	leunga	641	Date: 2000/05/10 21:17 EDT
411 :			Tag: leunga-20000510-moby-c--ssa
412 :			Description:
413 :
414 :			Various bug fixes and new features for C--, Moby and MLRISC optimizations.
415 :			None of these affect SML/NJ.
416 :
417 :			1. Register Allocation
418 :
419 :			a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
420 :			This module tries to remove local (i.e. basic block level) redundancies
421 :			during spilling.
422 :
423 :			b. A new framework for performing region based register allocation.
424 :			Not yet entirely functional.
425 :
426 :			2. X86
427 :
428 :			a. DefUse for POP was missing the stack pointer [found by Lal]
429 :			b. Reload for CALL was incorrect in X86Spill [found by John]
430 :			c. Various fixes in X86Spill so that it can be used correctly for
431 :			the new spilling module.
432 :
433 :			3. SSA/IR
434 :
435 :			a. New module ir/dj-dataflow.sml implements elimination based
436 :			data flow analysis.
437 :
438 :			4. MLRiscGen
439 :
440 :			a. Fix for gc type annotation
441 :
442 :			5. MDGen
443 :
444 :			Various fixes for machine description -> ml code translation. For ssa
445 :			only.
446 :
447 :			----------------------------------------------------------------------
448 :			Name: Allen Leung
449 :	leunga	640	Date: 2000/05/08 22:17 EDT
450 :			Tag: leunga-20000508-labexp
451 :			Description:
452 :
453 :			Fermin has found a few assembly problems with constant expressions
454 :			generated in LabelExp. Mostly, the problems involve extra parentheses,
455 :			which choke on dumb assemblers. This is his fix.
456 :
457 :			----------------------------------------------------------------------
458 :	dbm	635	Name: Dave MacQueen
459 :			Date: 2000/04/09 14:00 EDT
460 :			Tag: dbm-20000502-Version_110_28
461 :			Description:
462 :
463 :			1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
464 :
465 :			2. Updated config/version to 110.28
466 :
467 :			3. Updated config/srcarchiveurl
468 :
469 :			3. New boot files!
470 :			ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
471 :
472 :			----------------------------------------------------------------------
473 :	blume	634	Name: Matthias Blume
474 :			Date: 2000/05/01 19:05:00 JST
475 :			Tag: blume-20000501-noweb
476 :			Description:
477 :
478 :			A new noweb tool has been added. The existing system is entirely
479 :			unaffected by this, but some CM users have asked for renewed noweb
480 :			support. Everything is documented in the CM manual.
481 :
482 :			New (plugin) libraries:
483 :
484 :			noweb-tool.cm
485 :			nw-ext.cm
486 :
487 :			----------------------------------------------------------------------
488 :	dbm	633	Name: Dave MacQueen
489 :			Date: 2000/04/30 12:40PM EDT
490 :			Tag: dbm-20000430-bug_fixes
491 :			Description:
492 :
493 :			1. Fix for bug 1498
494 :			smlnj/src/system/Basis/Implementation/Unsafe/object.sig
495 :			smlnj/src/system/Basis/Implementation/Unsafe/object.sml
496 :			added toRealArray function
497 :			smlnj/src/compiler/MiscUtil/print/ppobj.sml
498 :			added check for tag Obj.RealArray to array printing case in ppObj
499 :
500 :			2. Fix for bug 1510
501 :			smlnj/src/compiler/Semant/types/typesutil.sml
502 :			fixed definition of dummyargs (used by equalTycon) so that
503 :			dummy args are distinct types
504 :
505 :			----------------------------------------------------------------------
506 :	blume	629	Name: Matthias Blume
507 :	blume	632	Date: 2000/04/30 01:00:00 JST
508 :			Tag: blume-20000430-versions
509 :			Description:
510 :
511 :			1. CM version numbering added. This is an implementation of Lal's
512 :			proposal for adding version numbers and version checking to .cm
513 :			files. Lal said that his proposal was just that -- a proposal.
514 :			For the time being I went ahead and implemented it so that people
515 :			can comment on it. Everything is completely backward-compatible
516 :			(except for the stable library format, i.e., new bootfiles!).
517 :
518 :			As usual, see the CM manual for details.
519 :
520 :			2. An alternative syntax for anchored paths has been implemented.
521 :			Dave has recently voiced the same concerns that I had when I did
522 :			this, so there should be some support. My take is that eventually
523 :			I will let support for the current syntax (where anchors are
524 :			"implicit") fade out in favor of the new, explicit syntax.
525 :			In order to be backward-compatible, both old and new syntax are
526 :			currently supported.
527 :
528 :			Again, see the CM manual for details.
529 :
530 :			3. Parallel make is trying to be slightly smarter: When the master
531 :			process finds a "bottleneck", i.e., when there is only one
532 :			compilation unit that can be compiled and everybody else is
533 :			waiting on it, then it will simply compile it directly instead
534 :			of clumsily telling one of the slaves to do it.
535 :
536 :			4. Support for "unsharing" added. This is necessary in order to be
537 :			able to have two different versions of the same library running
538 :			at the same time (e.g., for trying out a new MLRISC while still
539 :			having the old MLRISC linked into the current compiler, etc.)
540 :			See the CM manual.
541 :
542 :			5. Simple "makedepend" functionality added for generating Makefile
543 :			dependency information. (This is rather crude at the moment.
544 :			Expect some changes here in the future.)
545 :
546 :			6. ".fun" added as a recognized suffix for ML files. Also documented
547 :			explicitly in the manual that the fallback behavior (unknown suffix
548 :			-> ML file) is not an official feature!
