Annotation of /sml/branches/primop-branch-3/NOTES/HISTORY

Parent Directory | Revision Log

---

Revision 824 - (view) (download)
Original Path: sml/trunk/HISTORY

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 :	george	824
15 :	leunga	815	----------------------------------------------------------------------
16 :	george	824	Name: Lal George
17 :			Date: 2001/05/08 17:26:09 EDT
18 :			Tag: george-20010508-omit-frameptr
19 :			Description:
20 :
21 :			Changes to implement the omit-frame-pointer optimization to support
22 :			raw C calls. For now, there is only support on the Intel x86, but
23 :			other architectures will follow as more experience is gained with this.
24 :
25 :
26 :			----------------------------------------------------------------------
27 :	blume	818	Name: Matthias Blume
28 :	blume	819	Date: 2001/05/07 14:40:00 EDT
29 :			Tag: blume-20010507-proxies
30 :			Description:
31 :
32 :			I made into "proxy libraries" all libraries that qualify for such a
33 :			change. (A qualifying library is a library that has another library or
34 :			groups as its sole member and repeats that member's export list
35 :			verbatim. A proxy library avoids this repetition by omitting its export
36 :			list, effectively inheriting the list that its (only) member exports.
37 :			See the CM manual for more explanation.)
38 :			The main effect is that explicit export lists for these libraries
39 :			do not have to be kepts in sync, making maintenance a bit easier.
40 :
41 :			I also added copyright notices to many .cm-files.
42 :
43 :			Last but not least, I made a new set of bootfiles.
44 :
45 :			----------------------------------------------------------------------
46 :			Name: Matthias Blume
47 :	blume	818	Date: 2001/05/04 17:00:00 EDT
48 :			Tag: blume-20010504-cm-lsplit
49 :			Description:
50 :
51 :			0. John merged pending changes to $/smlnj-lib.cm
52 :
53 :			1. Allen's previous change accidentally backed out of one of Lal's
54 :			earlier changes. I undid this mistake (re-introducing Lal's change).
55 :
56 :			2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
57 :			within the compiler where applicable. There is some code simplification
58 :			because of that.
59 :
60 :			3. The "split" phase (in FLINT) is now part of the default list of phases.
61 :			Compiler.Control.LambdaSplitting.* can be used to globally control the
62 :			lambda-splitting (cross-module-inlining) engine. In addition to that,
63 :			it can now also be controlled on a per-source basis: CM has been taught
64 :			a new tool parameter applicable to ML source files.
65 :
66 :			- To turn lambda-splitting off completely:
67 :			local open Compiler.Control.LambdaSplitting in
68 :			val _ = set Off
69 :			end
70 :			- To make "no lambda-splitting" the global default (but allow per-source
71 :			overriding); this is the initial setting:
72 :			local open Compiler.Control.LambdaSplitting in
73 :			val _ = set (Default NONE)
74 :			end
75 :			- To make "lambda-splitting with aggressiveness a" the global default
76 :			(and allow per-source overriding):
77 :			local open Compiler.Control.LambdaSplitting in
78 :			val _ = set (Default (SOME a))
79 :			end
80 :
81 :			- To turn lambda-splitting off for a given ML souce file (say: a.sml)
82 :			write (in the respective .cm-file):
83 :			a.sml (lambdasplitting:off)
84 :			- To turn lambda-splitting for a.sml on with minimal aggressiveness:
85 :			a.sml (lambdasplitting:on)
86 :			- To turn lambda-splitting for a.sml on with aggressiveness <a> (where
87 :			<a> is a decimal non-negative integer):
88 :			a.sml (lambdasplitting:<a>)
89 :			- To turn lambda-splitting for a.sml on with maximal aggressiveness:
90 :			a.sml (lambdasplitting:infinity)
91 :			- To use the global default for a.sml:
92 :			a.sml (lambdasplitting:default)
93 :			or simply
94 :			a.sml
95 :
96 :			----------------------------------------------------------------------
97 :	leunga	815	Name: Allen Leung
98 :			Date: 2001/05/04 01:57:00 EDT
99 :			Tag: leunga-20010504-sync
100 :			Description:
101 :	leunga	788
102 :	leunga	815	MLRISC features.
103 :
104 :			1. Fix to CMPXCHG instructions.
105 :			2. Changed RA interface to allow annotations in callbacks.
106 :			3. Added a new method to the stream interface to allow annotations updates.
107 :
108 :	george	749	----------------------------------------------------------------------
109 :	blume	812	Name: Matthias Blume
110 :	blume	814	Date: 2001/05/01 11:45:00 EDT
111 :			Tag: blume-20010501-pcedittmp
112 :			Description:
113 :
114 :			Changed install.sh to use the current working directory instead of
115 :			/usr/tmp for a temporary file (pcedittmp). The previous choice
116 :			of /usr/tmp caused trouble with MacOS X because of file premission
117 :			problems.
118 :
119 :			----------------------------------------------------------------------
120 :			Name: Matthias Blume
121 :	blume	812	Date: 2001/04/20 11:10:00 EDT
122 :			Tag: blume-20010420-inMLflag
123 :			Description:
124 :
125 :			- added vp_limitPtrMask to vproc-state.h
126 :			(for use by the raw-C-calls mechanism to implement proper interrupt
127 :			handling)
128 :			- made the ML compiler aware of various data-structure offsets so it
129 :			can generate code for accessing the vp_inML flag and vp_limitPtrMask
130 :			- tweaked mlriscGen.sml to have it emit interrupt-handling code for
131 :			raw C-calls
132 :
133 :			----------------------------------------------------------------------
134 :	george	810	Name: Lal George
135 :			Date: 2001/04/20 09:15:28 EDT
136 :			Tag: george-20010420-macosX
137 :			Description:
138 :
139 :			- Changes to port to Mac OS X; Darwin.
140 :
141 :			- In the process I found that sqrt was broken on the PPC, because the
142 :			fsqrt instruction is not implemented.
143 :
144 :			----------------------------------------------------------------------
145 :	blume	799	Name: Matthias Blume
146 :	blume	808	Date: 2001/04/18 12:45:00 EDT
147 :			Tag: blume-20010418-ccalls
148 :			Description:
149 :
150 :			- fixed two off-by-4 errors in the x86-specific c-calls implementation
151 :			(this bug prevented structure arguments containing pointers from being
152 :			passed correctly)
153 :			- changed the raw-C-call code in mlriscGen.sml in such a way that
154 :			structure arguments are represented as a pointer to the beginning
155 :			of the structure (instead of having a series of synthesized arguments,
156 :			one for each structure member)
157 :
158 :			- made makeml script's verbosity level configurable via environment
159 :			variable (MAKEML_VERBOSITY)
160 :
161 :	blume	813	- eliminated placeholder implementations for f32l, w16s, i16s, and f32s
162 :	blume	808	in rawmem-x86.sml; we are now using the real thing
163 :
164 :			----------------------------------------------------------------------
165 :			Name: Matthias Blume
166 :	blume	806	Date: 2001/03/22 16:25:00 EST
167 :			Tag: blume-20010322-bootfiles
168 :			Description:
169 :
170 :			Created a new set of bootfiles (for your automatic installation convenience).
171 :
172 :			----------------------------------------------------------------------
173 :			Name: Matthias Blume
174 :	blume	805	Date: 2001/03/22 15:10:00 EST
175 :			Tag: blume-20010322-rawmem-parcm
176 :			Description:
177 :
178 :			1. All "raw memory access" primitives for the new FFI are implemented now
179 :			(at least on the x86).
180 :			2. Some further cleanup of CM's parallel make mechanism.
181 :
182 :			----------------------------------------------------------------------
183 :			Name: Matthias Blume
184 :	blume	801	Date: 2001/03/19 17:53:00 EST
185 :			Tag: blume-20010319-parallel
186 :			Description:
187 :
188 :			Parallel make (using compile servers) now works again.
189 :
190 :			To this end, CM.stabilize and CMB.make have been modified to work in
191 :			two passes when compile servers are attached:
192 :			1. Compile everything, do not perform stabilization; this pass
193 :			uses compile servers
194 :			2. Stabilize everything; this pass does not use compile servers
195 :			If there are no compile servers, the two passes are combined into one
196 :			(as before). Splitting the passes increases the inherent parallelism
197 :			in the dependency graph because the entire graph including all
198 :			libraries is available at the same time. This, in turn, improves
199 :			server utilization. The downside is that the master process will
200 :			have to do some extra work after compilation is done (because for
201 :			technical reasons it must re-read all the binfiles during stabilization).
202 :
203 :			----------------------------------------------------------------------
204 :			Name: Matthias Blume
205 :	blume	800	Date: 2001/03/16 12:22:00 EST
206 :			Tag: blume-20010316-bootfiles
207 :			Description:
208 :
209 :			Created a new set of bootfiles (for your automatic installation convenience).
210 :
211 :			----------------------------------------------------------------------
212 :			Name: Matthias Blume
213 :	blume	799	Date: 2001/03/16 11:00:00 EST
214 :			Tag: blume-20010316-MLTREE-fixup
215 :			Description:
216 :
217 :			This is a minor fixup for an (untagged) earlier commit by Allen.
218 :			(A file was missing).
219 :
220 :			----------------------------------------------------------------------
221 :	leunga	788	Name: Allen Leung
222 :	leunga	796	Date: Mon Mar 5 18:54:57 EST 2001
223 :			Tag: leunga-20010305-cut-support
224 :
225 :			1. New support for alternative control-flow in MLTREE.
226 :			Currently we support
227 :
228 :			FLOW_TO(CALL ...., [k1,...,kn])
229 :
230 :			This is needed for 'cuts to' in C-- and try/handle-like constructs
231 :			in Moby
232 :
233 :			New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
234 :
235 :			2. Register Allocator
236 :
237 :			Changes in interface [from Fermin, John]
238 :
239 :			3. Alpha 8-bit SLL support [Fermin]
240 :
241 :			4. All architectures
242 :
243 :			A new module (ClusterExpandCopies) for expanding parallel copies.
244 :
245 :			----------------------------------------------------------------------
246 :			Name: Allen Leung
247 :	leunga	788	Date: 2001/02/27 23:07:00 EST
248 :			Tag: leunga-20010227-minor-stuff
249 :
250 :			1. Alpha bug fix for CMOVNE
251 :			2. Handle mltree COND(..,FCMP ...,...)
252 :			3. Bug fix in simplifier
253 :
254 :			----------------------------------------------------------------------
255 :	blume	777	Name: Matthias Blume
256 :	blume	781	Date: 2001/01/30 17:50:00 EST
257 :			Tag: blume-20010130-sync
258 :			Description:
259 :
260 :			This is just a minor update to sync my devel branch with the main brach.
261 :			The only visible change is the addition of some README files.
262 :
263 :			----------------------------------------------------------------------
264 :			Name: Matthias Blume
265 :	blume	778	Date: 2001/01/12 23:30:00 JST
266 :			Tag: blume-20010112-bootfiles
267 :			Description:
268 :
269 :			Made a new set of bootfiles that goes with the current state of the
270 :			repository.
271 :
272 :			----------------------------------------------------------------------
273 :			Name: Matthias Blume
274 :	blume	777	Date: 2001/01/12 21:20:00 JST
275 :			Tag: blume-20010112-sync
276 :			Description:
277 :
278 :			I am just flushing out some minor changes that had accumulated in
279 :			my private branch in order to sync with the main tree. (This is
280 :			mainly because I had CVS trouble when trying to merge _into_ my
281 :			private branch.)
282 :
283 :			Most people should be completely unaffected by this.
284 :
285 :			----------------------------------------------------------------------
286 :	leunga	775	Name: Allen Leung
287 :			Date: Thu Jan 11 21:03:00 EST 2001
288 :			Tag: leunga-20010111-labexp=mltree
289 :			Description:
290 :
291 :			1. Removed the type LabelExp and replace it by MLTree.
292 :			2. Rewritten mltree-simplify with the pattern matcher tool.
293 :			3. There were some bugs in alpha code generator which would break
294 :			64-bit code generation.
295 :			4. Redo the tools to generate code with the
296 :			5. The CM files in MLRISC (and in src/system/smlnj/MLRISC)
297 :			are now generated by perl scripts.
298 :
299 :			----------------------------------------------------------------------
300 :	blume	771	Name: Matthias Blume
301 :	blume	774	Date: 2001/01/10 21:55:00 JST
302 :			Tag: blume-20010110-rcc
303 :			Description:
304 :
305 :			The RCC stuff now seems to work (but only on the x86).
306 :			This required hacking of the c-calls interface (and -implementation) in
307 :			MLRISC.
308 :
309 :			Normal compiler users should be unaffected.
310 :
311 :			----------------------------------------------------------------------
312 :			Name: Matthias Blume
313 :	blume	773	Date: 2001/01/09 01:20:00 JST
314 :			Tag: blume-20010109-rcc
315 :			Description:
316 :
317 :			This is a fairly big patch, flushing out a large number of pending
318 :			changes that I made to my development copy over the last couple of days.
319 :
320 :			Of practical relevance at this moment is a workaround for a pickling
321 :			bug that Allen ran into the other day. The cause of the bug itself is
322 :			still unknown and it might be hard to fix it properly, but the
323 :			workaround has some merits of its own (namely somewhat reducing pickling
324 :			overhead for certain libraries). Therefore, I think this solution should
325 :			be satisfactory at this time.
326 :
327 :			The rest of the changes (i.e., the vast majority) has to do with my
328 :			ongoing efforts of providing direct support for C function calls from
329 :			ML. At the moment there is a new primop "RAW_CCALL", typing magic
330 :			in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
331 :			case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
332 :			translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
333 :			handle RCC.
334 :
335 :			The last part (the changes to mlriscGen.sml) are still known to be
336 :			wrong on the x86 and not implemented on all other architectures. But
337 :			the infrastructure is in place. I had to change a few functor
338 :			signatures in the backend to be able to route the CCalls interface
339 :			from MLRISC there, and I had to specialize the mltree type (on the
340 :			x86) to include the necessary extensions. (The extensions themselves
341 :			were already there and redy to go in MLRISC/x86).
342 :
343 :			Everything should be very happy as soon as someone helps me with
344 :			mlriscGen.sml...
345 :
346 :			In any case, nothing of this should matter to anyone as long as the
347 :			new primop is not being used (which is going to be the case unless you
348 :			find it where I hid it :). The rest of the compiler is completely
349 :			unaffected.
