Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 594, Tue Apr 4 23:41:47 2000 UTC	revision 677, Mon Jun 26 00:56:56 2000 UTC
#	Line 12	Line 12
12	Tag: <post-commit CVS tag>	Tag: <post-commit CVS tag>
13	Description:	Description:
14	----------------------------------------------------------------------	----------------------------------------------------------------------
15		Name: Matthias Blume
16		Date: 2000/06/26 09:56:46 JST
17		Tag: blume-20000626-setup
18		Description:
19
20		CM: - setup-parameter to "sml" added; this can be used to run arbitrary
21		ML code before and after compiling a file (e.g., to set compiler
22		flags)
23
24		Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
25		- associated changes to btrace.sml (BTrace instrumentation pass)
26		- cleaner implementation of btimp.sml (BTrace tracing and report
27		module)
28
29		CM manual: * new path encoding documented
30		* description of setup-parameter to "sml" added
31
32		The biggest user-visible change to back-tracing is that it is no
33		longer necessary to compile all traced modules within the same
34		session.  (This was a real limitation.)
35
36		----------------------------------------------------------------------
37		Name: Matthias Blume
38		Date: 2000/06/24 12:40:00 JST
39		Tag: blume-20000624-startup
40		Description:
41
42		Fixes startup slowdown problem.  (I was calling SrcPath.sync a _tad_
43		bit too often -- to put it mildly. :)
44
45		----------------------------------------------------------------------
46		Name: Matthias Blume
47		Date: 2000/06/23 18:20:00 JST
48		Tag: blume-20000623-btrace
49		Description:
50
51		This updates adds a backtrace facility to aid programmers in debugging
52		their programs.  This involves the following changes:
53
54		1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
55		keeping track of the current call stack.  When programs are compiled
56		in a special mode, the compiler will insert calls to these hooks
57		into the user program.
58		"Hook" means that it is possible for different implementations of
59		back-tracing to register themselves (at different times).
60
61		2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
62		as an Absyn.dec->Absyn.dec rewrite.  Normally this phase is turned off.
63		It can be turned on using this call:
64		SMLofNJ.Internals.BTrace.mode (SOME true);
65		Turning it off again:
66		SMLofNJ.Internals.BTrace.mode (SOME false);
67		Querying the current status:
68		SMLofNJ.Internals.BTrace.mode NONE;
69		Annotated programs are about twice as big as normal ones, and they
70		run a factor of 2 to 4 slower with a dummy back-trace plugin (one
71		where all hooks do nothing).  The slowdown with a plugin that is
72		actually useful (such as the one supplied by default) is even greater,
73		but in the case of the default plugin it is still only an constant
74		factor (amortized).
75
76		3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
77		with a sub-structure BTrace for controlling back-tracing.  In particular,
78		the above-mentioned function "mode" controls whether the annotation
79		phase is invoked by the compiler.  Another important function is
80		"trigger": when called it aborts the current execution and causes
81		the top-level loop to print a full back-trace.
82
83		4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
84		for back-tracing.  It keeps track of the dynamic call stack and in
85		addition to that it keeps a partial history at each "level" of that
86		stack.  For example, if a tail-calls b, b tail-calls c, and c tail-calls
87		d and b (at separate times, dynamically), then the report will show:
88
89		GOTO   d
90		/c
91		GOTO  \b
92		CALL   a
93
94		This shows that there was an initial non-tail call of a, then a
95		tail-call to b or c, looping behavior in a cluster of functions that
96		consist of b and c, and then a goto from that cluster (i.e., either from
97		b or from c) to d.
98
99		Note that (depending on the user program) the amount of information
100		that the back-trace module has to keep track of at each level is bounded
101		by a constant.  Thus, the whole implementation has the same asymptotical
102		complexity as the original program (both in space and in time).
103
104		5. compiler/TopLevel/interact/evalloop.sml has been modified to
105		handle the special exception SMLofNJ.Internals.BTrace.BTrace
106		which is raised by the "trigger" function mentioned above.
107
108		Notes on usage:
109
110		- Annotated code works well together with unannotated code:
111		Unannotated calls simply do not show up at all in the backtrace.
112
113		- It is not a good idea to let modules that were annotated during
114		different sessions run at the same time.  This is because the compiler
115		chooses small integers to identify individual functions, and there
116		will be clashes if different modules were compiled in separate sessions.
117		(Nothing will crash, and you will even be told about the clashes, but
118		back-trace information will in general not be useful.)