549 :
550 :			7. Small changes to the pickler for stable libraries.
551 :
552 :			8. Several internal changes to CM (for cleanup/improvement).
553 :
554 :
555 :			!!!! NEW BINFILES !!!!
556 :
557 :			----------------------------------------------------------------------
558 :			Name: Matthias Blume
559 :	blume	631	Date: 2000/04/28 17:30:00 JST
560 :			Tag: blume-20000428-pathconfig
561 :			Description:
562 :
563 :			1. I changed config/install.sh to remove duplicate entries from the
564 :			lib/pathconfig file at the end. Moreover, the final version of
565 :			lib/pathconfig is sorted alphabetically. The same (sorting) is done
566 :			in src/system/installml.
567 :
568 :			2. The config/install.sh script now consistently uses relative
569 :			pathnames in lib/pathconfig whenever the anchor is in the lib
570 :			directory. (So far this was true for the libraries that come
571 :			pre-compiled and bundled as part of the bootfiles but not for
572 :			libraries that are compiled by the script itself.)
573 :
574 :			----------------------------------------------------------------------
575 :			Name: Matthias Blume
576 :	blume	629	Date: 2000/04/26 13:10:00 JST
577 :			Tag: blume-20000426-fun_suffix
578 :			Description:
579 :
580 :			Added ".fun" as a recognized file name suffix (for ML code).
581 :
582 :			----------------------------------------------------------------------
583 :	leunga	628	Name: Allen Leung
584 :			Date: 2000/04/25 17:00:00 EST
585 :			Tag: leunga-20000425-alpha-ra
586 :			Description:
587 :
588 :			1. Alpha
589 :
590 :			PSEUDOARITH was missing in AlphaRewrite. This causes an endless loop
591 :			in C--.
592 :
593 :			2. RA
594 :
595 :			Added a flag "ra-dump-size" to print out the size of the flowgraph
596 :			and the interference graph.
597 :
598 :			----------------------------------------------------------------------
599 :	dbm	627	Name: Dave MacQueen
600 :			Date: 2000/04/25/
601 :			Tag: dbm-20000425-mlyacc_doc_examples
602 :			Description:
603 :			Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
604 :			Updated all three examples in src/ml-yacc/examples to run
605 :			under 110.* using CM.make.
606 :
607 :			----------------------------------------------------------------------
608 :	leunga	624	Name: Allen Leung
609 :			Date: 2000/04/20 23:04:00 EST
610 :			Tag: leunga-20000420-ssa-c---stuff
611 :			Description:
612 :
613 :			This update synchronizes my repository with Yale's. Most of these
614 :			changes, however, do not affect SML/NJ at all (the RA is an exception).
615 :
616 :			1. Register Allocator
617 :
618 :			a. An improvement in the interference graph construction:
619 :			Given a copy
620 :
621 :			s <- t
622 :
623 :	leunga	628	no interference edge between s and t is added for this definition of s.
624 :	leunga	624
625 :			b. I've added two new spill heuristic modules that Fermin and I developed
626 :			(in the new library RA.cm). These are unused in SML/NJ but maybe
627 :			useful for others (Moby?)
628 :
629 :			2. X86
630 :
631 :			a. Various fixes in the backend provided by Fermin [C--] and Lal.
632 :
633 :			3. Alpha
634 :
635 :			a. Added the BSR instruction and code generation that goes with it [C--]
636 :			b. Other fixes too numerous to recount provided by Fermin [C--]
637 :
638 :			4. Regmaps
639 :
640 :			a. The regmaps are not initialized with the identity physical bindings
641 :			at creation time. This is unneeded.
642 :
643 :			5. MLRISC Optimizations
644 :
645 :			a. The DJ-Graph module can now compute the iterated dominance frontiers
646 :			intersects with liveness incrementally in linear time! Woohoo!
647 :			This is now used in my new SSA construction algorithm.
648 :
649 :			b. THe branch reorganization module is now smarter about linear chains of
650 :			basic blocks.
651 :
652 :
653 :			----------------------------------------------------------------------
654 :	blume	615	Name: Matthias Blume
655 :			Date: 2000/04/12 13:52:00 JST
656 :			Tag: blume_main_v110p27_1
657 :			Description:
658 :
659 :			Changed install.sh script to handle archive files without version number
660 :			and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
661 :			name of the boot file archive.
662 :
663 :			----------------------------------------------------------------------
664 :	dbm	607	Name: Dave MacQueen
665 :			Date: 2000/04/09 14:00 EDT
666 :			Tag: dbm-20000410-Version_110_27
667 :			Description:
668 :
669 :			1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
670 :
671 :			2. Updated src/config/version to 110.27
672 :
673 :			3. New boot files!
674 :
675 :			----------------------------------------------------------------------
676 :	leunga	606	Name: Allen Leung
677 :			Date: 2000/04/09 19:09:00 EST
678 :			Tag: leunga-20000409-misc
679 :			Description:
680 :
681 :			1. Yet another fix for x86 assembly for idivl, imull, mull and friends.
682 :
683 :			2. Miscellaneous improvements to MLRISC (unused in sml/nj)
684 :
685 :			----------------------------------------------------------------------
686 :	monnier	605	Name: Stefan
687 :			Date: 2000/04/07 10:00:00 EDT
688 :			Tag: monnier-20000406-branch-handling
689 :			Description:
690 :
691 :			Improved handling of branches (mostly those generated from
692 :			polymorphic equality), removed switchoff and changed the
693 :			default optimization settings (more cpsopt and less flintopt).