350 :
351 :			----------------------------------------------------------------------
352 :			Name: Matthias Blume
353 :	blume	772	Date: 2001/01/05 00:30:00 JST
354 :			Tag: blume-20010105-primops
355 :			Description:
356 :
357 :			Added some experimental support for work that I am doing right now.
358 :			These changes mostly concern added primops, but there is also a new
359 :			experimental C library in the runtime system (but currently not enabled
360 :			anywhere except on Linux/X86).
361 :
362 :			In the course of adding primops (and playing with them), I discovered that
363 :			Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
364 :			broken. (Zhong was very right he labeled this stuff as "major gross hack".)
365 :			To recover, I made type information in INL_PRIM mandatory and changed
366 :			prim.sml as well as built-in.sml accordingly. The InLine structure now
367 :			has complete, correct type information (i.e., no bottom types).
368 :
369 :			Since all these changes mean that we need new binfiles, I also bumped the
370 :			version number to 110.32.1.
371 :
372 :			----------------------------------------------------------------------
373 :			Name: Matthias Blume
374 :	blume	771	Date: 2000/12/30 22:10:00 JST
375 :			Tag: blume-20001230-various
376 :			Description:
377 :
378 :			Added proxy libraries for MLRISC and let MLRISC libraries refer
379 :			to each other using path anchors. (See CM manual for explanation.)
380 :
381 :			Updated CM documentation.
382 :
383 :			Fixed some bugs in CM.
384 :
385 :			Implemented "proxy" libraries (= syntactic sugar for CM).
386 :
387 :			Added "-quiet" option to makeml and changed runtime system accordingly.
388 :
389 :			Added cleanup handler for exportML to reset timers and compiler stats.
390 :
391 :			----------------------------------------------------------------------
392 :	george	761	Name: Lal George
393 :			Date: 2000/12/22 22:22:58 EST 2000
394 :			Tag: Release_110_32
395 :			Description:
396 :
397 :			Infinite precision used throughout MLRISC.
398 :			see MLRISC/mltree/machine-int.sig
399 :
400 :			----------------------------------------------------------------------
401 :	blume	756	Name: Matthias Blume
402 :	blume	760	Date: 2000/12/22 23:16:00 JST
403 :			Tag: blume-20001222-warn
404 :			Description:
405 :	blume	759
406 :	blume	760	Corrected wording and formatting of some CM warning message which I
407 :			broke in my previous patch.
408 :	blume	759
409 :			----------------------------------------------------------------------
410 :			Name: Matthias Blume
411 :	blume	763	Date: 2000/12/22 21:20:00 JST
412 :			Tag: blume-20001222-anchorenv
413 :			Description:
414 :
415 :			Fixed CM's handling of anchor environments in connection with CMB.make.
416 :
417 :			----------------------------------------------------------------------
418 :			Name: Matthias Blume
419 :	blume	758	Date: 2000/12/22 13:15:00 JST
420 :			Tag: blume-20001222-cleanup
421 :			Description:
422 :
423 :			Removed src/cm/ffi which does not (and did not) belong here.
424 :
425 :			----------------------------------------------------------------------
426 :			Name: Matthias Blume
427 :	blume	757	Date: 2000/12/21 23:55:00 JST
428 :			Tag: blume-20001221-exn
429 :			Description:
430 :
431 :			Probably most important: CM no longer silently swallows all exceptions
432 :			in the compiler.
433 :			Plus: some other minor CM changes. For example, CM now reports some
434 :			sizes for generated binfiles (code, data, envpickle, lambdapickle).
435 :
436 :			----------------------------------------------------------------------
437 :			Name: Matthias Blume
438 :	blume	756	Date: 2000/12/15 00:01:05 JST
439 :			Tag: blume-20001215-dirtool
440 :			Description:
441 :
442 :			- "dir" tool added.
443 :			- improvements and cleanup to Tools structure
444 :			- documentation updates
445 :
446 :			----------------------------------------------------------------------
447 :	leunga	752	Name: Allen Leung
448 :	leunga	755	Date: Thu Dec 14 03:45:24 EST 2000
449 :			Description:
450 :			Tag: leunga-20001214-int-inf
451 :			Description:
452 :
453 :			In IntInf, added these standard functions, which are missing from our
454 :			implementation:
455 :
456 :			andb : int * int -> int
457 :			xorb : int * int -> int
458 :			orb : int * int -> int
459 :			notb : int -> int
460 :			<< : int * word -> int
461 :			~>> : int * word -> int
462 :
463 :			Not tested, I hope they are correct.
464 :
465 :			----------------------------------------------------------------------
466 :			Name: Allen Leung
467 :	leunga	752	Date: Fri Dec 8 19:23:26 EST 2000
468 :			Description:
469 :			Tag: leunga-20001208-nowhere
470 :			Description:
471 :
472 :			Slight improvements to the 'nowhere' tool to handle OR-patterns,
473 :			to generate better error messages etc. Plus a brief manual.
474 :
475 :			----------------------------------------------------------------------
476 :	george	749	Name: Lal George
477 :			Date: 2000/12/08 09:54:02 EST 2000
478 :			Tag: Release_110_31
479 :			Description:
480 :	george	705
481 :	george	749	- Version 110.31
482 :	leunga	591	----------------------------------------------------------------------
483 :	leunga	744	Name: Allen Leung
484 :			Date: Thu Dec 7 22:01:04 EST 2000
485 :			Tag: leunga-20001207-cell-monster-hack
486 :			Description:
487 :
488 :			Major MLRISC internal changes. Affect all clients.
489 :			Summary:
490 :
491 :			1. Type CELLS.cell = int is now replaced by a datatype.
492 :			As a result, the old regmap is now gone. Almost all interfaces
493 :			in MLRISC change as a consequence.
494 :
495 :			2. A new brand version of machine description tool (v3.0) that generates
496 :			modules expecting the new interface. The old version is removed.
497 :
498 :			3. The RA interface has been further abstracted into two new functors.
499 :			RISC_RA and X86RA. These functors have much simpler interfaces.
500 :			[See also directory MLRISC/demo.]
501 :
502 :			4. Some other new source->source code generation tools are available:
503 :
504 :			a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
505 :			b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
506 :			I use this tool to generate the peephole optimizers---with the new
507 :			cell type changes, peephole rules are becoming difficult to write
508 :			without conditional pattern matching.
509 :
510 :			5. More Intmap -> IntHashTable change. Previous changes by Matthias didn't
511 :			cover the entire MLRISC source tree so many things broke.
512 :
513 :			6. CM files have been moved to the subdirectory MLRISC/cm.
514 :			They are moved because there are a lot of them and they clutter up the
515 :			root dir.
516 :
517 :			7. More detailed documentation to come...
518 :
519 :			NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
520 :			a makeml -rebuild first. This is because of other other
521 :			changes that Matthias has made (see below).
522 :
523 :
524 :			----------------------------------------------------------------------
525 :	blume	733	Name: Matthias Blume
526 :	blume	742	Date: 2000/11/30 23:12:00 JST
527 :			Tag: blume-20001130-filereorg
528 :			Description:
529 :
530 :			Some manual updates and some file reorganizations in CM.
531 :
532 :			----------------------------------------------------------------------
533 :			Name: Matthias Blume
534 :	blume	737	Date: 2000/11/24 17:45:00 JST
535 :			Tag: blume-20001124-link
536 :			Description:
537 :
538 :			Drastically improved link traversal code for the case that the dynamic
539 :			value was already loaded at bootstrap time. As a result, CM and CMB
540 :			now both load blazingly fast -- even on a very slow machine. Also,
541 :			memory consumption has been further reduced by this.
542 :
543 :			Warning: The format of the PIDMAP file has changed. THerefore, to
544 :			bootstrap you have to do this:
545 :
546 :			1. Run CMB.make
547 :			2. Make a symbolic link for the boot directory:
548 :			ln -s sml.boot.ARCH-OS xxx
549 :			3. "Rebuild" the boot directory:
550 :			./makeml -boot xxx -rebuild sml ; rm xxx
551 :			4. Boot normally:
552 :			./makeml
553 :
554 :			----------------------------------------------------------------------
555 :			Name: Matthias Blume
556 :	blume	735	Date: 2000/11/21 21:20:00 JST
557 :			Tag: blume-20001121-tools
558 :			Description:
559 :
560 :			Continued hacking on autoloading problem -- with success this time.
561 :			Also changed tool-plugin mechanism. See new CM manual.
562 :
563 :			----------------------------------------------------------------------
564 :			Name: Matthias Blume
565 :	blume	734	Date: 2000/11/19 14:30:00 JST
566 :			Tag: blume-20001119-autoload
567 :			Description:
568 :
569 :			Some hacking to make autoloading faster. Success for CMB, no success
570 :			so far for CM. There is a reduced structure CM' that autoloads faster.
571 :			(This is a temporary, non-documented hack to be eliminated again when
572 :			the general problem is solved.)
573 :
574 :			----------------------------------------------------------------------
575 :			Name: Matthias Blume
576 :	blume	733	Date: 2000/11/17 14:10:00 JST
577 :			Tag: blume-20001117-pickle-lib
578 :			Description:
579 :
580 :			1. Eliminated comp-lib.cm
581 :			2. Made pickle-lib.cm
582 :			3. Eliminated all uses of intset.sml (from comp-lib.cm)
583 :			4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
584 :			equivalent constructs from smlnj-lib.cm (INtHashTable).
585 :			5. Point 4. also goes for those uses of intmap.* in MLRISC.
586 :			Duplicated intmap modules thrown out.
587 :			6. Hunted down all duplicated SCC code and replaced it with
588 :			equivalent stuff (GraphSCCFn from smlnj-lib.cm).
589 :			7. Rewrote Feedback module.
590 :			8. Moved sortedlist.sml into viscomp-lib.cm. Eventually it
591 :			should be thrown out and equivalent modules from smlnj-lib.cm
592 :			should be used (IntRedBlackSet, IntListSet, ...).
593 :
594 :			Confirmed that compiler compiles to fixpoint.
595 :
596 :			----------------------------------------------------------------------
597 :	leunga	731	Name: Allen Leung
598 :			Date: 2000/11/10 18:00:00
599 :			Tag: leunga-20001110-new-x86-fp
600 :
601 :			A new x86 floating point code generator has been added.
602 :			By default this is turned off. To turn this on, do:
603 :
604 :			CM.autoload "$smlnj/compiler.cm";
605 :			Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
606 :
607 :			Changes:
608 :
609 :			1. Changed FTAN to FPTAN so that the assembly output is correct.
610 :			2. Changed the extension callback for FTANGENT to generate:
611 :
612 :			fptan
613 :			fstp %st(0)
614 :			instead of
615 :			fptan
616 :			fstpl ftempmem
617 :
618 :			3. Numerous assembly fixes for x86.
619 :
620 :			5. Cleaned up the machine code output module x86/x86MC.sml and added
621 :			support for a whole bunch of instructions and addressing modes:
622 :
623 :			fadd/fsub/fsubr/fmul/fdiv/fdivr %st, %st(n)
624 :			faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp %st, %st(n)
625 :			fadd/fsub/fsubr/fmul/fdiv/fdivr %st(n), %st
626 :			fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
627 :			fxch %st(n)
628 :			fld %st(n)
629 :			fst %st(n)
630 :			fst mem
631 :			fstp %st(n)
632 :			fucom %st(n)
633 :			fucomp %st(n)
634 :
635 :			All these are now generated when the fast fp mode is turned on.
636 :
637 :			6. Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
638 :
639 :			----------------------------------------------------------------------
640 :	blume	729	Name: Matthias Blume
641 :			Date: 2000/11/09 11:20:00 JST
642 :			Tag: blume-20001109-scc
643 :			Description:
644 :
645 :			Eliminated some code duplication:
646 :
647 :			1. Added "where" clause to GraphSCCFn in SML/NJ Library.
648 :			(Otherwise the functor is useless.)
649 :			2. Used GraphSCCFn where SCCUtilFun was used previously.
650 :			3. Got rid of SCCUtilFun (in comp-lib.cm).
651 :
652 :			----------------------------------------------------------------------
653 :	george	721	Name: Lal George
654 :			Date: 2000/11/06 09:02:21 EST 2000
655 :			Tag: Release_110_30
656 :			Description:
657 :
658 :			- Version 110.30
659 :			----------------------------------------------------------------------
660 :	blume	715	Name: Matthias Blume
661 :	blume	716	Date: 2000/11/04 14:45:00
662 :			Tag: blume-20001104-mlbuild
663 :			Description:
664 :
665 :			- Made ml-build faster on startup.
666 :			- Documentation fixes.
667 :
668 :			----------------------------------------------------------------------
669 :			Name: Matthias Blume
670 :	blume	715	Date: 2000/11/02 17:00:00 JST
671 :			Tag: blume-20001102-condcomp
672 :			Description:
673 :
674 :			- Small tweaks to pickler -- new BOOTFILES!
675 :			- Version bumped to 110.29.2.
676 :			- Added conditional compilation facility to init.cmi (see comment there).
677 :			----------------------------------------------------------------------
678 :	leunga	713	Name: Allen Leung
679 :			Date: 2000/10/23 19:31:00
680 :			Tag: leunga-20001023-demo-ra
681 :
682 :			1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
683 :			2. Test programs for the graph library updated
684 :			3. Some new MLRISC demo programs added
685 :
686 :			----------------------------------------------------------------------
687 :	blume	710	Name: Matthias Blume
688 :	blume	711	Date: 2000/08/31 22:15:00 JST
689 :			Tag: blume-20001017-errmsg
690 :			Description:
691 :
692 :			More error message grief: Where there used to be no messages, there
693 :			now were some that had bogus error regions. Fixed.
694 :
695 :			----------------------------------------------------------------------
696 :			Name: Matthias Blume
697 :	blume	710	Date: 2000/08/31 17:30:00 JST
698 :			Tag: blume-20001017-v110p29p1
699 :			Description:
700 :
701 :			I made a version 110.29.1 with new bootfiles.
702 :
703 :			Changes: Modified pickler/unpickler for faster and leaner unpickling.
704 :			CM documentation changes and a small bugfix in CM's error reporting.