119
120		- Back-tracing can be confused by callcc and capture.
121
122		- The only way of getting a back-trace right now is to explicitly
123		invoke the "trigger" function from your user program.  Eventually, we
124		should make every exception carry back-trace information (if
125		available).  But since this creates more overhead at "raise"-time
126		(similar to the current exnHistory overhead), I have not yet
127		implemented this.  (The implementation will be rather easy.)  With
128		exceptions carrying back-trace information, this facility will be even
129		more useful because users don't need to modify their programs...
130
131		- While it is possible to compile the compiler with back-trace
132		annotations turned on (I did it to get some confidence in
133		correctness), you must make absolutely sure that core.sml and
134		btimp.sml are compiled WITHOUT annotation!  (core.sml cannot actually
135		be compiled with annotation because there is no core access yet, but
136		if you compile btimp.sml with annotation, then the system will go into
137		an infinite recursion and crash.)
138		Since CM currently does not know about BTrace, the only way to turn
139		annotations on and off for different modules of the compiler is to
140		interrupt CMB.make, change the settings, and re-invoke it.  Of course,
141		this is awkward and clumsy.
142
143		Sample sessions:
144
145		Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
146		- SMLofNJ.Internals.BTrace.mode (SOME true);
147		[autoloading]
148		[autoloading done]
149		val it = false : bool
150		- structure X = struct
151		-     fun main n = let
152		-         fun a (x, 0) = d x
153		-           | a (x, n) = b (x, n - 1)
154		-         and b (x, n) = c (x, n)
155		-         and c (x, n) = a (x, n)
156		-         and d x = e (x, 3)
157		-         and e (x, 0) = f x
158		-           | e (x, n) = e (x, n - 1)
159		-         and f 0 = SMLofNJ.Internals.BTrace.trigger ()
160		-           | f n = n * g (n - 1)
161		-         and g n = a (n, 3)
162		-     in
163		-         f n
164		-     end
165		- end;
166		structure X : sig val main : int -> int end
167		- X.main 3;
168		*** BACK-TRACE ***
169		GOTO   stdIn:4.2-13.20: X.main[2].f
170		GOTO-( stdIn:4.2-13.20: X.main[2].e
171		GOTO   stdIn:4.2-13.20: X.main[2].d
172		/ stdIn:4.2-13.20: X.main[2].a
173		| stdIn:4.2-13.20: X.main[2].b
174		GOTO-\ stdIn:4.2-13.20: X.main[2].c
175		CALL   stdIn:4.2-13.20: X.main[2].g
176		GOTO   stdIn:4.2-13.20: X.main[2].f
177		GOTO-( stdIn:4.2-13.20: X.main[2].e
178		GOTO   stdIn:4.2-13.20: X.main[2].d
179		/ stdIn:4.2-13.20: X.main[2].a
180		| stdIn:4.2-13.20: X.main[2].b
181		GOTO-\ stdIn:4.2-13.20: X.main[2].c
182		CALL   stdIn:4.2-13.20: X.main[2].g
183		GOTO   stdIn:4.2-13.20: X.main[2].f
184		GOTO-( stdIn:4.2-13.20: X.main[2].e
185		GOTO   stdIn:4.2-13.20: X.main[2].d
186		/ stdIn:4.2-13.20: X.main[2].a
187		| stdIn:4.2-13.20: X.main[2].b
188		GOTO-\ stdIn:4.2-13.20: X.main[2].c
189		CALL   stdIn:4.2-13.20: X.main[2].g
190		GOTO   stdIn:4.2-13.20: X.main[2].f
191		CALL   stdIn:2.15-17.4: X.main[2]
192		-
193
194		(Note that because of a FLINt bug the above code currently does not
195		compile without BTrace turned on.)
196
197		Here is another example, using my modified Tiger compiler:
198
199		Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
200		- SMLofNJ.Internals.BTrace.mode (SOME true);
201		[autoloading]
202		[autoloading done]
203		val it = false : bool
204		- CM.make "sources.cm";
205		[autoloading]
206		...
207		[autoloading done]
208		[scanning sources.cm]
209		[parsing (sources.cm):parse.sml]
210		[creating directory CM/SKEL ...]
211		[parsing (sources.cm):tiger.lex.sml]
212		...
213		[wrote CM/sparc-unix/semant.sml]
214		[compiling (sources.cm):main.sml]
215		[wrote CM/sparc-unix/main.sml]
216		[New bindings added.]