694 :
695 :			----------------------------------------------------------------------
696 :	leunga	601	Name: Allen Leung
697 :	leunga	602	Date: 2000/04/06 01:30:00 EST
698 :			Tag: leunga-20000406-peephole-x86-SSA-2
699 :			Description:
700 :
701 :			Forgot a few files.
702 :
703 :			----------------------------------------------------------------------
704 :			Name: Allen Leung
705 :	leunga	601	Date: 2000/04/06 00:36:00 EST
706 :			Tag: leunga-20000406-peephole-x86-SSA
707 :			Description:
708 :
709 :			1. New Peephole code
710 :
711 :			2. Minor improvement to X86 instruction selection
712 :
713 :			3. Various fixes to SSA and machine description -> code translator
714 :
715 :			----------------------------------------------------------------------
716 :	blume	595	Name: Matthias Blume
717 :			Date: 2000/04/05 12:30:00 JST
718 :			Tag: blume_main_v110p26p2_3
719 :			Description:
720 :
721 :			This update just merges three minor cosmetic updates to CM's sources
722 :			to get ready for the 110.27 code freeze on Friday. No functionality
723 :			has changed.
724 :
725 :			----------------------------------------------------------------------
726 :	leunga	593	Name: Allen Leung
727 :	leunga	594	Date: 2000/04/04 19:39:00 EST
728 :			Tag: leunga-20000404-x86-asm
729 :			Description:
730 :
731 :			1. Fixed a problem in X86 assembly.
732 :
733 :			Things like
734 :
735 :			jmp %eax
736 :			jmp (%eax)
737 :
738 :			should be output as
739 :
740 :			jmp *%eax
741 :			jmp *(%eax)
742 :
743 :			2. Assembly output
744 :
745 :			Added a new flag
746 :
747 :			"asm-indent-copies" (default to false)
748 :
749 :			When this flag is on, parallel copies will be indented an extra level.
750 :
751 :			----------------------------------------------------------------------
752 :			Name: Allen Leung
753 :	leunga	593	Date: 2000/04/04 03:18:00 EST
754 :			Tag: leunga-20000404-C--Moby
755 :			Description:
756 :
757 :			All of these fixes are related to C--, Moby, and my own optimization
758 :			stuff; so they shouldn't affect SML/NJ.
759 :
760 :			1. X86
761 :
762 :			Various fixes related floating point, and extensions.
763 :
764 :			2. Alpha
765 :
766 :			Some extra patterns related to loads with signed/zero extension
767 :			provided by Fermin.
768 :
769 :			3. Assembly
770 :
771 :	monnier	659	When generating assembly, resolve the value of client defined constants,
772 :	leunga	593	instead of generating symbolic values. This is controlled by the
773 :			new flag "asm-resolve-constants", which is default to true.
774 :
775 :			4. Machine Descriptions
776 :
777 :			a. The precedence parser was slightly broken when parsing infixr symbols.
778 :			b. The type generalizing code had the bound variables reversed, resulting
779 :			in a problem during arity raising.
780 :			c. Various fixes in machine descriptions.
781 :
782 :			----------------------------------------------------------------------
783 :	blume	592	Name: Matthias Blume
784 :			Date: 2000/04/03 16:05:00 JST
785 :			Tag: blume_main_v110p26p2_2
786 :			Description:
787 :
788 :			I eliminated coreEnv from compInfo. Access to the "Core" structure is
789 :			now done via the ordinary static environment that is context to each
790 :			compilation unit.
791 :
792 :			To this end, I arranged that instead of "structure Core" as "structure
793 :			_Core" is bound in the pervasive environment. Core access is done via
794 :	monnier	659	_Core (which can never be accidentally rebound because _Core is not a
795 :	blume	592	legal surface-syntax symbol).
796 :
797 :			The current solution is much cleaner because the core environment is
798 :			now simply part of the pervasive environment which is part of every
799 :			compilation unit's context anyway. In particular, this eliminates all
800 :			special-case handling that was necessary until now in order to deal
801 :			with dynamic and symbolic parts of the core environment.
802 :
803 :			Remaining hackery (to bind the "magic" symbol _Core) is localized in the
804 :	monnier	659	compilation manager's bootstrap compiler (actually: in the "init group"
805 :	blume	592	handling). See the comments in src/system/smlnj/init/init.cmi for
806 :			more details.
807 :
808 :			I also tried to track down all mentions of "Core" (as string argument
809 :			to Symbol.strSymbol) in the compiler and replaced them with a
810 :			reference to the new CoreSym.coreSym. Seems cleaner since the actual
811 :			name appears in one place only.
812 :
813 :			Binfile and bootfile format have not changed, but the switchover from
814 :			the old "init.cmi" to the new one is a bit tricky, so I supplied new
815 :			bootfiles anyway.
816 :
817 :			----------------------------------------------------------------------
818 :	leunga	591	Name: Allen Leung
819 :			Date: 2000/04/02 21:17:00 EST
820 :			Tag: leunga-20000402-mltree
821 :			Description:
822 :	leunga	585
823 :	leunga	591	1. Renamed the constructor CALL in MLTREE by popular demand.
824 :			2. Added a bunch of files from my repository. These are currently
825 :			used by other non-SMLNJ backends.
826 :
827 :	leunga	576	----------------------------------------------------------------------
828 :	leunga	590	Name: Allen Leung
829 :			Date: 2000/03/31 21:15:00 EST
830 :			Tag: leunga-20000331-aliasing
831 :			Description:
832 :
833 :			This update contains a rewritten (and hopefully more correct) module
834 :			for extracting aliasing information from CPS.