705 :
706 :			----------------------------------------------------------------------
707 :	george	705	Name: Lal George
708 :			Date: 2000/09/27 14:42:35 EDT
709 :			Tag: george-20000927-nodestatus
710 :			Description:
711 :
712 :			Changed the type of the nodestatus, so that:
713 :
714 :			SPILLED(~1) is now SPILLED
715 :			SPILLED(m) where m>=0 is now MEMREG(m)
716 :			SPILLED(s) where s<~1 is now SPILL_LOC(~s)
717 :
718 :			----------------------------------------------------------------------
719 :	blume	697	Name: Matthias Blume
720 :	blume	703	Date: 2000/09/07 14:45:00 JST
721 :			Tag: blume-20000907-cmerrmsg
722 :			Description:
723 :
724 :			Small tweak to CM to avoid getting ML syntax error messages twice.
725 :
726 :			----------------------------------------------------------------------
727 :			Name: Matthias Blume
728 :	blume	701	Date: 2000/08/31 18:00:00 JST
729 :			Tag: blume-20000831-cvsbootfiles
730 :			Description:
731 :
732 :			New URL for boot files (because the 110.29 files on the BL server do
733 :			now work correctly with my updated install scripts for yacc and lex).
734 :
735 :			----------------------------------------------------------------------
736 :			Name: Matthias Blume
737 :	blume	697	Date: 2000/08/08 12:33:00 JST
738 :			Tag: blume-20000808-manual
739 :			Description:
740 :
741 :			Tiny update to CM manual.
742 :
743 :			----------------------------------------------------------------------
744 :	leunga	695	Name: Allen Leung
745 :			Date: 2000/08/7 19:31:00
746 :			Tag: leunga-20000807-a-whole-bunch-of-stuff
747 :
748 :			Moby, C--, SSA, x86, machine descriptions etc. Should only affect C--
749 :			and Mobdy.
750 :
751 :			1. x86
752 :
753 :			a. Fixes to peephole module by John and Dan.
754 :			b. Assembly fix to SETcc by Allen.
755 :			c. Fix to c-call by John.
756 :			d. Fix to spilling by John. (This one deals with the missing FSTPT case)
757 :			e. Instruction selection optimization to SETcc as suggested by John.
758 :
759 :			For example,
760 :
761 :			MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
762 :
763 :			should generate:
764 :
765 :			MOVL a, x
766 :			SUBL b, x
767 :			SHRL 31, x
768 :
769 :			2. IR stuff
770 :
771 :			A bunch of new DJ-graph related algorithms added. These
772 :			speed up SSA construction.
773 :
774 :			3. SSA + Scheduling
775 :
776 :			Added code for SSA and scheduling to the repository
777 :
778 :			----------------------------------------------------------------------
779 :	george	694	Name: Lal George
780 :			Date: 2000/07/27 11:53:14 EDT
781 :
782 :			Tag: lal-20000727-linux-ppc
783 :			Description:
784 :
785 :			Made changes to support Linux PPC.
786 :			p.s. I have confirmation that the 110.29 boot files work fine.
787 :
788 :			----------------------------------------------------------------------
789 :	blume	684	Name: Matthias Blume
790 :	blume	692	Date: 2000/07/27 17:40:00 JST
791 :			Tag: blume-20000727-scripts
792 :			Description:
793 :
794 :			!!!! WARNING !!!!
795 :			You must recompile the runtime system!
796 :			!!!! WARNING !!!!
797 :
798 :			This is basically another round of script-enhancements:
799 :
800 :			1. sml, ml-build, and ml-makedepend accept options -D and -U to define
801 :			and undefine CM preprocessor symbols.
802 :
803 :			2. ml-build avoids generating a new heap image if it finds that the
804 :			existing one is still ok. (The condition is that no ML file had to
805 :			be recompiled and all ML files are found to be older that the heap
806 :			file.)
807 :
808 :			To make this work smoothly, I also hacked the runtime system as
809 :			well as SMLofNJ.SysInfo to get access to the heap image suffix
810 :			(.sparc-solaris, ...) that is currently being used.
811 :
812 :			Moreover, the signature of CM.mk_standalone has changed. See the
813 :			CM manual.
814 :
815 :			3. ml-makedepend accepts additional options -n, -a, and -o. (See the
816 :			CM manual for details.)
817 :
818 :			4. More CM manual updates:
819 :			- all of the above has been documented.
820 :			- there is now a section describing the (CM-related) command line
821 :			arguments that are accepted by the "sml" command
822 :
823 :			----------------------------------------------------------------------
824 :			Name: Matthias Blume
825 :	blume	691	Date: 2000/07/25 16:20:00 JST
826 :			Tag: blume-20000725-makedepend
827 :			Description:
828 :
829 :			Added a script called ml-makedepend. This can be used in makefiles
830 :			for Unix' make in a way very similar to the "makedepend" command for
831 :			C.
832 :
833 :			The script internally uses function CM.sources.
834 :
835 :			Synopsis:
836 :
837 :			ml-makedepend [-f makefile] cmfile targetname
838 :
839 :			The default for the makefile is "makefile" (or "Makefile" should
840 :			"makefile" not exist).
841 :
842 :			ml-makedepend adds a cmfile/targetname-specific section to this
843 :			makefile (after removing the previous version of this section). The
844 :			section contains a single dependency specification with targetname on
845 :			the LHS (targetname is an arbitrary name), and a list of files derived
846 :			from the cmfile on the RHS. Some of the files on the RHS are
847 :			ARCH/OPSYS-specific. Therefore, ml-makedepend inserts references to
848 :			"make" variables $(ARCH) and $(OPSYS) in place of the corresponding
849 :			path names. The makefile writer is responsible for making sure that
850 :			these variables have correct at the time "make" is invoked.
851 :
852 :			----------------------------------------------------------------------
853 :			Name: Matthias Blume
854 :	blume	690	Date: 2000/07/22 23:30:00 JST
855 :			Tag: blume-20000722-urlupdate
856 :			Description:
857 :
858 :			Changed BOOT and config/srcarchiveurl to point to BL server:
859 :
860 :			ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
861 :
862 :			----------------------------------------------------------------------
863 :			Name: Matthias Blume
864 :	blume	686	Date: 2000/07/18 18:00:00 JST
865 :			Tag: blume-20000718-Version_110_29
866 :			Description:
867 :
868 :			1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
869 :
870 :			2. Updated config/version to 110.29
871 :
872 :			3. Updated config/srcarchiveurl
873 :
874 :			3. New boot files!
875 :			ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
876 :
877 :			----------------------------------------------------------------------
878 :			Name: Matthias Blume
879 :	blume	684	Date: 2000/07/11 13:58:00 JST
880 :			Tag: blume-20000711-doctypo
881 :			Description:
882 :
883 :			Fixed a few typos in CM manual.
884 :
885 :			----------------------------------------------------------------------
886 :	leunga	683	Name: Allen Leung
887 :			Date: 2000/06/15 00:38:00
888 :			Tag: leunga-20000704-sparc-x86
889 :
890 :			1. x86 peephole improvement sp += k; sp -= k => nop [from John]
891 :			2. fix to x86 RET bug [found by Dan Grossman]
892 :			3. sparc assembly bug fix for ticc instructions [found by Fermin]
893 :
894 :			Affects c-- and moby only
895 :
896 :			----------------------------------------------------------------------
897 :	blume	680	Name: Matthias Blume
898 :	blume	682	Date: 2000/07/04 15:26:00
899 :			Tag: blume-20000704-trigger
900 :			Description:
901 :
902 :			1. Improvements to CM manual.
903 :			2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
904 :			of getting a back-trace. The function, when called, raises an
905 :			internal exception which explicitly carries the full back-trace history,
906 :			so it is unaffected by any intervening handle-raise pairs ("trivial"
907 :			or not). The interactive loop will print that history once it arrives
908 :			at top level.
909 :			Short of having all exceptions implicitly carry the full history, the
910 :			recommended way of using this facility is:
911 :			- compile your program with instrumentation "on"
912 :			- run it, when it raises an exception, look at the history
913 :			- if the history is "cut off" because of some handler, go and modify
914 :			your program so that it explicitly calls BTrace.trigger
915 :			- recompile (still instrumented), and rerun; look at the full history
916 :
917 :			----------------------------------------------------------------------
918 :			Name: Matthias Blume
919 :			Date: 2000/07/03 15:36:00 JST
920 :	blume	680	Tag: blume-20000702-manual
921 :			Description:
922 :
923 :			Small corrections and updates to CM manual.
924 :
925 :			----------------------------------------------------------------------
926 :	blume	682	Name: Matthias Blume
927 :	blume	679	Date: 2000/06/29 16:04:00 JST
928 :			Tag: blume-20000629-yacctool
929 :			Description:
930 :
931 :			Changes:
932 :
933 :			1. Class "mlyacc" now takes separate arguments to pass options to
934 :			generated .sml- and .sig-files independently.
935 :			2. Corresponding CM manual updates.
936 :			3. BTrace module now also reports call sites. (However, for loop clusters
937 :			it only shows from where the cluster was entered.) There are associated
938 :			modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
939 :
940 :			----------------------------------------------------------------------
941 :	blume	675	Name: Matthias Blume
942 :	blume	678	Date: 2000/06/27 16:51:00 JST
943 :			Tag: blume-20000627-noweb
944 :			Description:
945 :
946 :			Changes:
947 :
948 :			1. Implemented "subdir" and "witness" options for noweb tool.
949 :			This caused some slight internal changes in CM's tool implementation.
950 :			2. Fixed bug in "tool plugin" mechanism. This is essentially cleaning
951 :			some remaining issues from earlier path anchor changes.
952 :			3. Updated CM manual accordingly.
953 :
954 :			4. Changed implementation of back-tracing so that I now consider it
955 :			ready for prime-time.
956 :
957 :			In particular, you don't have to explicitly trigger the back-trace
958 :			anymore. Instead, if you are running BTrace-instrumented code and
959 :			there is an uncaught exception (regardless of whether or not it was
960 :			raised in instrumented code), the top-level evalloop will print
961 :			the back-trace.
962 :
963 :			Features:
964 :
965 :			- Instrumented and uninstrumented code work together seemlessly.
966 :			(Of course, uninstrumented code is never mentioned in actual
967 :			back-traces.)
968 :
969 :			- Asymptotic time- and space-complexity of instrumented code is
970 :			equal to that of uninstrumented code. (This means that
971 :			tail-recursion is preserved by the instrumentation phase.)
972 :
973 :			- Modules whose code has been instrumented in different sessions
974 :			work together without problem.
975 :
976 :			- There is no penalty whatsoever on uninstrumented code.
977 :
978 :			- There is no penalty on "raise" expressions, even in
979 :			instrumented code.
980 :
981 :			A potential bug (or perhaps it is a feature, too):
982 :
983 :			A back-trace reaches no further than the outermost instrumented
984 :			non-trivial "raise". Here, a "trivial" raise is one that is the
985 :			sole RHS of a "handle" rule. Thus, back-traces reach trough
986 :
987 :			<exp> handle e => raise e
988 :
989 :			and even
990 :
991 :			<exp> handle Foo => raise Bar
992 :
993 :			and, of course, through
994 :
995 :			<exp> handle Foo => ...
996 :
997 :			if the exception was not Foo.
998 :
999 :			Back-traces always reach right through any un-instrumented code
1000 :			including any of its "handle" expressions, trivial or not.
1001 :
1002 :			To try this out, do the following:
1003 :
1004 :			- Erase all existing binfiles for your program.
1005 :			(You may keep binfiles for those modules where you think you
1006 :			definitely don't need back-tracing.)
1007 :			- Turn on back-trace instrumentation:
1008 :			SMLofNJ.Internals.BTrace.mode (SOME true);
1009 :			- Recompile your program. (I.e., run "CM.make" or "use".)
1010 :			- You may now turn instrumentation off again (if you want):
1011 :			SMLofNJ.Internals.BTrace.mode (SOME false);
1012 :			- Run your program as usual. If it raises an exception that
1013 :			reaches the interactive toplevel, then a back-trace will
1014 :			automatically be printed. After that, the toplevel loop
1015 :			will print the exception history as usual.
1016 :
1017 :			----------------------------------------------------------------------
1018 :			Name: Matthias Blume
1019 :	blume	677	Date: 2000/06/26 09:56:46 JST
1020 :			Tag: blume-20000626-setup
1021 :			Description:
1022 :
1023 :			CM: - setup-parameter to "sml" added; this can be used to run arbitrary
1024 :			ML code before and after compiling a file (e.g., to set compiler
1025 :			flags)
1026 :
1027 :			Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
1028 :			- associated changes to btrace.sml (BTrace instrumentation pass)
1029 :			- cleaner implementation of btimp.sml (BTrace tracing and report
1030 :			module)
1031 :
1032 :			CM manual: * new path encoding documented
1033 :			* description of setup-parameter to "sml" added
1034 :
1035 :			The biggest user-visible change to back-tracing is that it is no
1036 :			longer necessary to compile all traced modules within the same
1037 :			session. (This was a real limitation.)
1038 :
1039 :			----------------------------------------------------------------------
1040 :			Name: Matthias Blume
1041 :	blume	676	Date: 2000/06/24 12:40:00 JST
1042 :			Tag: blume-20000624-startup
1043 :			Description:
1044 :
1045 :			Fixes startup slowdown problem. (I was calling SrcPath.sync a _tad_
1046 :			bit too often -- to put it mildly. :)
1047 :
1048 :			----------------------------------------------------------------------
1049 :			Name: Matthias Blume
1050 :	blume	675	Date: 2000/06/23 18:20:00 JST
1051 :			Tag: blume-20000623-btrace
1052 :			Description:
1053 :
1054 :			This updates adds a backtrace facility to aid programmers in debugging
1055 :			their programs. This involves the following changes:
1056 :
1057 :			1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
1058 :			keeping track of the current call stack. When programs are compiled
1059 :			in a special mode, the compiler will insert calls to these hooks
1060 :			into the user program.
1061 :			"Hook" means that it is possible for different implementations of
1062 :			back-tracing to register themselves (at different times).
1063 :
1064 :			2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
1065 :			as an Absyn.dec->Absyn.dec rewrite. Normally this phase is turned off.
