Annotation of /sml/trunk/HISTORY

Parent Directory | Revision Log

---

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

---

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