835 :
836 :			To turn on this feature:
837 :
838 :			Compiler.Control.CG.memDisambiguate := true
839 :
840 :			To pretty print the region information with assembly
841 :
842 :			Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
843 :
844 :			To control how many levels of aliasing information are printed, use:
845 :
846 :			Compiler.Control.MLRISC.getInt "points-to-show-level" := n
847 :
848 :			The default of n is 3.
849 :
850 :			----------------------------------------------------------------------
851 :	dbm	589	Name: David MacQueen
852 :			Date: 2000/03/31 11:15:00 EST
853 :			Tag: dbm-20000331-runtime_fix
854 :			Description:
855 :
856 :			This update contains:
857 :
858 :			1. runtime/c-lib/c-libraries.c
859 :			includes added in revision 1.2 caused compilation errors on hppa-hpux
860 :
861 :			2. fix for bug 1556
862 :			system/Basis/Implementation/NJ/internal-signals.sml
863 :
864 :			----------------------------------------------------------------------
865 :	blume	587	Name: Matthias Blume
866 :	blume	588	Date: 2000/03/31 18:00:00 JST
867 :			Tag: blume_main_v110p26p2_1
868 :			Description:
869 :
870 :			This update contains:
871 :
872 :			1. A small change to CM's handling of stable libraries:
873 :			CM now maintains one "global" modmap that is used for all stable
874 :			libraries. The use of such a global modmap maximizes sharing and
875 :			minimizes the need for re-traversing parts of environments during
876 :			modmap construction. (However, this has minor impact since modmap
877 :			construction seems to account for just one percent or less of total
878 :			compile time.)
879 :
880 :			2. I added a "genmap" phase to the statistics. This is where I got the
881 :			"one percent" number (see above).
882 :
883 :			3. CM's new tool parameter mechanism just became _even_ better. :)
884 :			- The parser understands named parameters and recursive options.
885 :			- The "make" and "shell" tools use these new features.
886 :			(This makes it a lot easier to cascade these tools.)
887 :			- There is a small syntax change: named parameters use a
888 :
889 :			<name> : ( <option> ... ) or
890 :			<name> : <string>
891 :
892 :			syntax. Previously, named parameters were implemented in an
893 :			ad-hoc fashion by each tool individually (by parsing strings)
894 :			and had the form
895 :
896 :			<name>=<string>
897 :
898 :			See the CM manual for a full description of these issues.
899 :
900 :			----------------------------------------------------------------------
901 :			Name: Matthias Blume
902 :	blume	587	Date: 2000/03/30 18:00:00 JST
903 :			Tag: blume_main_v110p26p2_0
904 :			Description:
905 :
906 :			!!!!! WARNING !!!!!!
907 :			!! New binfiles !!
908 :			!!!!!!!!!!!!!!!!!!!!
909 :
910 :			This update contains:
911 :
912 :			1. Moderate changes to CM:
913 :
914 :			- Changes to CM's tools mechanism. In particular, it is now possible
915 :			to have tools that accept additional "command line" parameters
916 :			(specified in the .cm file at each instance where the tool's class is
917 :			used).
918 :
919 :	monnier	659	This was done to accommodate the new "make" and "shell" tools which
920 :			facilitate fairly seamless hookup to portions of code managed using
921 :	blume	587	Makefiles or Shell scripts.
922 :
923 :			There are no classes "shared" or "private" anymore. Instead, the
924 :			sharing annotation is now a parameter to the "sml" class.
925 :
926 :			There is a bit of generic machinery for implementing one's own
927 :			tools that accept command-line parameters. However, I am not yet fully
928 :			satisfied with that part, so expect changes here in the future.
929 :
930 :			All existing tools are described in the CM manual.
931 :
932 :	monnier	659	- Slightly better error handling. (CM now suppresses many followup
933 :	blume	587	error messages that tended to be more annoying than helpful.)
934 :
935 :			2. Major changes to the compiler's static environment data structures.
936 :
937 :			- no CMStaticEnv anymore.
938 :			- no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
939 :			but it is called Environment), no conversions between different
940 :			kinds of static environments
941 :
942 :			- There is still a notion of a "modmap", but such modmaps are generated
943 :			on demand at the time when they are needed. This sounds slow, but I
944 :			sped up the code that generates modmaps enough for this not to lead to
945 :			a slowdown of the compiler (at least I didn't detect any).
946 :
947 :			- To facilitate rapid modmap generation, static environments now
948 :			contain an (optional) "modtree" structure. Modtree annotations are
949 :			constructed by the unpickler during unpickling. (This means that
950 :			the elaborator does not have to worry about modtrees at all.)
951 :			Modtrees have the advantage that they are compositional in the same
952 :			way as the environment data structure itself is compositional.
953 :			As a result, modtrees never hang on to parts of an environment that
954 :			has already been rendered "stale" by filtering or rebinding.
955 :
956 :			- I went through many, many trials and errors before arriving at the
957 :			current solution. (The initial idea of "linkpaths" did not work.)
958 :			But the result of all this is that I have touched a lot of files that
959 :			depend on the "modules" and "types" data structures (most of the
960 :			elaborator). There were a lot of changes during my "linkpath" trials
961 :			that could have been reverted to their original state but weren't.
962 :			Please, don't be too harsh on me for messing with this code a bit more
963 :			than what was strictly necessary... (I _did_ resist the tempation
964 :			of doing any "global reformatting" to avoid an untimely death at
965 :			Dave's hands. :)
966 :
967 :			- One positive aspect of the previous point: At least I made sure that
968 :			all files that I touched now compile without warnings (other than
969 :			"polyEqual").