1066 :			It can be turned on using this call:
1067 :			SMLofNJ.Internals.BTrace.mode (SOME true);
1068 :			Turning it off again:
1069 :			SMLofNJ.Internals.BTrace.mode (SOME false);
1070 :			Querying the current status:
1071 :			SMLofNJ.Internals.BTrace.mode NONE;
1072 :			Annotated programs are about twice as big as normal ones, and they
1073 :			run a factor of 2 to 4 slower with a dummy back-trace plugin (one
1074 :			where all hooks do nothing). The slowdown with a plugin that is
1075 :			actually useful (such as the one supplied by default) is even greater,
1076 :			but in the case of the default plugin it is still only an constant
1077 :			factor (amortized).
1078 :
1079 :			3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
1080 :			with a sub-structure BTrace for controlling back-tracing. In particular,
1081 :			the above-mentioned function "mode" controls whether the annotation
1082 :			phase is invoked by the compiler. Another important function is
1083 :			"trigger": when called it aborts the current execution and causes
1084 :			the top-level loop to print a full back-trace.
1085 :
1086 :			4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
1087 :			for back-tracing. It keeps track of the dynamic call stack and in
1088 :			addition to that it keeps a partial history at each "level" of that
1089 :			stack. For example, if a tail-calls b, b tail-calls c, and c tail-calls
1090 :			d and b (at separate times, dynamically), then the report will show:
1091 :
1092 :			GOTO d
1093 :			/c
1094 :			GOTO \b
1095 :			CALL a
1096 :
1097 :			This shows that there was an initial non-tail call of a, then a
1098 :			tail-call to b or c, looping behavior in a cluster of functions that
1099 :			consist of b and c, and then a goto from that cluster (i.e., either from
1100 :			b or from c) to d.
1101 :
1102 :			Note that (depending on the user program) the amount of information
1103 :			that the back-trace module has to keep track of at each level is bounded
1104 :			by a constant. Thus, the whole implementation has the same asymptotical
1105 :			complexity as the original program (both in space and in time).
1106 :
1107 :			5. compiler/TopLevel/interact/evalloop.sml has been modified to
1108 :			handle the special exception SMLofNJ.Internals.BTrace.BTrace
1109 :			which is raised by the "trigger" function mentioned above.
1110 :
1111 :			Notes on usage:
1112 :
1113 :			- Annotated code works well together with unannotated code:
1114 :			Unannotated calls simply do not show up at all in the backtrace.
1115 :
1116 :			- It is not a good idea to let modules that were annotated during
1117 :			different sessions run at the same time. This is because the compiler
1118 :			chooses small integers to identify individual functions, and there
1119 :			will be clashes if different modules were compiled in separate sessions.
1120 :			(Nothing will crash, and you will even be told about the clashes, but
1121 :			back-trace information will in general not be useful.)
1122 :
1123 :			- Back-tracing can be confused by callcc and capture.
1124 :
1125 :			- The only way of getting a back-trace right now is to explicitly
1126 :			invoke the "trigger" function from your user program. Eventually, we
1127 :			should make every exception carry back-trace information (if
1128 :			available). But since this creates more overhead at "raise"-time
1129 :			(similar to the current exnHistory overhead), I have not yet
1130 :			implemented this. (The implementation will be rather easy.) With
1131 :			exceptions carrying back-trace information, this facility will be even
1132 :			more useful because users don't need to modify their programs...
1133 :
1134 :			- While it is possible to compile the compiler with back-trace
1135 :			annotations turned on (I did it to get some confidence in
1136 :			correctness), you must make absolutely sure that core.sml and
1137 :			btimp.sml are compiled WITHOUT annotation! (core.sml cannot actually
1138 :			be compiled with annotation because there is no core access yet, but
1139 :			if you compile btimp.sml with annotation, then the system will go into
1140 :			an infinite recursion and crash.)
1141 :			Since CM currently does not know about BTrace, the only way to turn
1142 :			annotations on and off for different modules of the compiler is to
1143 :			interrupt CMB.make, change the settings, and re-invoke it. Of course,
1144 :			this is awkward and clumsy.
1145 :
1146 :			Sample sessions:
1147 :
1148 :			Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
1149 :			- SMLofNJ.Internals.BTrace.mode (SOME true);
1150 :			[autoloading]
1151 :			[autoloading done]
1152 :			val it = false : bool
1153 :			- structure X = struct
1154 :			- fun main n = let
1155 :			- fun a (x, 0) = d x
1156 :			- | a (x, n) = b (x, n - 1)
1157 :			- and b (x, n) = c (x, n)
1158 :			- and c (x, n) = a (x, n)
1159 :			- and d x = e (x, 3)
1160 :			- and e (x, 0) = f x
1161 :			- | e (x, n) = e (x, n - 1)
1162 :			- and f 0 = SMLofNJ.Internals.BTrace.trigger ()
1163 :			- | f n = n * g (n - 1)
1164 :			- and g n = a (n, 3)
1165 :			- in
1166 :			- f n
1167 :			- end
1168 :			- end;
1169 :			structure X : sig val main : int -> int end
1170 :			- X.main 3;
1171 :			*** BACK-TRACE ***
1172 :			GOTO stdIn:4.2-13.20: X.main[2].f
1173 :			GOTO-( stdIn:4.2-13.20: X.main[2].e
1174 :			GOTO stdIn:4.2-13.20: X.main[2].d
1175 :			/ stdIn:4.2-13.20: X.main[2].a
1176 :			| stdIn:4.2-13.20: X.main[2].b
1177 :			GOTO-\ stdIn:4.2-13.20: X.main[2].c
1178 :			CALL stdIn:4.2-13.20: X.main[2].g
1179 :			GOTO stdIn:4.2-13.20: X.main[2].f
1180 :			GOTO-( stdIn:4.2-13.20: X.main[2].e
1181 :			GOTO stdIn:4.2-13.20: X.main[2].d
1182 :			/ stdIn:4.2-13.20: X.main[2].a
1183 :			| stdIn:4.2-13.20: X.main[2].b
1184 :			GOTO-\ stdIn:4.2-13.20: X.main[2].c
1185 :			CALL stdIn:4.2-13.20: X.main[2].g
1186 :			GOTO stdIn:4.2-13.20: X.main[2].f
1187 :			GOTO-( stdIn:4.2-13.20: X.main[2].e
1188 :			GOTO stdIn:4.2-13.20: X.main[2].d
1189 :			/ stdIn:4.2-13.20: X.main[2].a
1190 :			| stdIn:4.2-13.20: X.main[2].b
1191 :			GOTO-\ stdIn:4.2-13.20: X.main[2].c
1192 :			CALL stdIn:4.2-13.20: X.main[2].g
1193 :			GOTO stdIn:4.2-13.20: X.main[2].f
1194 :			CALL stdIn:2.15-17.4: X.main[2]
1195 :			-
1196 :
1197 :			(Note that because of a FLINt bug the above code currently does not
1198 :			compile without BTrace turned on.)
1199 :
1200 :			Here is another example, using my modified Tiger compiler:
1201 :
1202 :			Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
1203 :			- SMLofNJ.Internals.BTrace.mode (SOME true);
1204 :			[autoloading]
1205 :			[autoloading done]
1206 :			val it = false : bool
1207 :			- CM.make "sources.cm";
1208 :			[autoloading]
1209 :			...
1210 :			[autoloading done]
1211 :			[scanning sources.cm]
1212 :			[parsing (sources.cm):parse.sml]
1213 :			[creating directory CM/SKEL ...]
1214 :			[parsing (sources.cm):tiger.lex.sml]
1215 :			...
1216 :			[wrote CM/sparc-unix/semant.sml]
1217 :			[compiling (sources.cm):main.sml]
1218 :			[wrote CM/sparc-unix/main.sml]
1219 :			[New bindings added.]
1220 :			val it = true : bool
1221 :			- Main.compile ("../testcases/merge.tig", "foo.out");
1222 :			*** BACK-TRACE ***
1223 :			CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
1224 :			CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1225 :			CALL lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
1226 :			GOTO lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
1227 :			CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1228 :			CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1229 :			CALL lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
1230 :			/ lib/semant.sml:411.65-543.8: SemantFun[2].transDec
1231 :			CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
1232 :			CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1233 :			CALL lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
1234 :			CALL main.sml:1.18-118.4: Main.compile[2]
1235 :			-
1236 :
1237 :			----------------------------------------------------------------------
1238 :	blume	673	Name: Matthias Blumen
1239 :			Date: 2000/06/21 18:00:00 JST
1240 :			Tag: blume-20000621-manual
1241 :			Description:
1242 :
1243 :			CM manual update: Path environments documented.
1244 :
1245 :			----------------------------------------------------------------------
1246 :	blume	666	Name: Matthias Blume
1247 :	blume	672	Date: 2000/06/19 13:40:00
1248 :			Tag: blume-20000619-manual
1249 :			Description:
1250 :
1251 :			CM manual and system/README update. This only covers the fact that
1252 :			there are no more implicit anchors. (Path environments and the "bind"
1253 :			option to "cm" have yet to be documented.)
1254 :
1255 :			----------------------------------------------------------------------
1256 :			Name: Matthias Blume
1257 :	blume	671	Date: 2000/06/19 11:05:00 JST
1258 :			Tag: blume-20000619-chdir-bugfix
1259 :			Description:
1260 :
1261 :			Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
1262 :
1263 :			----------------------------------------------------------------------
1264 :			Name: Matthias Blume
1265 :	blume	670	Date: 2000/06/18 22:00:10 JST
1266 :			Tag: blume-20000618-implicit-anchors-really-gone
1267 :			Description:
1268 :
1269 :			I updates the previous HISTORY entry where I forgot to mention that
1270 :			implicit anchors are no longer with us.
1271 :
1272 :			The current update also gets rid of the (now useless) controller
1273 :			CM.Control.implicit_anchors.
1274 :
1275 :			----------------------------------------------------------------------
1276 :			Name: Matthias Blume
1277 :	blume	666	Date: 2000/06/16 17:30:00 JST
1278 :			Tag: blume-20000616-anchorenv
1279 :			Description:
1280 :
1281 :			This patch implements the long anticipated (just kidding :) "anchor
1282 :			environment" mechanism. In the course of doing this, I also
1283 :			re-implemented CM's internal "SrcPath" module from scratch. The new
1284 :			one should be more robust in certain boundary cases. In any case, it
1285 :			is a lot cleaner than its predecessor (IMHO).
1286 :
1287 :			This time, although there is yet another boot file format change, I
1288 :			kept the unpickler backward-compatible. As a result, no new bootfiles
1289 :			are necessary and bootstrapping is straightforward. (You cannot read
1290 :			new bootfiles into an old system, but the other way around is no
1291 :			problem.)
1292 :
1293 :			Visible changes:
1294 :
1295 :	blume	670	** 0. Implicit path anchors (without the leading $-symbol) are no
1296 :			longer recognized at all. This means that such path names are not
1297 :			illegal either. For example, the name basis.cm simply refers to a
1298 :			local file called "basis.cm" (i.e, the name is an ordinary path
1299 :			relative to .cm-files directory). Or, to put it differently, only
1300 :			names that start with $ are anchored paths.
1301 :	blume	666
1302 :	blume	670	** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
1303 :			vanished.
1304 :
1305 :			John (Reppy) had critizised this as soon as I originally proposed and
1306 :	blume	666	implemented it, but at that time I did not really deeply believe
1307 :			him. :) Now I came full-circle because I need the $<singlearc> syntax
1308 :			in another place where it cannot be seen as an abbreviation for
1309 :			$/<singlearc>. To avoid the confusion, $<singlearc> now means what it
1310 :			seems to mean (i.e., it "expands" into the corresponding anchor
1311 :			value).
1312 :
1313 :			However, when paths are used as members in CM description files, it
1314 :			continues to be true that there must be at least another arc after the
1315 :			anchor. This is now enforced separately during semantic analysis
1316 :	blume	670	(i.e., from a lexical/syntactical point of view, the notation is ok.)
1317 :	blume	666
1318 :			** 2. The "cm" class now accepts an option "bind". The option's value
1319 :			is a sub-option list of precisely two items -- one labeled "anchor"
1320 :			and the other one labeled "value". As you might expect, "anchor" is
1321 :			used to specify an anchor name to be bound, and "value" specifies what
1322 :			the anchor is being bound to.
1323 :
1324 :			The value must be a directory name and can be given in either standard
1325 :			syntax (including the possibility that it is itself an anchored path)
1326 :			or native syntax.
1327 :
1328 :			Examples:
1329 :
1330 :			foo.cm (bind:(anchor:bar value:$mystuff/bar))
1331 :			lib.cm (bind:(anchor:a value:"H:\\x\\y\\z")) (* only works under windows *)
1332 :
1333 :			and so on.
1334 :
1335 :			The meaning of this is that the .cm-file will be processed with an
1336 :			augmented anchor environment where the given anchor(s) is/are bound to
1337 :			the given values(s).
1338 :
1339 :			The rationale for having this feature is this: Suppose you are trying
1340 :			to use two different (already stable) libraries a.cm and b.cm (that
1341 :			you perhaps didn't write yourself). Further, suppose each of these
1342 :			two libraries internally uses its own auxiliary library $aux/lib.cm.
1343 :			Normally you would now have a problem because the anchor "lib" can not
1344 :			be bound to more than one value globally. Therefore, the project that
1345 :			uses both a.cm and b.cm must locally redirect the anchor to some other
1346 :			place:
1347 :
1348 :			a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
1349 :			b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
1350 :
1351 :			This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
1352 :			/usr/lib/smlnj/b-stuff/aux.cm, respectively.
1353 :
1354 :			Hard-wiring path names is a bit inflexible (and CM will verbosely warn
1355 :			you when you do so at the time of CM.stabilize). Therefore, you can
1356 :			also use an anchored path as the value:
1357 :
1358 :			a.cm (bind:(anchor:lib value:$a-lib))
1359 :			b.cm (bind:(anchor:lib value:$b-lib))
1360 :
1361 :			Now you can globally configure (using the usual CM.Anchor.anchor or
1362 :			pathconfig machinery) bindings for "a-lib" and "b-lib". Since "lib"
1363 :			itself is always locally bound, setting it globally is no longer
1364 :			meaningful or necessary (but it does not hurt either). In fact, "lib"
1365 :			can still be used as a global anchor for separate purposes. As a
1366 :			matter of fact, one can locally define "lib" in terms of a global
1367 :			"lib":
1368 :
1369 :			a.cm (bind:(anchor:lib value:$lib/a))
1370 :			b.cm (bind:(anchor:lib value:$lib/b))
1371 :
1372 :			** 3: The encoding of path names has changed. This affects the way
1373 :			path names are shown in CM's progress report and also the internal
1374 :			protocol encoding used for parallel make.