217		val it = true : bool
218		- Main.compile ("../testcases/merge.tig", "foo.out");
219		*** BACK-TRACE ***
220		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
221		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
222		CALL   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
223		GOTO   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
224		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
225		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
226		CALL   lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
227		/ lib/semant.sml:411.65-543.8: SemantFun[2].transDec
228		CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
229		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
230		CALL   lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
231		CALL   main.sml:1.18-118.4: Main.compile[2]
232		-
233
234		----------------------------------------------------------------------
235		Name: Matthias Blumen
236		Date: 2000/06/21 18:00:00 JST
237		Tag: blume-20000621-manual
238		Description:
239
240		CM manual update: Path environments documented.
241
242		----------------------------------------------------------------------
243		Name: Matthias Blume
244		Date: 2000/06/19 13:40:00
245		Tag: blume-20000619-manual
246		Description:
247
248		CM manual and system/README update.  This only covers the fact that
249		there are no more implicit anchors.  (Path environments and the "bind"
250		option to "cm" have yet to be documented.)
251
252		----------------------------------------------------------------------
253		Name: Matthias Blume
254		Date: 2000/06/19 11:05:00 JST
255		Tag: blume-20000619-chdir-bugfix
256		Description:
257
258		Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
259
260		----------------------------------------------------------------------
261		Name: Matthias Blume
262		Date: 2000/06/18 22:00:10 JST
263		Tag: blume-20000618-implicit-anchors-really-gone
264		Description:
265
266		I updates the previous HISTORY entry where I forgot to mention that
267		implicit anchors are no longer with us.
268
269		The current update also gets rid of the (now useless) controller
270		CM.Control.implicit_anchors.
271
272		----------------------------------------------------------------------
273		Name: Matthias Blume
274		Date: 2000/06/16 17:30:00 JST
275		Tag: blume-20000616-anchorenv
276		Description:
277
278		This patch implements the long anticipated (just kidding :) "anchor
279		environment" mechanism.  In the course of doing this, I also
280		re-implemented CM's internal "SrcPath" module from scratch.  The new
281		one should be more robust in certain boundary cases.  In any case, it
282		is a lot cleaner than its predecessor (IMHO).
283
284		This time, although there is yet another boot file format change, I
285		kept the unpickler backward-compatible.  As a result, no new bootfiles
286		are necessary and bootstrapping is straightforward.  (You cannot read
287		new bootfiles into an old system, but the other way around is no
288		problem.)
289
290		Visible changes:
291
292		** 0. Implicit path anchors (without the leading $-symbol) are no
293		longer recognized at all. This means that such path names are not
294		illegal either.  For example, the name basis.cm simply refers to a
295		local file called "basis.cm" (i.e, the name is an ordinary path
296		relative to .cm-files directory).  Or, to put it differently, only
297		names that start with $ are anchored paths.
298
299		** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
300		vanished.
301
302		John (Reppy) had critizised this as soon as I originally proposed and
303		implemented it, but at that time I did not really deeply believe
304		him. :) Now I came full-circle because I need the $<singlearc> syntax
305		in another place where it cannot be seen as an abbreviation for
306		$/<singlearc>.  To avoid the confusion, $<singlearc> now means what it
307		seems to mean (i.e., it "expands" into the corresponding anchor
308		value).
309
310		However, when paths are used as members in CM description files, it
311		continues to be true that there must be at least another arc after the
312		anchor.  This is now enforced separately during semantic analysis
313		(i.e., from a lexical/syntactical point of view, the notation is ok.)
314
315		** 2. The "cm" class now accepts an option "bind".  The option's value
316		is a sub-option list of precisely two items -- one labeled "anchor"
317		and the other one labeled "value".  As you might expect, "anchor" is
318		used to specify an anchor name to be bound, and "value" specifies what
319		the anchor is being bound to.
320
321		The value must be a directory name and can be given in either standard
322		syntax (including the possibility that it is itself an anchored path)
323		or native syntax.
324
325		Examples:
326
327		foo.cm (bind:(anchor:bar value:$mystuff/bar))
328		lib.cm (bind:(anchor:a value:"H:\\x\\y\\z"))  (* only works under windows *)
329
330		and so on.
331
332		The meaning of this is that the .cm-file will be processed with an
333		augmented anchor environment where the given anchor(s) is/are bound to
334		the given values(s).