970 :
971 :			- compiler now tends to run "leaner" (i.e., ties up less memory in
972 :			redundant modmaps)
973 :
974 :			----------------------------------------------------------------------
975 :	leunga	580	Name: Allen Leung
976 :	leunga	585	Date: 2000/03/29 18:00:00
977 :			Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
978 :	leunga	586	Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
979 :	leunga	585	Description:
980 :
981 :			This update contains *MAJOR* changes to the way code is generated from CPS
982 :			in the module mlriscGen, and in various backend modules.
983 :
984 :			CHANGES
985 :			=======
986 :
987 :			1. MLRiscGen: forward propagation fix.
988 :
989 :			There was a bug in forward propagation introduced at about the same time
990 :			as the MLRISC x86 backend, which prohibits coalescing to be
991 :			performed effectively in loops.
992 :
993 :			Effect: speed up of loops in RISC architectures.
994 :			By itself, this actually slowed down certain benchmarks on the x86.
995 :
996 :			2. MLRiscGen: forward propagating addresses from consing.
997 :
998 :			I've changed the way consing code is generated. Basically I separated
999 :			out the initialization part:
1000 :
1001 :			store tag, offset(allocptr)
1002 :			store elem1, offset+4(allocptr)
1003 :			store elem2, offset+8(allocptr)
1004 :			...
1005 :			store elemn, offset+4n(allocptr)
1006 :
1007 :			and the address computation part:
1008 :
1009 :			celladdr <- offset+4+alloctpr
1010 :
1011 :			and move the address computation part
1012 :
1013 :			Effect: register pressure is generally lower as a result. This
1014 :			makes compilation of certain expressions much faster, such as
1015 :			long lists with non-trivial elements.
1016 :
1017 :			[(0,0), (0,0), .... (0,0)]
1018 :
1019 :			3. MLRiscGen: base pointer elimination.
1020 :
1021 :			As part of the linkage mechanism, we generate the sequence:
1022 :
1023 :			L: ... <- start of the code fragment
1024 :
1025 :			L1:
1026 :			base pointer <- linkreg - L1 + L
1027 :
1028 :			The base pointer was then used for computing relocatable addresses
1029 :			in the code fragment. Frequently (such as in lots of continuations)
1030 :			this is not needed. We now eliminate this sequence whenever possible.
1031 :
1032 :			For compile time efficiency, I'm using a very stupid local heuristic.
1033 :			But in general, this should be done as a control flow analysis.
1034 :
1035 :			Effect: Smaller code size. Speed up of most programs.
1036 :
1037 :			4. Hppa back end
1038 :
1039 :			Long jumps in span dependence resolution used to depend on the existence
1040 :			of the base pointer.
1041 :
1042 :			A jump to a long label L was expanded into the following sequence:
1043 :
1044 :			LDIL %hi(L-8192), %r29
1045 :			LDO %lo(L-8192)(%r29), %r29
1046 :			ADD %r29, baseptr, %r29
1047 :			BV,n %r0(%r29)
1048 :
1049 :			In the presence of change (3) above, this will not work. I've changed
1050 :			it so that the following sequence of instructions are generated, which
1051 :			doesn't mention the base pointer at all:
1052 :
1053 :			BL,n L', %r29 /* branch and link, L' + 4 -> %r29 */
1054 :			L': ADDIL L-(L'+4), %r29 /* Compute address of L */
1055 :			BV,n %r0(%r29) /* Jump */
1056 :
1057 :			5. Alpha back end
1058 :
1059 :			New alpha instructions LDB/LDW have been added, as per Fermin's
1060 :			suggestions. This is unrelated to all other changes.
1061 :
1062 :			6. X86 back end
1063 :
1064 :			I've changed andl to testl in the floating point test sequence
1065 :			whenever appropriate. The Intel optimization guide states that
1066 :	monnier	659	testl is preferable to andl.
1067 :	leunga	585
1068 :			7. RA (x86 only)
1069 :
1070 :			I've improved the spill propagation algorithm, using an approximation
1071 :			of maximal weighted independent sets. This seems to be necessary to
1072 :			alleviate the negative effect in light of the slow down in (1).
1073 :
1074 :			I'll write down the algorithm one of these days.
1075 :
1076 :			8. MLRiscGen: frequencies
1077 :
1078 :			I've added an annotation that states that all call gc blocks have zero
1079 :			execution frequencies. This improves register allocation on the x86.
1080 :
1081 :			BENCHMARKS
1082 :			==========
1083 :
1084 :			I've only perform the comparison on 110.25.
1085 :
1086 :			The platforms are:
1087 :
1088 :			HPPA A four processor HP machine (E9000) with 5G of memory.
1089 :			X86 A 300Hhz Pentium II with 128M of memory, and
1090 :			SPARC An Ultra sparc 2 with 512M of memory.
1091 :
1092 :			I used the following parameters for the SML benchmarks:
1093 :
1094 :			@SMLalloc
1095 :			HPPA 256k
1096 :			SPARC 512k
1097 :			X86 256k
1098 :
1099 :			COMPILATION TIME
1100 :			----------------
1101 :			Here are the numbers comparing the compilation times of the compilers.
1102 :			I've only compared 110.25 compiling the new sources versus
1103 :			a fixpoint version of the new compiler compiling the same.
1104 :
1105 :			110.25 New
1106 :			Total Time in RA Spill+Reload Total Time In RA Spill+Reload
1107 :			HPPA 627s 116s 2684+3584 599s 95s 1003+1879
1108 :			SPARC 892s 173s 2891+3870 708s 116s 1004+1880
1109 :			X86 999s 315s 94006+130691 987s 296s 108877+141957
1110 :
1111 :			110.25 New
1112 :			Code Size Code Size
1113 :			HPPA 8596736 8561421
1114 :			SPARC 8974299 8785143
1115 :			X86 9029180 8716783
1116 :
1117 :			So in summary, things are at least as good as before. Dramatic
1118 :			reduction in compilation is obtained on the Sparc; I can't explain it,
1119 :			but it is reproducible. Perhaps someone should try to reproduce this
1120 :			on their own machines.