1375 :
1376 :			The encoding now uses one or more ':'-separated segments. Each
1377 :			segments corresponds to a file that has been specified relative to the
1378 :			file given by its preceding segment. The first segment is either
1379 :			relative to the CWD, absolute, or anchored. Each segment itself is
1380 :			basically a Unix pathname; all segments but the first are relative.
1381 :
1382 :			Example:
1383 :
1384 :			$foo/bar/baz.cm:a/b/c.sml
1385 :
1386 :			This path denotes the file bar/a/b/c.sml relative to the directory
1387 :			denoted by anchor "foo". Notice that the encoding also includes
1388 :			baz.cm which is the .cm-file that listed a/b/c.sml. As usual, such
1389 :			paths are resolved relative to the .cm-files directory, so baz.cm must
1390 :			be ignored to get the "real" pathname.
1391 :
1392 :			To make this fact more obvious, CM puts the names of such "virtual
1393 :			arcs" into parentheses when they appear in progress reports. (No
1394 :			parentheses will appear in the internal protocol encoding.) Thus,
1395 :			what you really see is:
1396 :
1397 :			$foo/bar/(baz.cm):a/b/c.sml
1398 :
1399 :			I find this notation to be much more informative than before.
1400 :
1401 :			Another new feature of the encoding is that special characters
1402 :			including parentheses, colons, (back)slashes, and white space are
1403 :			written as \ddd (where ddd is the decimal encoding of the character).
1404 :
1405 :	blume	670	*** The CM manual still needs to be updated.
1406 :
1407 :	blume	666	----------------------------------------------------------------------
1408 :	leunga	660	Name: Allen Leung
1409 :			Date: 2000/06/15 00:38:00
1410 :			Tag: leunga-20000615-x86-peephole
1411 :
1412 :			x86 Peephole fix by Fermin. Affects c-- and moby only.
1413 :
1414 :			----------------------------------------------------------------------
1415 :	blume	658	Name: Matthias Blume
1416 :			Date: 2000/06/12 11:40:00
1417 :			Tag: blume-20000612-parmakefix
1418 :			Description:
1419 :
1420 :			More cleanup after changing the file naming scheme: This time I
1421 :			repaired the parallel make mechanism for CMB.make which I broke earlier.
1422 :
1423 :			----------------------------------------------------------------------
1424 :	leunga	657	Name: Allen Leung
1425 :			Date: 2000/06/09 01:25:00
1426 :			Tag: leunga-20000609-various
1427 :
1428 :			None of these things should affect normal SML/NJ operations
1429 :
1430 :			1. Peephole improvements provided by Fermin (c--)
1431 :			2. New annotation DEFUSE for adding extra dependence (moby)
1432 :			3. New X86 LOCK instructions (moby)
1433 :			4. New machine description language for reservation tables (scheduling)
1434 :			5. Fixes to various optimization/analysis modules (branch chaining, dominator
1435 :			trees etc.)
1436 :			6. I've changed the CM files so that they can work with versions
1437 :			110.0.6, 110.25 and 110.28
1438 :
1439 :			----------------------------------------------------------------------
1440 :	blume	652	Name: Matthias Blume
1441 :	blume	656	Date: 2000/06/09 12:40:00
1442 :			Tag: blume-20000609-log
1443 :			Description:
1444 :
1445 :			- Removed all(?) remaining RCS Log entries from sources.
1446 :
1447 :			- Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
1448 :			anchored paths).
1449 :
1450 :			----------------------------------------------------------------------
1451 :			Name: Matthias Blume
1452 :	blume	653	Date: 2000/06/07 17:00:00 JST
1453 :			Tag: blume-20000607-no-implicit-anchors
1454 :			Description:
1455 :
1456 :			1. This update changes the default setting for
1457 :			CM.Control.implicit_anchors from true to false. This means that
1458 :			implicit anchors are no longer permitted by default. I also tried to
1459 :			make sure that nothing else still relies on implicit anchors.
1460 :			(This is the next step on the schedule towards a CM that does not even
1461 :			have the notion of implicit anchors anymore.)
1462 :
1463 :			2. More CM manual updates.
1464 :
1465 :			3. I managed to track down and fix the pickling bug I mentioned last
1466 :			time. Because of the previously existing workaround, this entails no
1467 :			immediate practical changes.
1468 :
1469 :			----------------------------------------------------------------------
1470 :			Name: Matthias Blume
1471 :	blume	652	Date: 2000/06/06 11:15:00 JST
1472 :			Tag: blume-20000606-lazierpickle
1473 :			Description:
1474 :
1475 :			!!!! NEW BOOT FILES !!!!
1476 :
1477 :			* The main purpose of this update is to make library pickles lazier in
1478 :			order to reduce the initial space penalty for autoloading a library.
1479 :			As a result, it is now possible to have $smlnj/compiler.cm
1480 :			pre-registered. This should take care of the many complaints or
1481 :			inquiries about missing structure Compiler. This required changes to
1482 :			CM's internal data structures and small tweaks to some algorithms.
1483 :
1484 :			As a neat additional effect, it is no longer necessary (for the sake
1485 :			of lean heap image files) to distinguish between a "minimal" CM and a
1486 :			"full" CM. Now, there is only one CM (i.e., the "full" version:
1487 :			$smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
1488 :			interactive top level. ($smlnj/cm/minimal.cm is gone.)
1489 :
1490 :			To make the life of compiler-hackers easier, "makeml" now also
1491 :			pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm). In other
1492 :			words, after you bootstrap a new sml for the first time, you will not
1493 :			have to autoload $smlnj/cmb.cm again afterwards. (The first time
1494 :			around you will still have to do it, though.)
1495 :
1496 :			* A second change consists of major updates to the CM manual. There
1497 :			are now several appendices with summary information and also a full
1498 :			specification of the CM description file syntax.
1499 :
1500 :			* In directory src/system I added the script "allcross". This script
1501 :			invokes sml and cross-compiles the compiler for all supported
1502 :			architectures. (Useful when providing a new set of boot files.)
1503 :
1504 :			* There seems to be a latent bug in my "lazy pickles" mechanism. I
1505 :			added a small tweak to pickle-util.sml to work around this problem,
1506 :			but it is not a proper fix yet. I will investigate further. (The
1507 :			effect of the bug was an inflation of library pickle size.)
1508 :
1509 :			* Version number increased to 110.28.1 (to avoid compatibility problems).
1510 :
1511 :			----------------------------------------------------------------------
1512 :	leunga	646	Name: Allen Leung
1513 :	leunga	648	Date: 2000/05/25 17:28 EDT
1514 :			Tag: leunga-20000525-ra
1515 :			Description:
1516 :
1517 :			Fixed a bug in freezing phase of the register allocator.
1518 :
1519 :			----------------------------------------------------------------------
1520 :			Name: Allen Leung
1521 :	leunga	647	Date: 2000/05/15 22:53 EDT
1522 :	leunga	646	Tag: leunga-20000515-alpha-x86-ra
1523 :			Description:
1524 :
1525 :			1. Alpha
1526 :
1527 :			Slight cleanup. Removed the instruction SGNXL
1528 :
1529 :			2. X86
1530 :
1531 :			Added the following instructions to the instruction set:
1532 :
1533 :			ROLx, RORx,
1534 :			BTx, BTSx, BTLx, BTRx,
1535 :			XCHGx, and variants with the LOCK prefix
1536 :
1537 :			3. Register Allocation
1538 :
1539 :			The module ra-rewrite-with-renaming has been improved.
1540 :
1541 :	leunga	647	These have no effect on SML/NJ.
1542 :
1543 :	leunga	646	----------------------------------------------------------------------
1544 :	blume	642	Name: Matthias Blume
1545 :	blume	645	Date: 2000/05/15 16:20:00 JST
1546 :			Tag: blume-20000515-lightrebuild
1547 :			Description:
1548 :
1549 :			1. I added an alternative to "-rebuild" to "makeml". The difference is
1550 :			that prior to calling CMB.make' the CM-variable "LIGHT" will be
1551 :			defined. In effect, the command will not build any cross-compiler
1552 :			backends and therefore finish more quickly.
1553 :
1554 :			The "fixpt" script also takes a "-light" switch to be able to use
1555 :			this new facility while compiling for a fixpoint.
1556 :
1557 :			2. I replaced all mentions of anchored paths in group owner specifications
1558 :			with simple relative paths (usually starting with "..").
1559 :			The rationale is that a library's internal workings should not be
1560 :			compromised by the lack of some anchor. (An anchor is necessary
1561 :			for someone who wants to refer to the library by an anchored path,
1562 :			but it should not be necessary to build the same library in the first
1563 :			place.)
1564 :
1565 :			3. I changed the way CM's tool mechanism determines the shell command
1566 :			string used for things like ml-yacc etc. so that it does not break
1567 :			when CM.Control.implicit_anchors is turned off.
1568 :
1569 :			----------------------------------------------------------------------
1570 :			Name: Matthias Blume
1571 :	blume	644	Date: 2000/05/12 18:20:00 JST
1572 :			Tag: blume-20000512-ml-build
1573 :			Description:
1574 :
1575 :			Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
1576 :			getting installed properly (by config/install.sh).
1577 :
1578 :			----------------------------------------------------------------------
1579 :			Name: Matthias Blume
1580 :	blume	643	Date: 2000/05/12 17:30:00 JST
1581 :			Tag: blume-20000512-anchors
1582 :			Description:
1583 :
1584 :			!!! NEW BOOT FILES !!!
1585 :
1586 :			This change is in preparation of fading out support for "implicitly
1587 :			anchored path names". I went through all sources and used the
1588 :			explicit (and relatively new) $-notation. See system/README and the
1589 :			CM manual for more info on this.
1590 :
1591 :			I also modified the anchoring scheme for some things such as "smlnj",
1592 :			"MLRISC", "cm", etc. to take advantage of the fact that explicit
1593 :			anchors are more expressive: anchor name and first arc do not have to
1594 :			coincide. This entails the following user-visible change:
1595 :
1596 :			You have to write $smlnj/foo/bar instead of smlnj/foo/bar. In
1597 :			particular, when you fire up sml with a command-line argument, say,
1598 :			e.g.:
1599 :
1600 :			sml '$smlnj/cmb.cm'
1601 :
1602 :			At the ML toplevel prompt:
1603 :
1604 :			CM.autoload "$smlnj/cmb.cm";
1605 :
1606 :			There is also a new controller in CM.Control that can be used to turn
1607 :			off all remaining support for implicit anchors by saying:
1608 :
1609 :			CM.autoload "$smlnj/
1610 :			#set CM.Control.implicit_anchors false;
1611 :
1612 :			This causes CM to reject implicitly anchored paths. This is (for the
1613 :			time being) less permissive than the "final" version where there will
1614 :			be no more such implicit anchors and relative paths will be just that:
1615 :			relative.
1616 :
1617 :			The next step (version after next version?) will be to make the
1618 :			default for CM.Control.implicit_anchors false. After the dust has
1619 :			settled, I can then produce the "final" version of this...
1620 :
1621 :			Note: Since bootstrapping is a bit tricky, I provided new boot files.
1622 :
1623 :			----------------------------------------------------------------------
1624 :			Name: Matthias Blume
1625 :	blume	642	Date: 2000/05/11 16:30:00 JST
1626 :			Tag: blume-20000511-sources
1627 :			Description:
1628 :
1629 :			The main change is that I added function CM.sources as a generalized
1630 :			version of the earlier CM.makedepend. This entails the following
1631 :			additional changes:
1632 :
1633 :			- CM.makedepend has been dropped.
1634 :
1635 :			- CM manual has been updated.
1636 :
1637 :			- TOOLS signature and API have been changed.
1638 :
1639 :			----------------------------------------------------------------------
1640 :	leunga	640	Name: Allen Leung
1641 :	leunga	641	Date: 2000/05/10 21:17 EDT
1642 :			Tag: leunga-20000510-moby-c--ssa
1643 :			Description:
1644 :
1645 :			Various bug fixes and new features for C--, Moby and MLRISC optimizations.
1646 :			None of these affect SML/NJ.
1647 :
1648 :			1. Register Allocation
1649 :
1650 :			a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
1651 :			This module tries to remove local (i.e. basic block level) redundancies
1652 :			during spilling.
1653 :
1654 :			b. A new framework for performing region based register allocation.
1655 :			Not yet entirely functional.
1656 :
1657 :			2. X86
1658 :
1659 :			a. DefUse for POP was missing the stack pointer [found by Lal]
1660 :			b. Reload for CALL was incorrect in X86Spill [found by John]
1661 :			c. Various fixes in X86Spill so that it can be used correctly for
1662 :			the new spilling module.
1663 :
1664 :			3. SSA/IR
1665 :
1666 :			a. New module ir/dj-dataflow.sml implements elimination based
1667 :			data flow analysis.
1668 :
1669 :			4. MLRiscGen
1670 :
1671 :			a. Fix for gc type annotation
1672 :
1673 :			5. MDGen
1674 :
1675 :			Various fixes for machine description -> ml code translation. For ssa
1676 :			only.
1677 :
1678 :			----------------------------------------------------------------------
1679 :			Name: Allen Leung
1680 :	leunga	640	Date: 2000/05/08 22:17 EDT
1681 :			Tag: leunga-20000508-labexp
1682 :			Description:
1683 :
1684 :			Fermin has found a few assembly problems with constant expressions
1685 :			generated in LabelExp. Mostly, the problems involve extra parentheses,
1686 :			which choke on dumb assemblers. This is his fix.
1687 :
1688 :			----------------------------------------------------------------------
1689 :	dbm	635	Name: Dave MacQueen
1690 :			Date: 2000/04/09 14:00 EDT
1691 :			Tag: dbm-20000502-Version_110_28
1692 :			Description:
1693 :
1694 :			1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
1695 :
1696 :			2. Updated config/version to 110.28
1697 :
1698 :			3. Updated config/srcarchiveurl
1699 :
1700 :			3. New boot files!
1701 :			ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
1702 :
1703 :			----------------------------------------------------------------------
1704 :	blume	634	Name: Matthias Blume
1705 :			Date: 2000/05/01 19:05:00 JST
1706 :			Tag: blume-20000501-noweb
1707 :			Description:
1708 :
1709 :			A new noweb tool has been added. The existing system is entirely
1710 :			unaffected by this, but some CM users have asked for renewed noweb
1711 :			support. Everything is documented in the CM manual.