335
336		The rationale for having this feature is this: Suppose you are trying
337		to use two different (already stable) libraries a.cm and b.cm (that
338		you perhaps didn't write yourself).  Further, suppose each of these
339		two libraries internally uses its own auxiliary library $aux/lib.cm.
340		Normally you would now have a problem because the anchor "lib" can not
341		be bound to more than one value globally.  Therefore, the project that
342		uses both a.cm and b.cm must locally redirect the anchor to some other
343		place:
344
345		a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
346		b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
347
348		This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
349		/usr/lib/smlnj/b-stuff/aux.cm, respectively.
350
351		Hard-wiring path names is a bit inflexible (and CM will verbosely warn
352		you when you do so at the time of CM.stabilize).  Therefore, you can
353		also use an anchored path as the value:
354
355		a.cm (bind:(anchor:lib value:$a-lib))
356		b.cm (bind:(anchor:lib value:$b-lib))
357
358		Now you can globally configure (using the usual CM.Anchor.anchor or
359		pathconfig machinery) bindings for "a-lib" and "b-lib".  Since "lib"
360		itself is always locally bound, setting it globally is no longer
361		meaningful or necessary (but it does not hurt either).  In fact, "lib"
362		can still be used as a global anchor for separate purposes.  As a
363		matter of fact, one can locally define "lib" in terms of a global
364		"lib":
365
366		a.cm (bind:(anchor:lib value:$lib/a))
367		b.cm (bind:(anchor:lib value:$lib/b))
368
369		** 3: The encoding of path names has changed.  This affects the way
370		path names are shown in CM's progress report and also the internal
371		protocol encoding used for parallel make.
372
373		The encoding now uses one or more ':'-separated segments.  Each
374		segments corresponds to a file that has been specified relative to the
375		file given by its preceding segment.  The first segment is either
376		relative to the CWD, absolute, or anchored.  Each segment itself is
377		basically a Unix pathname; all segments but the first are relative.
378
379		Example:
380
381		$foo/bar/baz.cm:a/b/c.sml
382
383		This path denotes the file bar/a/b/c.sml relative to the directory
384		denoted by anchor "foo".  Notice that the encoding also includes
385		baz.cm which is the .cm-file that listed a/b/c.sml.  As usual, such
386		paths are resolved relative to the .cm-files directory, so baz.cm must
387		be ignored to get the "real" pathname.
388
389		To make this fact more obvious, CM puts the names of such "virtual
390		arcs" into parentheses when they appear in progress reports. (No
391		parentheses will appear in the internal protocol encoding.)  Thus,
392		what you really see is:
393
394		$foo/bar/(baz.cm):a/b/c.sml
395
396		I find this notation to be much more informative than before.
397
398		Another new feature of the encoding is that special characters
399		including parentheses, colons, (back)slashes, and white space are
400		written as \ddd (where ddd is the decimal encoding of the character).
401
402		*** The CM manual still needs to be updated.
403
404		----------------------------------------------------------------------
405		Name: Allen Leung
406		Date: 2000/06/15 00:38:00
407		Tag: leunga-20000615-x86-peephole
408
409		x86 Peephole fix by Fermin.  Affects c-- and moby only.
410
411		----------------------------------------------------------------------
412		Name: Matthias Blume
413		Date: 2000/06/12 11:40:00
414		Tag: blume-20000612-parmakefix
415		Description:
416
417		More cleanup after changing the file naming scheme: This time I
418		repaired the parallel make mechanism for CMB.make which I broke earlier.
419
420		----------------------------------------------------------------------
421		Name: Allen Leung
422		Date: 2000/06/09 01:25:00
423		Tag: leunga-20000609-various
424
425		None of these things should affect normal SML/NJ operations
426
427		1. Peephole improvements provided by Fermin (c--)
428		2. New annotation DEFUSE for adding extra dependence (moby)
429		3. New X86 LOCK instructions (moby)
430		4. New machine description language for reservation tables (scheduling)
431		5. Fixes to various optimization/analysis modules (branch chaining, dominator
432		trees etc.)
433		6. I've changed the CM files so that they can work with versions
434		110.0.6, 110.25 and 110.28
435
436		----------------------------------------------------------------------
437		Name: Matthias Blume
438		Date: 2000/06/09 12:40:00
439		Tag: blume-20000609-log
440		Description:
441
442		- Removed all(?) remaining RCS Log entries from sources.
443
444		- Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
445		anchored paths).