1121 :
1122 :			SML BENCHMARKS
1123 :			--------------
1124 :
1125 :			On the average, all benchmarks perform at least as well as before.
1126 :
1127 :			HPPA Compilation Time Spill+Reload Run Time
1128 :			110.25 New 110.25 New 110.25 New
1129 :
1130 :			barnesHut 3.158 3.015 4.75% 1+1 0+0 2.980 2.922 2.00%
1131 :			boyer 6.152 5.708 7.77% 0+0 0+0 0.218 0.213 2.34%
1132 :			count-graphs 1.168 1.120 4.32% 0+0 0+0 22.705 23.073 -1.60%
1133 :			fft 0.877 0.792 10.74% 1+3 1+3 0.602 0.587 2.56%
1134 :			knuthBendix 3.180 2.857 11.32% 0+0 0+0 0.675 0.662 2.02%
1135 :			lexgen 6.190 5.290 17.01% 0+0 0+0 0.913 0.788 15.86%
1136 :			life 0.803 0.703 14.22% 25+25 0+0 0.153 0.140 9.52%
1137 :			logic 2.048 2.007 2.08% 6+6 1+1 4.133 4.008 3.12%
1138 :			mandelbrot 0.077 0.080 -4.17% 0+0 0+0 0.765 0.712 7.49%
1139 :			mlyacc 22.932 20.937 9.53% 154+181 32+57 0.468 0.430 8.91%
1140 :			nucleic 5.183 5.060 2.44% 2+2 0+0 0.125 0.120 4.17%
1141 :			ratio-regions 3.357 3.142 6.84% 0+0 0+0 116.225 113.173 2.70%
1142 :			ray 1.283 1.290 -0.52% 0+0 0+0 2.887 2.855 1.11%
1143 :			simple 6.307 6.032 4.56% 28+30 5+7 3.705 3.658 1.28%
1144 :			tsp 0.888 0.862 3.09% 0+0 0+0 7.040 6.893 2.13%
1145 :			vliw 24.378 23.455 3.94% 106+127 25+45 2.758 2.707 1.91%
1146 :			--------------------------------------------------------------------------
1147 :			Average 6.12% 4.09%
1148 :
1149 :			SPARC Compilation Time Spill+Reload Run Time
1150 :			110.25 New 110.25 New 110.25 New
1151 :
1152 :			barnesHut 3.778 3.592 5.20% 2+2 0+0 3.648 3.453 5.65%
1153 :			boyer 6.632 6.110 8.54% 0+0 0+0 0.258 0.242 6.90%
1154 :			count-graphs 1.435 1.325 8.30% 0+0 0+0 33.672 34.737 -3.07%
1155 :			fft 0.980 0.940 4.26% 3+9 2+6 0.838 0.827 1.41%
1156 :			knuthBendix 3.590 3.138 14.39% 0+0 0+0 0.962 0.967 -0.52%
1157 :			lexgen 6.593 6.072 8.59% 1+1 0+0 1.077 1.078 -0.15%
1158 :			life 0.972 0.868 11.90% 26+26 0+0 0.143 0.140 2.38%
1159 :			logic 2.525 2.387 5.80% 7+7 1+1 5.625 5.158 9.05%
1160 :			mandelbrot 0.090 0.093 -3.57% 0+0 0+0 0.855 0.728 17.39%
1161 :			mlyacc 26.732 23.827 12.19% 162+189 32+57 0.550 0.560 -1.79%
1162 :			nucleic 6.233 6.197 0.59% 3+3 0+0 0.163 0.173 -5.77%
1163 :			ratio-regions 3.780 3.507 7.79% 0+0 0+0 133.993 131.035 2.26%
1164 :			ray 1.595 1.550 2.90% 1+1 0+0 3.440 3.418 0.63%
1165 :			simple 6.972 6.487 7.48% 29+32 5+7 3.523 3.525 -0.05%
1166 :			tsp 1.115 1.063 4.86% 0+0 0+0 7.393 7.265 1.77%
1167 :			vliw 27.765 24.818 11.87% 110+135 25+45 2.265 2.135 6.09%
1168 :			----------------------------------------------------------------------------
1169 :			Average 6.94% 2.64%
1170 :
1171 :			X86 Compilation Time Spill+Reload Run Time
1172 :			110.25 New 110.25 New 110.25 New
1173 :
1174 :			barnesHut 5.530 5.420 2.03% 593+893 597+915 3.532 3.440 2.66%
1175 :			boyer 8.768 7.747 13.19% 493+199 301+289 0.327 0.297 10.11%
1176 :			count-graphs 2.040 2.010 1.49% 298+394 315+457 26.578 28.660 -7.26%
1177 :			fft 1.327 1.302 1.92% 112+209 115+210 1.055 0.962 9.71%
1178 :			knuthBendix 5.218 5.475 -4.69% 451+598 510+650 0.928 0.932 -0.36%
1179 :			lexgen 9.970 9.623 3.60% 1014+841 1157+885 0.947 0.928 1.97%
1180 :			life 1.183 1.183 0.00% 162+182 145+148 0.127 0.103 22.58%
1181 :			logic 3.285 3.512 -6.45% 514+684 591+836 5.682 5.577 1.88%
1182 :			mandelbrot 0.147 0.143 2.33% 38+41 33+54 0.703 0.690 1.93%
1183 :			mlyacc 35.457 32.763 8.22% 3496+4564 3611+4860 0.552 0.550 0.30%
1184 :			nucleic 7.100 6.888 3.07% 239+168 201+158 0.175 0.173 0.96%
1185 :			ratio-regions 6.388 6.843 -6.65% 1182+257 981+300 120.142 120.345 -0.17%
1186 :			ray 2.332 2.338 -0.29% 346+398 402+494 3.593 3.540 1.51%
1187 :			simple 9.912 9.903 0.08% 1475+941 1579+1168 3.057 3.178 -3.83%
1188 :			tsp 1.623 1.532 5.98% 266+200 250+211 8.045 7.878 2.12%
1189 :			vliw 33.947 35.470 -4.29% 2629+2774 2877+3171 2.072 1.890 9.61%
1190 :			----------------------------------------------------------------------------
1191 :			Average 1.22% 3.36%
1192 :
1193 :			----------------------------------------------------------------------
1194 :			Name: Allen Leung
1195 :	leunga	583	Date: 2000/03/23 16:25:00
1196 :			Tag: leunga-20000323-fix_x86_alpha
1197 :			Description:
1198 :
1199 :			1. X86 fixes/changes
1200 :
1201 :			a. The old code generated for SETcc was completely wrong.