1712 :
1713 :			New (plugin) libraries:
1714 :
1715 :			noweb-tool.cm
1716 :			nw-ext.cm
1717 :
1718 :			----------------------------------------------------------------------
1719 :	dbm	633	Name: Dave MacQueen
1720 :			Date: 2000/04/30 12:40PM EDT
1721 :			Tag: dbm-20000430-bug_fixes
1722 :			Description:
1723 :
1724 :			1. Fix for bug 1498
1725 :			smlnj/src/system/Basis/Implementation/Unsafe/object.sig
1726 :			smlnj/src/system/Basis/Implementation/Unsafe/object.sml
1727 :			added toRealArray function
1728 :			smlnj/src/compiler/MiscUtil/print/ppobj.sml
1729 :			added check for tag Obj.RealArray to array printing case in ppObj
1730 :
1731 :			2. Fix for bug 1510
1732 :			smlnj/src/compiler/Semant/types/typesutil.sml
1733 :			fixed definition of dummyargs (used by equalTycon) so that
1734 :			dummy args are distinct types
1735 :
1736 :			----------------------------------------------------------------------
1737 :	blume	629	Name: Matthias Blume
1738 :	blume	632	Date: 2000/04/30 01:00:00 JST
1739 :			Tag: blume-20000430-versions
1740 :			Description:
1741 :
1742 :			1. CM version numbering added. This is an implementation of Lal's
1743 :			proposal for adding version numbers and version checking to .cm
1744 :			files. Lal said that his proposal was just that -- a proposal.
1745 :			For the time being I went ahead and implemented it so that people
1746 :			can comment on it. Everything is completely backward-compatible
1747 :			(except for the stable library format, i.e., new bootfiles!).
1748 :
1749 :			As usual, see the CM manual for details.
1750 :
1751 :			2. An alternative syntax for anchored paths has been implemented.
1752 :			Dave has recently voiced the same concerns that I had when I did
1753 :			this, so there should be some support. My take is that eventually
1754 :			I will let support for the current syntax (where anchors are
1755 :			"implicit") fade out in favor of the new, explicit syntax.
1756 :			In order to be backward-compatible, both old and new syntax are
1757 :			currently supported.
1758 :
1759 :			Again, see the CM manual for details.
1760 :
1761 :			3. Parallel make is trying to be slightly smarter: When the master
1762 :			process finds a "bottleneck", i.e., when there is only one
1763 :			compilation unit that can be compiled and everybody else is
1764 :			waiting on it, then it will simply compile it directly instead
1765 :			of clumsily telling one of the slaves to do it.
1766 :
1767 :			4. Support for "unsharing" added. This is necessary in order to be
1768 :			able to have two different versions of the same library running
1769 :			at the same time (e.g., for trying out a new MLRISC while still
1770 :			having the old MLRISC linked into the current compiler, etc.)
1771 :			See the CM manual.
1772 :
1773 :			5. Simple "makedepend" functionality added for generating Makefile
1774 :			dependency information. (This is rather crude at the moment.
1775 :			Expect some changes here in the future.)
1776 :
1777 :			6. ".fun" added as a recognized suffix for ML files. Also documented
1778 :			explicitly in the manual that the fallback behavior (unknown suffix
1779 :			-> ML file) is not an official feature!
1780 :
1781 :			7. Small changes to the pickler for stable libraries.
1782 :
1783 :			8. Several internal changes to CM (for cleanup/improvement).
1784 :
1785 :
1786 :			!!!! NEW BINFILES !!!!
1787 :
1788 :			----------------------------------------------------------------------
1789 :			Name: Matthias Blume
1790 :	blume	631	Date: 2000/04/28 17:30:00 JST
1791 :			Tag: blume-20000428-pathconfig
1792 :			Description:
1793 :
1794 :			1. I changed config/install.sh to remove duplicate entries from the
1795 :			lib/pathconfig file at the end. Moreover, the final version of
1796 :			lib/pathconfig is sorted alphabetically. The same (sorting) is done
1797 :			in src/system/installml.
1798 :
1799 :			2. The config/install.sh script now consistently uses relative
1800 :			pathnames in lib/pathconfig whenever the anchor is in the lib
1801 :			directory. (So far this was true for the libraries that come
1802 :			pre-compiled and bundled as part of the bootfiles but not for
1803 :			libraries that are compiled by the script itself.)
1804 :
1805 :			----------------------------------------------------------------------
1806 :			Name: Matthias Blume
1807 :	blume	629	Date: 2000/04/26 13:10:00 JST
1808 :			Tag: blume-20000426-fun_suffix
1809 :			Description:
1810 :
1811 :			Added ".fun" as a recognized file name suffix (for ML code).
1812 :
1813 :			----------------------------------------------------------------------
1814 :	leunga	628	Name: Allen Leung
1815 :			Date: 2000/04/25 17:00:00 EST
1816 :			Tag: leunga-20000425-alpha-ra
1817 :			Description:
1818 :
1819 :			1. Alpha
1820 :
1821 :			PSEUDOARITH was missing in AlphaRewrite. This causes an endless loop
1822 :			in C--.
1823 :
1824 :			2. RA
1825 :
1826 :			Added a flag "ra-dump-size" to print out the size of the flowgraph
1827 :			and the interference graph.
1828 :
1829 :			----------------------------------------------------------------------
1830 :	dbm	627	Name: Dave MacQueen
1831 :			Date: 2000/04/25/
1832 :			Tag: dbm-20000425-mlyacc_doc_examples
1833 :			Description:
1834 :			Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
1835 :			Updated all three examples in src/ml-yacc/examples to run
1836 :			under 110.* using CM.make.
1837 :
1838 :			----------------------------------------------------------------------
1839 :	leunga	624	Name: Allen Leung
1840 :			Date: 2000/04/20 23:04:00 EST
1841 :			Tag: leunga-20000420-ssa-c---stuff
1842 :			Description:
1843 :
1844 :			This update synchronizes my repository with Yale's. Most of these
1845 :			changes, however, do not affect SML/NJ at all (the RA is an exception).
1846 :
1847 :			1. Register Allocator
1848 :
1849 :			a. An improvement in the interference graph construction:
1850 :			Given a copy
1851 :
1852 :			s <- t
1853 :
1854 :	leunga	628	no interference edge between s and t is added for this definition of s.
1855 :	leunga	624
1856 :			b. I've added two new spill heuristic modules that Fermin and I developed
1857 :			(in the new library RA.cm). These are unused in SML/NJ but maybe
1858 :			useful for others (Moby?)
1859 :
1860 :			2. X86
1861 :
1862 :			a. Various fixes in the backend provided by Fermin [C--] and Lal.
1863 :
1864 :			3. Alpha
1865 :
1866 :			a. Added the BSR instruction and code generation that goes with it [C--]
1867 :			b. Other fixes too numerous to recount provided by Fermin [C--]
1868 :
1869 :			4. Regmaps
1870 :
1871 :			a. The regmaps are not initialized with the identity physical bindings
1872 :			at creation time. This is unneeded.
1873 :
1874 :			5. MLRISC Optimizations
1875 :
1876 :			a. The DJ-Graph module can now compute the iterated dominance frontiers
1877 :			intersects with liveness incrementally in linear time! Woohoo!
1878 :			This is now used in my new SSA construction algorithm.
1879 :
1880 :			b. THe branch reorganization module is now smarter about linear chains of
1881 :			basic blocks.
1882 :
1883 :
1884 :			----------------------------------------------------------------------
1885 :	blume	615	Name: Matthias Blume
1886 :			Date: 2000/04/12 13:52:00 JST
1887 :			Tag: blume_main_v110p27_1
1888 :			Description:
1889 :
1890 :			Changed install.sh script to handle archive files without version number
1891 :			and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
1892 :			name of the boot file archive.
1893 :
1894 :			----------------------------------------------------------------------
1895 :	dbm	607	Name: Dave MacQueen
1896 :			Date: 2000/04/09 14:00 EDT
1897 :			Tag: dbm-20000410-Version_110_27
1898 :			Description:
1899 :
1900 :			1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
1901 :
1902 :			2. Updated src/config/version to 110.27
1903 :
1904 :			3. New boot files!
1905 :
1906 :			----------------------------------------------------------------------
1907 :	leunga	606	Name: Allen Leung
1908 :			Date: 2000/04/09 19:09:00 EST
1909 :			Tag: leunga-20000409-misc
1910 :			Description:
1911 :
1912 :			1. Yet another fix for x86 assembly for idivl, imull, mull and friends.
1913 :
1914 :			2. Miscellaneous improvements to MLRISC (unused in sml/nj)
1915 :
1916 :			----------------------------------------------------------------------
1917 :	monnier	605	Name: Stefan
1918 :			Date: 2000/04/07 10:00:00 EDT
1919 :			Tag: monnier-20000406-branch-handling
1920 :			Description:
1921 :
1922 :			Improved handling of branches (mostly those generated from
1923 :			polymorphic equality), removed switchoff and changed the
1924 :			default optimization settings (more cpsopt and less flintopt).
1925 :
1926 :			----------------------------------------------------------------------
1927 :	leunga	601	Name: Allen Leung
1928 :	leunga	602	Date: 2000/04/06 01:30:00 EST
1929 :			Tag: leunga-20000406-peephole-x86-SSA-2
1930 :			Description:
1931 :
1932 :			Forgot a few files.
1933 :
1934 :			----------------------------------------------------------------------
1935 :			Name: Allen Leung
1936 :	leunga	601	Date: 2000/04/06 00:36:00 EST
1937 :			Tag: leunga-20000406-peephole-x86-SSA
1938 :			Description:
1939 :
1940 :			1. New Peephole code
1941 :
1942 :			2. Minor improvement to X86 instruction selection
1943 :
1944 :			3. Various fixes to SSA and machine description -> code translator
1945 :
1946 :			----------------------------------------------------------------------
1947 :	blume	595	Name: Matthias Blume
1948 :			Date: 2000/04/05 12:30:00 JST
1949 :			Tag: blume_main_v110p26p2_3
1950 :			Description:
1951 :
1952 :			This update just merges three minor cosmetic updates to CM's sources
1953 :			to get ready for the 110.27 code freeze on Friday. No functionality
1954 :			has changed.
1955 :
1956 :			----------------------------------------------------------------------
1957 :	leunga	593	Name: Allen Leung
1958 :	leunga	594	Date: 2000/04/04 19:39:00 EST
1959 :			Tag: leunga-20000404-x86-asm
1960 :			Description:
1961 :
1962 :			1. Fixed a problem in X86 assembly.
1963 :
1964 :			Things like
1965 :
1966 :			jmp %eax
1967 :			jmp (%eax)
1968 :
1969 :			should be output as
1970 :
1971 :			jmp *%eax
1972 :			jmp *(%eax)
1973 :
1974 :			2. Assembly output
1975 :
1976 :			Added a new flag
1977 :
1978 :			"asm-indent-copies" (default to false)
1979 :
1980 :			When this flag is on, parallel copies will be indented an extra level.
1981 :
1982 :			----------------------------------------------------------------------
1983 :			Name: Allen Leung
1984 :	leunga	593	Date: 2000/04/04 03:18:00 EST
1985 :			Tag: leunga-20000404-C--Moby
1986 :			Description:
1987 :
1988 :			All of these fixes are related to C--, Moby, and my own optimization
1989 :			stuff; so they shouldn't affect SML/NJ.
1990 :
1991 :			1. X86
1992 :
1993 :			Various fixes related floating point, and extensions.
1994 :
1995 :			2. Alpha
1996 :
1997 :			Some extra patterns related to loads with signed/zero extension
1998 :			provided by Fermin.
1999 :
2000 :			3. Assembly
2001 :
2002 :	monnier	659	When generating assembly, resolve the value of client defined constants,
2003 :	leunga	593	instead of generating symbolic values. This is controlled by the
2004 :			new flag "asm-resolve-constants", which is default to true.
2005 :
2006 :			4. Machine Descriptions
2007 :
2008 :			a. The precedence parser was slightly broken when parsing infixr symbols.
2009 :			b. The type generalizing code had the bound variables reversed, resulting
2010 :			in a problem during arity raising.
2011 :			c. Various fixes in machine descriptions.
2012 :
2013 :			----------------------------------------------------------------------
2014 :	blume	592	Name: Matthias Blume
2015 :			Date: 2000/04/03 16:05:00 JST
2016 :			Tag: blume_main_v110p26p2_2
2017 :			Description:
2018 :
2019 :			I eliminated coreEnv from compInfo. Access to the "Core" structure is
2020 :			now done via the ordinary static environment that is context to each
2021 :			compilation unit.
2022 :
2023 :			To this end, I arranged that instead of "structure Core" as "structure
2024 :			_Core" is bound in the pervasive environment. Core access is done via
2025 :	monnier	659	_Core (which can never be accidentally rebound because _Core is not a
2026 :	blume	592	legal surface-syntax symbol).
2027 :
2028 :			The current solution is much cleaner because the core environment is
2029 :			now simply part of the pervasive environment which is part of every
2030 :			compilation unit's context anyway. In particular, this eliminates all
2031 :			special-case handling that was necessary until now in order to deal
2032 :			with dynamic and symbolic parts of the core environment.
2033 :
2034 :			Remaining hackery (to bind the "magic" symbol _Core) is localized in the
2035 :	monnier	659	compilation manager's bootstrap compiler (actually: in the "init group"
2036 :	blume	592	handling). See the comments in src/system/smlnj/init/init.cmi for
2037 :			more details.
2038 :
2039 :			I also tried to track down all mentions of "Core" (as string argument
2040 :			to Symbol.strSymbol) in the compiler and replaced them with a
2041 :			reference to the new CoreSym.coreSym. Seems cleaner since the actual
2042 :			name appears in one place only.
2043 :
2044 :			Binfile and bootfile format have not changed, but the switchover from
2045 :			the old "init.cmi" to the new one is a bit tricky, so I supplied new
2046 :			bootfiles anyway.
2047 :
2048 :			----------------------------------------------------------------------
2049 :	leunga	591	Name: Allen Leung
2050 :			Date: 2000/04/02 21:17:00 EST
2051 :			Tag: leunga-20000402-mltree
2052 :			Description:
2053 :	leunga	585
2054 :	leunga	591	1. Renamed the constructor CALL in MLTREE by popular demand.