446
447		----------------------------------------------------------------------
448		Name: Matthias Blume
449		Date: 2000/06/07 17:00:00 JST
450		Tag: blume-20000607-no-implicit-anchors
451		Description:
452
453		1. This update changes the default setting for
454		CM.Control.implicit_anchors from true to false.  This means that
455		implicit anchors are no longer permitted by default.  I also tried to
456		make sure that nothing else still relies on implicit anchors.
457		(This is the next step on the schedule towards a CM that does not even
458		have the notion of implicit anchors anymore.)
459
460		2. More CM manual updates.
461
462		3. I managed to track down and fix the pickling bug I mentioned last
463		time.  Because of the previously existing workaround, this entails no
464		immediate practical changes.
465
466		----------------------------------------------------------------------
467		Name: Matthias Blume
468		Date: 2000/06/06 11:15:00 JST
469		Tag: blume-20000606-lazierpickle
470		Description:
471
472		!!!! NEW BOOT FILES !!!!
473
474		* The main purpose of this update is to make library pickles lazier in
475		order to reduce the initial space penalty for autoloading a library.
476		As a result, it is now possible to have $smlnj/compiler.cm
477		pre-registered.  This should take care of the many complaints or
478		inquiries about missing structure Compiler.  This required changes to
479		CM's internal data structures and small tweaks to some algorithms.
480
481		As a neat additional effect, it is no longer necessary (for the sake
482		of lean heap image files) to distinguish between a "minimal" CM and a
483		"full" CM.  Now, there is only one CM (i.e., the "full" version:
484		$smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
485		interactive top level. ($smlnj/cm/minimal.cm is gone.)
486
487		To make the life of compiler-hackers easier, "makeml" now also
488		pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm).  In other
489		words, after you bootstrap a new sml for the first time, you will not
490		have to autoload $smlnj/cmb.cm again afterwards.  (The first time
491		around you will still have to do it, though.)
492
493		* A second change consists of major updates to the CM manual.  There
494		are now several appendices with summary information and also a full
495		specification of the CM description file syntax.
496
497		* In directory src/system I added the script "allcross".  This script
498		invokes sml and cross-compiles the compiler for all supported
499		architectures.  (Useful when providing a new set of boot files.)
500
501		* There seems to be a latent bug in my "lazy pickles" mechanism.  I
502		added a small tweak to pickle-util.sml to work around this problem,
503		but it is not a proper fix yet.  I will investigate further.  (The
504		effect of the bug was an inflation of library pickle size.)
505
506		* Version number increased to 110.28.1 (to avoid compatibility problems).
507
508		----------------------------------------------------------------------
509		Name: Allen Leung
510		Date: 2000/05/25 17:28 EDT
511		Tag: leunga-20000525-ra
512		Description:
513
514		Fixed a bug in freezing phase of the register allocator.
515
516		----------------------------------------------------------------------
517		Name: Allen Leung
518		Date: 2000/05/15 22:53 EDT
519		Tag: leunga-20000515-alpha-x86-ra
520		Description:
521
522		1. Alpha
523
524		Slight cleanup.  Removed the instruction SGNXL
525
526		2. X86
527
528		Added the following instructions to the instruction set:
529
530		ROLx, RORx,
531		BTx, BTSx, BTLx, BTRx,
532		XCHGx, and variants with the LOCK prefix
533
534		3. Register Allocation
535
536		The module ra-rewrite-with-renaming has been improved.
537
538		These have no effect on SML/NJ.
539
540		----------------------------------------------------------------------
541		Name: Matthias Blume
542		Date: 2000/05/15 16:20:00 JST
543		Tag: blume-20000515-lightrebuild
544		Description:
545
546		1. I added an alternative to "-rebuild" to "makeml".  The difference is
547		that prior to calling CMB.make' the CM-variable "LIGHT" will be
548		defined.  In effect, the command will not build any cross-compiler
549		backends and therefore finish more quickly.
550
551		The "fixpt" script also takes a "-light" switch to be able to use
552		this new facility while compiling for a fixpoint.
553
554		2. I replaced all mentions of anchored paths in group owner specifications
555		with simple relative paths (usually starting with "..").
556		The rationale is that a library's internal workings should not be
557		compromised by the lack of some anchor.  (An anchor is necessary
558		for someone who wants to refer to the library by an anchored path,
559		but it should not be necessary to build the same library in the first
560		place.)