1202 :			The Intel optimization guide is VERY misleading.
1203 :
1204 :			2. ALPHA fixes/changes
1205 :
1206 :			a. Added the instructions LDBU, LDWU, STB, STW as per Fermin's suggestion.
1207 :			b. Added a new mode byteWordLoadStores to the functor parameter to Alpha()
1208 :			c. Added reassociation code for address computation.
1209 :
1210 :			----------------------------------------------------------------------
1211 :			Name: Allen Leung
1212 :	leunga	580	Date: 2000/03/22 01:23:00
1213 :			Tag: leunga-20000322-fix_x86_hppa_ra
1214 :			Description:
1215 :
1216 :			1. X86 fixes/changes
1217 :
1218 :			a. x86Rewrite bug with MUL3 (found by Lal)
1219 :			b. Added the instructions FSTS, FSTL
1220 :
1221 :			2. PA-RISC fixes/changes
1222 :
1223 :			a. B label should not be a delay slot candidate! Why did this work?
1224 :			b. ADDT(32, REG(32, r), LI n) now generates one instruction instead of two,
1225 :			as it should be.
1226 :			c. The assembly syntax for fstds and fstdd was wrong.
1227 :			d. Added the composite instruction COMICLR/LDO, which is the immediate
1228 :			operand variant of COMCLR/LDO.
1229 :
1230 :			3. Generic MLRISC
1231 :
1232 :			a. shuffle.sml rewritten to be slightly more efficient
1233 :			b. DIV bug in mltree-simplify fixed (found by Fermin)
1234 :
1235 :			4. Register Allocator
1236 :
1237 :			a. I now release the interference graph earlier during spilling.
1238 :			May improve memory usage.
1239 :
1240 :			----------------------------------------------------------------------
1241 :	blume	577	Name: Matthias Blume
1242 :	blume	578	Date: 2000/03/14 14:15:32
1243 :			Tag: blume_main_v110p26p1_2
1244 :			Description:
1245 :
1246 :			1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
1247 :			additional argument called "template" which is an optional string that
1248 :	monnier	659	specifies the layout of the tool command line. See the CM manual for
1249 :	blume	578	explanation.
1250 :
1251 :	monnier	659	2. A special-purpose tool can be "registered" by simply dropping the
1252 :	blume	578	corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
1253 :			directory where the .cm file lives that uses this tool. (The
1254 :			behavior/misfeature until now was to look for the tool description
1255 :			files in the current working directory.) As before, tool description
1256 :			files could also be anchored -- in which case they can live anywhere
1257 :			they like. Following the recent e-mail discussion, this change should
1258 :			make it easier to have special-purpose tools that are shipped together
1259 :			with the sources of the program that uses them.
1260 :
1261 :			----------------------------------------------------------------------
1262 :			Name: Matthias Blume
1263 :	blume	577	Date: 2000/03/10 07:48:34
1264 :			Tag: blume_main_v110p26p1_1
1265 :			Description:
1266 :
1267 :			I added a re-written version of Dave's fixpt script to src/system.
1268 :			Changes relative to the original version:
1269 :			- sh-ified (not everybody has ksh)
1270 :			- automatically figures out which architecture it runs on
1271 :			- uses ./makeml a bit more cleverly
1272 :			- never invokes ./installml (and, thus, does not clobber your
1273 :			good and working installation of sml in case something goes wrong)
1274 :			- accepts max iteration count using option "-iter <n>"
1275 :			- accepts a "base" name using option "-base <base>"
1276 :
1277 :			It does not build any extraneous heap images but directly rebuilds
1278 :			bin- and boot-hierarchies using makeml's "-rebuild" switch. Finally,
1279 :			it can incorporate existing bin- and boot- hierarchies. For example,
1280 :			suppose the base is set to "sml" (which is the default). Then it
1281 :			successively builds
1282 :
1283 :			sml.bin.<arch>-unix and sml.boot.<arch>-unix
1284 :			then sml1.bin.<arch>-unix and sml1.boot.<arch>-unix
1285 :			then sml2.bin.<arch>-unix and sml2.boot.<arch>-unix
1286 :			...