2055 :			2. Added a bunch of files from my repository. These are currently
2056 :			used by other non-SMLNJ backends.
2057 :
2058 :	leunga	576	----------------------------------------------------------------------
2059 :	leunga	590	Name: Allen Leung
2060 :			Date: 2000/03/31 21:15:00 EST
2061 :			Tag: leunga-20000331-aliasing
2062 :			Description:
2063 :
2064 :			This update contains a rewritten (and hopefully more correct) module
2065 :			for extracting aliasing information from CPS.
2066 :
2067 :			To turn on this feature:
2068 :
2069 :			Compiler.Control.CG.memDisambiguate := true
2070 :
2071 :			To pretty print the region information with assembly
2072 :
2073 :			Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
2074 :
2075 :			To control how many levels of aliasing information are printed, use:
2076 :
2077 :			Compiler.Control.MLRISC.getInt "points-to-show-level" := n
2078 :
2079 :			The default of n is 3.
2080 :
2081 :			----------------------------------------------------------------------
2082 :	dbm	589	Name: David MacQueen
2083 :			Date: 2000/03/31 11:15:00 EST
2084 :			Tag: dbm-20000331-runtime_fix
2085 :			Description:
2086 :
2087 :			This update contains:
2088 :
2089 :			1. runtime/c-lib/c-libraries.c
2090 :			includes added in revision 1.2 caused compilation errors on hppa-hpux
2091 :
2092 :			2. fix for bug 1556
2093 :			system/Basis/Implementation/NJ/internal-signals.sml
2094 :
2095 :			----------------------------------------------------------------------
2096 :	blume	587	Name: Matthias Blume
2097 :	blume	588	Date: 2000/03/31 18:00:00 JST
2098 :			Tag: blume_main_v110p26p2_1
2099 :			Description:
2100 :
2101 :			This update contains:
2102 :
2103 :			1. A small change to CM's handling of stable libraries:
2104 :			CM now maintains one "global" modmap that is used for all stable
2105 :			libraries. The use of such a global modmap maximizes sharing and
2106 :			minimizes the need for re-traversing parts of environments during
2107 :			modmap construction. (However, this has minor impact since modmap
2108 :			construction seems to account for just one percent or less of total
2109 :			compile time.)
2110 :
2111 :			2. I added a "genmap" phase to the statistics. This is where I got the
2112 :			"one percent" number (see above).
2113 :
2114 :			3. CM's new tool parameter mechanism just became _even_ better. :)
2115 :			- The parser understands named parameters and recursive options.
2116 :			- The "make" and "shell" tools use these new features.
2117 :			(This makes it a lot easier to cascade these tools.)
2118 :			- There is a small syntax change: named parameters use a
2119 :
2120 :			<name> : ( <option> ... ) or
2121 :			<name> : <string>
2122 :
2123 :			syntax. Previously, named parameters were implemented in an
2124 :			ad-hoc fashion by each tool individually (by parsing strings)
2125 :			and had the form
2126 :
2127 :			<name>=<string>
2128 :
2129 :			See the CM manual for a full description of these issues.
2130 :
2131 :			----------------------------------------------------------------------
2132 :			Name: Matthias Blume
2133 :	blume	587	Date: 2000/03/30 18:00:00 JST
2134 :			Tag: blume_main_v110p26p2_0
2135 :			Description:
2136 :
2137 :			!!!!! WARNING !!!!!!
2138 :			!! New binfiles !!
2139 :			!!!!!!!!!!!!!!!!!!!!
2140 :
2141 :			This update contains:
2142 :
2143 :			1. Moderate changes to CM:
2144 :
2145 :			- Changes to CM's tools mechanism. In particular, it is now possible
2146 :			to have tools that accept additional "command line" parameters
2147 :			(specified in the .cm file at each instance where the tool's class is
2148 :			used).
2149 :
2150 :	monnier	659	This was done to accommodate the new "make" and "shell" tools which
2151 :			facilitate fairly seamless hookup to portions of code managed using
2152 :	blume	587	Makefiles or Shell scripts.
2153 :
2154 :			There are no classes "shared" or "private" anymore. Instead, the
2155 :			sharing annotation is now a parameter to the "sml" class.
2156 :
2157 :			There is a bit of generic machinery for implementing one's own
2158 :			tools that accept command-line parameters. However, I am not yet fully
2159 :			satisfied with that part, so expect changes here in the future.
2160 :
2161 :			All existing tools are described in the CM manual.
2162 :
2163 :	monnier	659	- Slightly better error handling. (CM now suppresses many followup
2164 :	blume	587	error messages that tended to be more annoying than helpful.)
2165 :
2166 :			2. Major changes to the compiler's static environment data structures.
2167 :
2168 :			- no CMStaticEnv anymore.
2169 :			- no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
2170 :			but it is called Environment), no conversions between different
2171 :			kinds of static environments
2172 :
2173 :			- There is still a notion of a "modmap", but such modmaps are generated
2174 :			on demand at the time when they are needed. This sounds slow, but I
2175 :			sped up the code that generates modmaps enough for this not to lead to
2176 :			a slowdown of the compiler (at least I didn't detect any).
2177 :
2178 :			- To facilitate rapid modmap generation, static environments now
2179 :			contain an (optional) "modtree" structure. Modtree annotations are
2180 :			constructed by the unpickler during unpickling. (This means that
2181 :			the elaborator does not have to worry about modtrees at all.)
2182 :			Modtrees have the advantage that they are compositional in the same
2183 :			way as the environment data structure itself is compositional.
2184 :			As a result, modtrees never hang on to parts of an environment that
2185 :			has already been rendered "stale" by filtering or rebinding.
2186 :
2187 :			- I went through many, many trials and errors before arriving at the
2188 :			current solution. (The initial idea of "linkpaths" did not work.)
2189 :			But the result of all this is that I have touched a lot of files that
2190 :			depend on the "modules" and "types" data structures (most of the
2191 :			elaborator). There were a lot of changes during my "linkpath" trials
2192 :			that could have been reverted to their original state but weren't.
2193 :			Please, don't be too harsh on me for messing with this code a bit more
2194 :			than what was strictly necessary... (I _did_ resist the tempation
2195 :			of doing any "global reformatting" to avoid an untimely death at
2196 :			Dave's hands. :)
2197 :
2198 :			- One positive aspect of the previous point: At least I made sure that
2199 :			all files that I touched now compile without warnings (other than
2200 :			"polyEqual").
2201 :
2202 :			- compiler now tends to run "leaner" (i.e., ties up less memory in
2203 :			redundant modmaps)
2204 :
2205 :			----------------------------------------------------------------------
2206 :	leunga	580	Name: Allen Leung
2207 :	leunga	585	Date: 2000/03/29 18:00:00
2208 :			Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
2209 :	leunga	586	Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
2210 :	leunga	585	Description:
2211 :
2212 :			This update contains *MAJOR* changes to the way code is generated from CPS
2213 :			in the module mlriscGen, and in various backend modules.
2214 :
2215 :			CHANGES
2216 :			=======
2217 :
2218 :			1. MLRiscGen: forward propagation fix.
2219 :
2220 :			There was a bug in forward propagation introduced at about the same time
2221 :			as the MLRISC x86 backend, which prohibits coalescing to be
2222 :			performed effectively in loops.
2223 :
2224 :			Effect: speed up of loops in RISC architectures.
2225 :			By itself, this actually slowed down certain benchmarks on the x86.
2226 :
2227 :			2. MLRiscGen: forward propagating addresses from consing.
2228 :
2229 :			I've changed the way consing code is generated. Basically I separated
2230 :			out the initialization part:
2231 :
2232 :			store tag, offset(allocptr)
2233 :			store elem1, offset+4(allocptr)
2234 :			store elem2, offset+8(allocptr)
2235 :			...
2236 :			store elemn, offset+4n(allocptr)
2237 :
2238 :			and the address computation part:
2239 :
2240 :			celladdr <- offset+4+alloctpr
2241 :
2242 :			and move the address computation part
2243 :
2244 :			Effect: register pressure is generally lower as a result. This
2245 :			makes compilation of certain expressions much faster, such as
2246 :			long lists with non-trivial elements.
2247 :
2248 :			[(0,0), (0,0), .... (0,0)]
2249 :
2250 :			3. MLRiscGen: base pointer elimination.
2251 :
2252 :			As part of the linkage mechanism, we generate the sequence:
2253 :
2254 :			L: ... <- start of the code fragment
2255 :
2256 :			L1:
2257 :			base pointer <- linkreg - L1 + L
2258 :
2259 :			The base pointer was then used for computing relocatable addresses
2260 :			in the code fragment. Frequently (such as in lots of continuations)
2261 :			this is not needed. We now eliminate this sequence whenever possible.
2262 :
2263 :			For compile time efficiency, I'm using a very stupid local heuristic.
2264 :			But in general, this should be done as a control flow analysis.
2265 :
2266 :			Effect: Smaller code size. Speed up of most programs.
2267 :
2268 :			4. Hppa back end
2269 :
2270 :			Long jumps in span dependence resolution used to depend on the existence
2271 :			of the base pointer.
2272 :
2273 :			A jump to a long label L was expanded into the following sequence:
2274 :
2275 :			LDIL %hi(L-8192), %r29
2276 :			LDO %lo(L-8192)(%r29), %r29
2277 :			ADD %r29, baseptr, %r29
2278 :			BV,n %r0(%r29)
2279 :
2280 :			In the presence of change (3) above, this will not work. I've changed
2281 :			it so that the following sequence of instructions are generated, which
2282 :			doesn't mention the base pointer at all:
2283 :
2284 :			BL,n L', %r29 /* branch and link, L' + 4 -> %r29 */
2285 :			L': ADDIL L-(L'+4), %r29 /* Compute address of L */
2286 :			BV,n %r0(%r29) /* Jump */
2287 :
2288 :			5. Alpha back end
2289 :
2290 :			New alpha instructions LDB/LDW have been added, as per Fermin's
2291 :			suggestions. This is unrelated to all other changes.
2292 :
2293 :			6. X86 back end
2294 :
2295 :			I've changed andl to testl in the floating point test sequence
2296 :			whenever appropriate. The Intel optimization guide states that
2297 :	monnier	659	testl is preferable to andl.
2298 :	leunga	585
2299 :			7. RA (x86 only)
2300 :
2301 :			I've improved the spill propagation algorithm, using an approximation
2302 :			of maximal weighted independent sets. This seems to be necessary to
2303 :			alleviate the negative effect in light of the slow down in (1).
2304 :
2305 :			I'll write down the algorithm one of these days.
2306 :
2307 :			8. MLRiscGen: frequencies
2308 :
2309 :			I've added an annotation that states that all call gc blocks have zero
2310 :			execution frequencies. This improves register allocation on the x86.
2311 :
2312 :			BENCHMARKS
2313 :			==========
2314 :
2315 :			I've only perform the comparison on 110.25.
2316 :
2317 :			The platforms are:
2318 :
2319 :			HPPA A four processor HP machine (E9000) with 5G of memory.
2320 :			X86 A 300Hhz Pentium II with 128M of memory, and
2321 :			SPARC An Ultra sparc 2 with 512M of memory.
2322 :
2323 :			I used the following parameters for the SML benchmarks:
2324 :
2325 :			@SMLalloc
2326 :			HPPA 256k
2327 :			SPARC 512k
2328 :			X86 256k
2329 :
2330 :			COMPILATION TIME
2331 :			----------------
2332 :			Here are the numbers comparing the compilation times of the compilers.
2333 :			I've only compared 110.25 compiling the new sources versus
2334 :			a fixpoint version of the new compiler compiling the same.
2335 :
2336 :			110.25 New
2337 :			Total Time in RA Spill+Reload Total Time In RA Spill+Reload
2338 :			HPPA 627s 116s 2684+3584 599s 95s 1003+1879
2339 :			SPARC 892s 173s 2891+3870 708s 116s 1004+1880
2340 :			X86 999s 315s 94006+130691 987s 296s 108877+141957
2341 :
2342 :			110.25 New
2343 :			Code Size Code Size
2344 :			HPPA 8596736 8561421
2345 :			SPARC 8974299 8785143
2346 :			X86 9029180 8716783
2347 :
2348 :			So in summary, things are at least as good as before. Dramatic
2349 :			reduction in compilation is obtained on the Sparc; I can't explain it,
2350 :			but it is reproducible. Perhaps someone should try to reproduce this
2351 :			on their own machines.
2352 :
2353 :			SML BENCHMARKS
2354 :			--------------
2355 :
2356 :			On the average, all benchmarks perform at least as well as before.