561
562		3. I changed the way CM's tool mechanism determines the shell command
563		string used for things like ml-yacc etc. so that it does not break
564		when CM.Control.implicit_anchors is turned off.
565
566		----------------------------------------------------------------------
567		Name: Matthias Blume
568		Date: 2000/05/12 18:20:00 JST
569		Tag: blume-20000512-ml-build
570		Description:
571
572		Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
573		getting installed properly (by config/install.sh).
574
575		----------------------------------------------------------------------
576		Name: Matthias Blume
577		Date: 2000/05/12 17:30:00 JST
578		Tag: blume-20000512-anchors
579		Description:
580
581		!!! NEW BOOT FILES !!!
582
583		This change is in preparation of fading out support for "implicitly
584		anchored path names".  I went through all sources and used the
585		explicit (and relatively new) $-notation.  See system/README and the
586		CM manual for more info on this.
587
588		I also modified the anchoring scheme for some things such as "smlnj",
589		"MLRISC", "cm", etc. to take advantage of the fact that explicit
590		anchors are more expressive: anchor name and first arc do not have to
591		coincide.  This entails the following user-visible change:
592
593		You have to write $smlnj/foo/bar instead of smlnj/foo/bar.  In
594		particular, when you fire up sml with a command-line argument, say,
595		e.g.:
596
597		sml '$smlnj/cmb.cm'
598
599		At the ML toplevel prompt:
600
601		CM.autoload "$smlnj/cmb.cm";
602
603		There is also a new controller in CM.Control that can be used to turn
604		off all remaining support for implicit anchors by saying:
605
606		CM.autoload "$smlnj/
607		#set CM.Control.implicit_anchors false;
608
609		This causes CM to reject implicitly anchored paths.  This is (for the
610		time being) less permissive than the "final" version where there will
611		be no more such implicit anchors and relative paths will be just that:
612		relative.
613
614		The next step (version after next version?) will be to make the
615		default for CM.Control.implicit_anchors false.  After the dust has
616		settled, I can then produce the "final" version of this...
617
618		Note: Since bootstrapping is a bit tricky, I provided new boot files.
619
620		----------------------------------------------------------------------
621		Name: Matthias Blume
622		Date: 2000/05/11 16:30:00 JST
623		Tag: blume-20000511-sources
624		Description:
625
626		The main change is that I added function CM.sources as a generalized
627		version of the earlier CM.makedepend.  This entails the following
628		additional changes:
629
630		- CM.makedepend has been dropped.
631
632		- CM manual has been updated.
633
634		- TOOLS signature and API have been changed.
635
636		----------------------------------------------------------------------
637		Name: Allen Leung
638		Date: 2000/05/10 21:17 EDT
639		Tag: leunga-20000510-moby-c--ssa
640		Description:
641
642		Various bug fixes and new features for C--, Moby and MLRISC optimizations.
643		None of these affect SML/NJ.
644
645		1. Register Allocation
646
647		a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
648		This module tries to remove local (i.e. basic block level) redundancies
649		during spilling.
650
651		b. A new framework for performing region based register allocation.
652		Not yet entirely functional.
653
654		2. X86
655
656		a. DefUse for POP was missing the stack pointer [found by Lal]
657		b. Reload for CALL was incorrect in X86Spill [found by John]
658		c. Various fixes in X86Spill so that it can be used correctly for
659		the new spilling module.
660
661		3. SSA/IR
662
663		a. New module ir/dj-dataflow.sml implements elimination based
664		data flow analysis.
665
666		4. MLRiscGen
667
668		a. Fix for gc type annotation
669
670		5. MDGen
671
672		Various fixes for machine description -> ml code translation.  For ssa
673		only.
674
675		----------------------------------------------------------------------
676		Name: Allen Leung
677		Date: 2000/05/08 22:17 EDT
678		Tag: leunga-20000508-labexp
679		Description:
680
681		Fermin has found a few assembly problems with constant expressions
682		generated in LabelExp.  Mostly, the problems involve extra parentheses,
683		which choke on dumb assemblers.  This is his fix.
684
685		----------------------------------------------------------------------
686		Name: Dave MacQueen
687		Date: 2000/04/09 14:00 EDT
688		Tag: dbm-20000502-Version_110_28
689		Description:
690
691		1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
692
693		2. Updated config/version to 110.28
694
695		3. Updated config/srcarchiveurl
696
697		3. New boot files!