1287 :			then sml<n>.bin.<arch>-unix and sml<n>.boot.<arch>-unix
1288 :
1289 :			and so on. If any of these already exist, it will just use what's
1290 :			there. In particular, many people will have the initial set of bin
1291 :			and boot files around, so this saves time for at least one full
1292 :			rebuild. Having sets of the form <base><k>.{bin,boot}.<arch>-unix for
1293 :			<k>=1,2,... is normally not a good idea when invoking fixpt. However,
1294 :			they might be the result of an earlier partial run of fixpt (which
1295 :	monnier	659	perhaps got accidentally killed). In this case, fixpt will quickly
1296 :	blume	577	move through what exists before continuing where it left off earlier,
1297 :			and, thus, saves a lot of time.
1298 :
1299 :			----------------------------------------------------------------------
1300 :	leunga	576	Name: Allen Leung
1301 :			Date: 00/03/10 02:20:00
1302 :			Tag: leunga-20000310-fix_x86_asm_ra
1303 :			Description:
1304 :	dbm	570
1305 :	leunga	576	More assembly output problems involving the indexed addressing mode
1306 :			on the x86 have been found and corrected. Thanks to Fermin Reig for the
1307 :			fix.
1308 :
1309 :			The interface and implementation of the register allocator have been changed
1310 :			slightly to accommodate the possibility to skip the register allocation
1311 :			phases completely and go directly to memory allocation. This is needed
1312 :			for C-- use.
1313 :
1314 :	dbm	570	----------------------------------------------------------------------
1315 :	blume	572	Name: Matthias Blume
1316 :	blume	575	Date: 00/03/09 10:23:53
1317 :			Tag: blume_main_v110p26p1_0
1318 :			Description:
1319 :
1320 :			* Complete re-organization of library names. Many libraries have been
1321 :			consolidated so that they share the same path anchor. For example,
1322 :			all MLRISC-related libraries are anchored at MLRISC, most libraries that
1323 :			are SML/NJ-specific are under "smlnj". Notice that names like
1324 :			host-cmb.cm or host-compiler.cm no longer exist. See system/README
1325 :			for a complete description of the new naming scheme. Quick reference:
1326 :
1327 :			host-cmb.cm -> smlnj/cmb.cm
1328 :			host-compiler.cm -> smlnj/compiler.cm
1329 :			full-cm.cm -> smlnj/cm.cm
1330 :			<arch>-<os>.cm -> smlnj/cmb/<arch>-<os>.cm
1331 :			<arch>-compiler.cm -> smlnj/compiler/<arch>.cm
1332 :
1333 :			* Bug fixes in CM.
1334 :			- exceptions in user code are being passed through (i.e., reach top level)
1335 :			- more bugs in paranoia mode fixed
1336 :			- bug related to checking group owners fixed
1337 :
1338 :			* New install.sh script that automagically fetches archive files:
1339 :			The new file config/srcarchiveurl must contain the URL of the
1340 :			(remote) directory that contains bin files (or other source archives).
1341 :			If install.sh does not find the archive locally, it tries to get
1342 :			it from that remote directory.
1343 :			This should simplify installation further: For machines that have
1344 :			access to the internet, just fetch <version>-config.tgz, unpack it,
1345 :	monnier	659	edit config/targets, and go (run config/install.sh). The script will
1346 :	blume	575	fetch everything else that it might need all by itself.
1347 :
1348 :			For CVS users, this mechanism is not relevant for source archives, but
1349 :			it is convenient for getting new sets of binfiles.
1350 :
1351 :			Archives should be tar files compressed with either gzip, compress, or
1352 :			bzip2. The script recognizes .tgz, .tar, tar.gz, tz, .tar.Z, and .tar.bz2.
1353 :
1354 :			----------------------------------------------------------------------
1355 :			Name: Matthias Blume
1356 :	blume	572	Date: 2000/03/07 04:01:04
1357 :			Tag: blume_main_v110_26_2
1358 :	dbm	570	Description:
1359 :	blume	572	- size info in BOOTLIST
1360 :			* no fixed upper limits for number of bootfiles or length of
1361 :			bootfile names in runtime
1362 :			* falling back to old behavior if no BOOTLIST size info found
1363 :			- allocation size heuristics in .run-sml
1364 :			* tries to read cache size from /proc/cpuinfo (this is important for
1365 :			small-cache Celeron systems!)
1366 :			- install.sh robustified
1367 :			- CM manual updates
1368 :			- paranoid mode
1369 :			* no more CMB.deliver() (i.e., all done by CMB.make())
1370 :			* can re-use existing sml.boot.* files
1371 :			* init.cmi now treated as library
1372 :			* library stamps for consistency checks
1373 :			- sml.boot.<arch>-<os>/PIDMAP file
1374 :			* This file is read by the CM startup code. This is used to minimize
1375 :			the amount of dynamic state that needs to be stowed away for the
1376 :			purpose of sharing between interactive system and user code.
1377 :			- CM.Anchor.anchor instead of CM.Anchor.{set,cancel}
1378 :			* Upon request by Elsa. Anchors now controlled by get-set-pair
1379 :			like most other CM state variables.
1380 :			- Compiler.CMSA eliminated
1381 :			* No longer supported by CM anyway.
1382 :			- fixed bugs in pickler that kept biting Stefan
1383 :			* past refs to past refs (was caused by the possibility that
1384 :			ad-hoc sharing is more discriminating than hash-cons sharing)
1385 :			* integer overflow on LargeInt.minInt
1386 :			- ml-{lex,yacc} build scripts now use new mechanism
1387 :			for building standalone programs
1388 :			- fixed several gcc -Wall warnings that were caused by missing header
1389 :			files, missing initializations, etc., in runtime (not all warnings
1390 :			eliminated, though)

---

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