2357 :
2358 :			HPPA Compilation Time Spill+Reload Run Time
2359 :			110.25 New 110.25 New 110.25 New
2360 :
2361 :			barnesHut 3.158 3.015 4.75% 1+1 0+0 2.980 2.922 2.00%
2362 :			boyer 6.152 5.708 7.77% 0+0 0+0 0.218 0.213 2.34%
2363 :			count-graphs 1.168 1.120 4.32% 0+0 0+0 22.705 23.073 -1.60%
2364 :			fft 0.877 0.792 10.74% 1+3 1+3 0.602 0.587 2.56%
2365 :			knuthBendix 3.180 2.857 11.32% 0+0 0+0 0.675 0.662 2.02%
2366 :			lexgen 6.190 5.290 17.01% 0+0 0+0 0.913 0.788 15.86%
2367 :			life 0.803 0.703 14.22% 25+25 0+0 0.153 0.140 9.52%
2368 :			logic 2.048 2.007 2.08% 6+6 1+1 4.133 4.008 3.12%
2369 :			mandelbrot 0.077 0.080 -4.17% 0+0 0+0 0.765 0.712 7.49%
2370 :			mlyacc 22.932 20.937 9.53% 154+181 32+57 0.468 0.430 8.91%
2371 :			nucleic 5.183 5.060 2.44% 2+2 0+0 0.125 0.120 4.17%
2372 :			ratio-regions 3.357 3.142 6.84% 0+0 0+0 116.225 113.173 2.70%
2373 :			ray 1.283 1.290 -0.52% 0+0 0+0 2.887 2.855 1.11%
2374 :			simple 6.307 6.032 4.56% 28+30 5+7 3.705 3.658 1.28%
2375 :			tsp 0.888 0.862 3.09% 0+0 0+0 7.040 6.893 2.13%
2376 :			vliw 24.378 23.455 3.94% 106+127 25+45 2.758 2.707 1.91%
2377 :			--------------------------------------------------------------------------
2378 :			Average 6.12% 4.09%
2379 :
2380 :			SPARC Compilation Time Spill+Reload Run Time
2381 :			110.25 New 110.25 New 110.25 New
2382 :
2383 :			barnesHut 3.778 3.592 5.20% 2+2 0+0 3.648 3.453 5.65%
2384 :			boyer 6.632 6.110 8.54% 0+0 0+0 0.258 0.242 6.90%
2385 :			count-graphs 1.435 1.325 8.30% 0+0 0+0 33.672 34.737 -3.07%
2386 :			fft 0.980 0.940 4.26% 3+9 2+6 0.838 0.827 1.41%
2387 :			knuthBendix 3.590 3.138 14.39% 0+0 0+0 0.962 0.967 -0.52%
2388 :			lexgen 6.593 6.072 8.59% 1+1 0+0 1.077 1.078 -0.15%
2389 :			life 0.972 0.868 11.90% 26+26 0+0 0.143 0.140 2.38%
2390 :			logic 2.525 2.387 5.80% 7+7 1+1 5.625 5.158 9.05%
2391 :			mandelbrot 0.090 0.093 -3.57% 0+0 0+0 0.855 0.728 17.39%
2392 :			mlyacc 26.732 23.827 12.19% 162+189 32+57 0.550 0.560 -1.79%
2393 :			nucleic 6.233 6.197 0.59% 3+3 0+0 0.163 0.173 -5.77%
2394 :			ratio-regions 3.780 3.507 7.79% 0+0 0+0 133.993 131.035 2.26%
2395 :			ray 1.595 1.550 2.90% 1+1 0+0 3.440 3.418 0.63%
2396 :			simple 6.972 6.487 7.48% 29+32 5+7 3.523 3.525 -0.05%
2397 :			tsp 1.115 1.063 4.86% 0+0 0+0 7.393 7.265 1.77%
2398 :			vliw 27.765 24.818 11.87% 110+135 25+45 2.265 2.135 6.09%
2399 :			----------------------------------------------------------------------------
2400 :			Average 6.94% 2.64%
2401 :
2402 :			X86 Compilation Time Spill+Reload Run Time
2403 :			110.25 New 110.25 New 110.25 New
2404 :
2405 :			barnesHut 5.530 5.420 2.03% 593+893 597+915 3.532 3.440 2.66%
2406 :			boyer 8.768 7.747 13.19% 493+199 301+289 0.327 0.297 10.11%
2407 :			count-graphs 2.040 2.010 1.49% 298+394 315+457 26.578 28.660 -7.26%
2408 :			fft 1.327 1.302 1.92% 112+209 115+210 1.055 0.962 9.71%
2409 :			knuthBendix 5.218 5.475 -4.69% 451+598 510+650 0.928 0.932 -0.36%
2410 :			lexgen 9.970 9.623 3.60% 1014+841 1157+885 0.947 0.928 1.97%
2411 :			life 1.183 1.183 0.00% 162+182 145+148 0.127 0.103 22.58%
2412 :			logic 3.285 3.512 -6.45% 514+684 591+836 5.682 5.577 1.88%
2413 :			mandelbrot 0.147 0.143 2.33% 38+41 33+54 0.703 0.690 1.93%
2414 :			mlyacc 35.457 32.763 8.22% 3496+4564 3611+4860 0.552 0.550 0.30%
2415 :			nucleic 7.100 6.888 3.07% 239+168 201+158 0.175 0.173 0.96%
2416 :			ratio-regions 6.388 6.843 -6.65% 1182+257 981+300 120.142 120.345 -0.17%
2417 :			ray 2.332 2.338 -0.29% 346+398 402+494 3.593 3.540 1.51%
2418 :			simple 9.912 9.903 0.08% 1475+941 1579+1168 3.057 3.178 -3.83%
2419 :			tsp 1.623 1.532 5.98% 266+200 250+211 8.045 7.878 2.12%
2420 :			vliw 33.947 35.470 -4.29% 2629+2774 2877+3171 2.072 1.890 9.61%
2421 :			----------------------------------------------------------------------------
2422 :			Average 1.22% 3.36%
2423 :
2424 :			----------------------------------------------------------------------
2425 :			Name: Allen Leung
2426 :	leunga	583	Date: 2000/03/23 16:25:00
2427 :			Tag: leunga-20000323-fix_x86_alpha
2428 :			Description:
2429 :
2430 :			1. X86 fixes/changes
2431 :
2432 :			a. The old code generated for SETcc was completely wrong.
2433 :			The Intel optimization guide is VERY misleading.
2434 :
2435 :			2. ALPHA fixes/changes
2436 :
2437 :			a. Added the instructions LDBU, LDWU, STB, STW as per Fermin's suggestion.
2438 :			b. Added a new mode byteWordLoadStores to the functor parameter to Alpha()
2439 :			c. Added reassociation code for address computation.
2440 :
2441 :			----------------------------------------------------------------------
2442 :			Name: Allen Leung
2443 :	leunga	580	Date: 2000/03/22 01:23:00
2444 :			Tag: leunga-20000322-fix_x86_hppa_ra
2445 :			Description:
2446 :
2447 :			1. X86 fixes/changes
2448 :
2449 :			a. x86Rewrite bug with MUL3 (found by Lal)
2450 :			b. Added the instructions FSTS, FSTL
2451 :
2452 :			2. PA-RISC fixes/changes
2453 :
2454 :			a. B label should not be a delay slot candidate! Why did this work?
2455 :			b. ADDT(32, REG(32, r), LI n) now generates one instruction instead of two,
2456 :			as it should be.
2457 :			c. The assembly syntax for fstds and fstdd was wrong.
2458 :			d. Added the composite instruction COMICLR/LDO, which is the immediate
2459 :			operand variant of COMCLR/LDO.
2460 :
2461 :			3. Generic MLRISC
2462 :
2463 :			a. shuffle.sml rewritten to be slightly more efficient
2464 :			b. DIV bug in mltree-simplify fixed (found by Fermin)
2465 :
2466 :			4. Register Allocator
2467 :
2468 :			a. I now release the interference graph earlier during spilling.
2469 :			May improve memory usage.
2470 :
2471 :			----------------------------------------------------------------------
2472 :	blume	577	Name: Matthias Blume
2473 :	blume	578	Date: 2000/03/14 14:15:32
2474 :			Tag: blume_main_v110p26p1_2
2475 :			Description:
2476 :
2477 :			1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
2478 :			additional argument called "template" which is an optional string that
2479 :	monnier	659	specifies the layout of the tool command line. See the CM manual for
2480 :	blume	578	explanation.
2481 :
2482 :	monnier	659	2. A special-purpose tool can be "registered" by simply dropping the
2483 :	blume	578	corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
2484 :			directory where the .cm file lives that uses this tool. (The
2485 :			behavior/misfeature until now was to look for the tool description
2486 :			files in the current working directory.) As before, tool description
2487 :			files could also be anchored -- in which case they can live anywhere
2488 :			they like. Following the recent e-mail discussion, this change should
2489 :			make it easier to have special-purpose tools that are shipped together
2490 :			with the sources of the program that uses them.
2491 :
2492 :			----------------------------------------------------------------------
2493 :			Name: Matthias Blume
2494 :	blume	577	Date: 2000/03/10 07:48:34
2495 :			Tag: blume_main_v110p26p1_1
2496 :			Description:
2497 :
2498 :			I added a re-written version of Dave's fixpt script to src/system.
2499 :			Changes relative to the original version:
2500 :			- sh-ified (not everybody has ksh)
2501 :			- automatically figures out which architecture it runs on
2502 :			- uses ./makeml a bit more cleverly
2503 :			- never invokes ./installml (and, thus, does not clobber your
2504 :			good and working installation of sml in case something goes wrong)
2505 :			- accepts max iteration count using option "-iter <n>"
2506 :			- accepts a "base" name using option "-base <base>"
2507 :
2508 :			It does not build any extraneous heap images but directly rebuilds
2509 :			bin- and boot-hierarchies using makeml's "-rebuild" switch. Finally,
2510 :			it can incorporate existing bin- and boot- hierarchies. For example,
2511 :			suppose the base is set to "sml" (which is the default). Then it
2512 :			successively builds
2513 :
2514 :			sml.bin.<arch>-unix and sml.boot.<arch>-unix
2515 :			then sml1.bin.<arch>-unix and sml1.boot.<arch>-unix
2516 :			then sml2.bin.<arch>-unix and sml2.boot.<arch>-unix
2517 :			...
2518 :			then sml<n>.bin.<arch>-unix and sml<n>.boot.<arch>-unix
2519 :
2520 :			and so on. If any of these already exist, it will just use what's
2521 :			there. In particular, many people will have the initial set of bin
2522 :			and boot files around, so this saves time for at least one full
2523 :			rebuild. Having sets of the form <base><k>.{bin,boot}.<arch>-unix for
2524 :			<k>=1,2,... is normally not a good idea when invoking fixpt. However,
2525 :			they might be the result of an earlier partial run of fixpt (which
2526 :	monnier	659	perhaps got accidentally killed). In this case, fixpt will quickly
2527 :	blume	577	move through what exists before continuing where it left off earlier,
2528 :			and, thus, saves a lot of time.
2529 :
2530 :			----------------------------------------------------------------------
2531 :	leunga	576	Name: Allen Leung
2532 :			Date: 00/03/10 02:20:00
2533 :			Tag: leunga-20000310-fix_x86_asm_ra
2534 :			Description:
2535 :	dbm	570
2536 :	leunga	576	More assembly output problems involving the indexed addressing mode
2537 :			on the x86 have been found and corrected. Thanks to Fermin Reig for the
2538 :			fix.
2539 :
2540 :			The interface and implementation of the register allocator have been changed
2541 :			slightly to accommodate the possibility to skip the register allocation
2542 :			phases completely and go directly to memory allocation. This is needed
2543 :			for C-- use.
2544 :
2545 :	dbm	570	----------------------------------------------------------------------
2546 :	blume	572	Name: Matthias Blume
2547 :	blume	575	Date: 00/03/09 10:23:53
2548 :			Tag: blume_main_v110p26p1_0
2549 :			Description:
2550 :
2551 :			* Complete re-organization of library names. Many libraries have been
2552 :			consolidated so that they share the same path anchor. For example,
2553 :			all MLRISC-related libraries are anchored at MLRISC, most libraries that
2554 :			are SML/NJ-specific are under "smlnj". Notice that names like
2555 :			host-cmb.cm or host-compiler.cm no longer exist. See system/README
2556 :			for a complete description of the new naming scheme. Quick reference:
2557 :
2558 :			host-cmb.cm -> smlnj/cmb.cm
2559 :			host-compiler.cm -> smlnj/compiler.cm
2560 :			full-cm.cm -> smlnj/cm.cm
2561 :			<arch>-<os>.cm -> smlnj/cmb/<arch>-<os>.cm
2562 :			<arch>-compiler.cm -> smlnj/compiler/<arch>.cm
2563 :
2564 :			* Bug fixes in CM.
2565 :			- exceptions in user code are being passed through (i.e., reach top level)
2566 :			- more bugs in paranoia mode fixed
2567 :			- bug related to checking group owners fixed
2568 :
2569 :			* New install.sh script that automagically fetches archive files:
2570 :			The new file config/srcarchiveurl must contain the URL of the
2571 :			(remote) directory that contains bin files (or other source archives).
2572 :			If install.sh does not find the archive locally, it tries to get
2573 :			it from that remote directory.
2574 :			This should simplify installation further: For machines that have
2575 :			access to the internet, just fetch <version>-config.tgz, unpack it,
2576 :	monnier	659	edit config/targets, and go (run config/install.sh). The script will
2577 :	blume	575	fetch everything else that it might need all by itself.
2578 :
2579 :			For CVS users, this mechanism is not relevant for source archives, but
2580 :			it is convenient for getting new sets of binfiles.
2581 :
2582 :			Archives should be tar files compressed with either gzip, compress, or
2583 :			bzip2. The script recognizes .tgz, .tar, tar.gz, tz, .tar.Z, and .tar.bz2.
2584 :
2585 :			----------------------------------------------------------------------
2586 :			Name: Matthias Blume
2587 :	blume	572	Date: 2000/03/07 04:01:04
2588 :			Tag: blume_main_v110_26_2
2589 :	dbm	570	Description:
2590 :	blume	572	- size info in BOOTLIST
2591 :			* no fixed upper limits for number of bootfiles or length of
2592 :			bootfile names in runtime
2593 :			* falling back to old behavior if no BOOTLIST size info found
2594 :			- allocation size heuristics in .run-sml
2595 :			* tries to read cache size from /proc/cpuinfo (this is important for
2596 :			small-cache Celeron systems!)
2597 :			- install.sh robustified
2598 :			- CM manual updates
2599 :			- paranoid mode
2600 :			* no more CMB.deliver() (i.e., all done by CMB.make())
2601 :			* can re-use existing sml.boot.* files
2602 :			* init.cmi now treated as library
2603 :			* library stamps for consistency checks
2604 :			- sml.boot.<arch>-<os>/PIDMAP file
2605 :			* This file is read by the CM startup code. This is used to minimize
2606 :			the amount of dynamic state that needs to be stowed away for the
2607 :			purpose of sharing between interactive system and user code.
2608 :			- CM.Anchor.anchor instead of CM.Anchor.{set,cancel}
2609 :			* Upon request by Elsa. Anchors now controlled by get-set-pair
2610 :			like most other CM state variables.
2611 :			- Compiler.CMSA eliminated
2612 :			* No longer supported by CM anyway.
2613 :			- fixed bugs in pickler that kept biting Stefan
2614 :			* past refs to past refs (was caused by the possibility that
2615 :			ad-hoc sharing is more discriminating than hash-cons sharing)
2616 :			* integer overflow on LargeInt.minInt
2617 :			- ml-{lex,yacc} build scripts now use new mechanism
2618 :			for building standalone programs
2619 :			- fixed several gcc -Wall warnings that were caused by missing header
2620 :			files, missing initializations, etc., in runtime (not all warnings
2621 :			eliminated, though)

---

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