698		ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
699
700		----------------------------------------------------------------------
701		Name: Matthias Blume
702		Date: 2000/05/01 19:05:00 JST
703		Tag: blume-20000501-noweb
704		Description:
705
706		A new noweb tool has been added.  The existing system is entirely
707		unaffected by this, but some CM users have asked for renewed noweb
708		support.  Everything is documented in the CM manual.
709
710		New (plugin) libraries:
711
712		noweb-tool.cm
713		nw-ext.cm
714
715		----------------------------------------------------------------------
716		Name: Dave MacQueen
717		Date: 2000/04/30 12:40PM EDT
718		Tag: dbm-20000430-bug_fixes
719		Description:
720
721		1. Fix for bug 1498
722		smlnj/src/system/Basis/Implementation/Unsafe/object.sig
723		smlnj/src/system/Basis/Implementation/Unsafe/object.sml
724		added toRealArray function
725		smlnj/src/compiler/MiscUtil/print/ppobj.sml
726		added check for tag Obj.RealArray to array printing case in ppObj
727
728		2. Fix for bug 1510
729		smlnj/src/compiler/Semant/types/typesutil.sml
730		fixed definition of dummyargs (used by equalTycon) so that
731		dummy args are distinct types
732
733		----------------------------------------------------------------------
734		Name: Matthias Blume
735		Date: 2000/04/30 01:00:00 JST
736		Tag: blume-20000430-versions
737		Description:
738
739		1. CM version numbering added.  This is an implementation of Lal's
740		proposal for adding version numbers and version checking to .cm
741		files.  Lal said that his proposal was just that -- a proposal.
742		For the time being I went ahead and implemented it so that people
743		can comment on it.  Everything is completely backward-compatible
744		(except for the stable library format, i.e., new bootfiles!).
745
746		As usual, see the CM manual for details.
747
748		2. An alternative syntax for anchored paths has been implemented.
749		Dave has recently voiced the same concerns that I had when I did
750		this, so there should be some support.  My take is that eventually
751		I will let support for the current syntax (where anchors are
752		"implicit") fade out in favor of the new, explicit syntax.
753		In order to be backward-compatible, both old and new syntax are
754		currently supported.
755
756		Again, see the CM manual for details.
757
758		3. Parallel make is trying to be slightly smarter:  When the master
759		process finds a "bottleneck", i.e., when there is only one
760		compilation unit that can be compiled and everybody else is
761		waiting on it, then it will simply compile it directly instead
762		of clumsily telling one of the slaves to do it.
763
764		4. Support for "unsharing" added.  This is necessary in order to be
765		able to have two different versions of the same library running
766		at the same time (e.g., for trying out a new MLRISC while still
767		having the old MLRISC linked into the current compiler, etc.)
768		See the CM manual.
769
770		5. Simple "makedepend" functionality added for generating Makefile
771		dependency information.  (This is rather crude at the moment.
772		Expect some changes here in the future.)
773
774		6. ".fun" added as a recognized suffix for ML files. Also documented
775		explicitly in the manual that the fallback behavior (unknown suffix
776		-> ML file) is not an official feature!
777
778		7. Small changes to the pickler for stable libraries.
779
780		8. Several internal changes to CM (for cleanup/improvement).
781
782
783		!!!! NEW BINFILES !!!!
784
785		----------------------------------------------------------------------
786		Name: Matthias Blume
787		Date: 2000/04/28 17:30:00 JST
788		Tag: blume-20000428-pathconfig
789		Description:
790
791		1. I changed config/install.sh to remove duplicate entries from the
792		lib/pathconfig file at the end.  Moreover, the final version of
793		lib/pathconfig is sorted alphabetically.  The same (sorting) is done
794		in src/system/installml.
795
796		2. The config/install.sh script now consistently uses relative
797		pathnames in lib/pathconfig whenever the anchor is in the lib
798		directory.  (So far this was true for the libraries that come
799		pre-compiled and bundled as part of the bootfiles but not for
800		libraries that are compiled by the script itself.)
801
802		----------------------------------------------------------------------
803		Name: Matthias Blume
804		Date: 2000/04/26 13:10:00 JST
805		Tag: blume-20000426-fun_suffix
806		Description:
807
808		Added ".fun" as a recognized file name suffix (for ML code).
809
810		----------------------------------------------------------------------
811		Name: Allen Leung
812		Date: 2000/04/25 17:00:00 EST
813		Tag: leunga-20000425-alpha-ra
814		Description:
815
816		1. Alpha
817
818		PSEUDOARITH was missing in AlphaRewrite.  This causes an endless loop
819		in C--.
820
821		2. RA
822
823		Added a flag "ra-dump-size" to print out the size of the flowgraph
824		and the interference graph.
825
826		----------------------------------------------------------------------
827		Name: Dave MacQueen
828		Date: 2000/04/25/
829		Tag: dbm-20000425-mlyacc_doc_examples
830		Description:
831		Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
832		Updated all three examples in src/ml-yacc/examples to run
833		under 110.* using CM.make.
834
835		----------------------------------------------------------------------
836		Name: Allen Leung
837		Date: 2000/04/20 23:04:00 EST
838		Tag: leunga-20000420-ssa-c---stuff
839		Description:
840
841		This update synchronizes my repository with Yale's.  Most of these
842		changes, however, do not affect SML/NJ at all (the RA is an exception).
843
844		1. Register Allocator
845
846		a. An improvement in the interference graph construction:
847		Given a copy
848
849		s <- t
850
851		no interference edge between s and t is added for this definition of s.
852
853		b. I've added two new spill heuristic modules that Fermin and I developed
854		(in the new library RA.cm). These are unused in SML/NJ but maybe
855		useful for others (Moby?)
856
857		2. X86
858
859		a. Various fixes in the backend provided by Fermin [C--] and Lal.
860
861		3. Alpha
862
863		a. Added the BSR instruction and code generation that goes with it [C--]
864		b. Other fixes too numerous to recount provided by Fermin [C--]
865
866		4. Regmaps
867
868		a. The regmaps are not initialized with the identity physical bindings
869		at creation time.  This is unneeded.
870
871		5. MLRISC Optimizations
872
873		a. The DJ-Graph module can now compute the iterated dominance frontiers
874		intersects with liveness incrementally in linear time! Woohoo!
875		This is now used in my new SSA construction algorithm.
876
877		b. THe branch reorganization module is now smarter about linear chains of
878		basic blocks.
879
880
881		----------------------------------------------------------------------
882		Name: Matthias Blume
883		Date: 2000/04/12 13:52:00 JST
884		Tag: blume_main_v110p27_1
885		Description:
886
887		Changed install.sh script to handle archive files without version number
888		and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
889		name of the boot file archive.
890
891		----------------------------------------------------------------------
892		Name: Dave MacQueen
893		Date: 2000/04/09 14:00 EDT
894		Tag: dbm-20000410-Version_110_27
895		Description:
896
897		1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
898
899		2. Updated src/config/version to 110.27
900
901		3. New boot files!
902
903		----------------------------------------------------------------------
904		Name: Allen Leung
905		Date: 2000/04/09 19:09:00 EST
906		Tag: leunga-20000409-misc
907		Description:
908
909		1.  Yet another fix for x86 assembly for idivl, imull, mull and friends.
910
911		2.  Miscellaneous improvements to MLRISC (unused in sml/nj)
912
913		----------------------------------------------------------------------
914		Name: Stefan
915		Date: 2000/04/07 10:00:00 EDT
916		Tag: monnier-20000406-branch-handling
917		Description:
918
919		Improved handling of branches (mostly those generated from
920		polymorphic equality), removed switchoff and changed the
921		default optimization settings (more cpsopt and less flintopt).
922
923		----------------------------------------------------------------------
924		Name: Allen Leung
925		Date: 2000/04/06 01:30:00 EST
926		Tag: leunga-20000406-peephole-x86-SSA-2
927		Description:
928
929		Forgot a few files.
930
931		----------------------------------------------------------------------
932		Name: Allen Leung
933		Date: 2000/04/06 00:36:00 EST
934		Tag: leunga-20000406-peephole-x86-SSA
935		Description:
936
937		1.  New Peephole code
938
939		2.  Minor improvement to X86 instruction selection
940
941		3.  Various fixes to SSA and machine description -> code translator
942
943		----------------------------------------------------------------------
944		Name: Matthias Blume
945		Date: 2000/04/05 12:30:00 JST
946		Tag: blume_main_v110p26p2_3
947		Description:
948
949		This update just merges three minor cosmetic updates to CM's sources
950		to get ready for the 110.27 code freeze on Friday.  No functionality
951		has changed.
952
953		----------------------------------------------------------------------
954	Name: Allen Leung	Name: Allen Leung
955	Date: 2000/04/04 19:39:00 EST	Date: 2000/04/04 19:39:00 EST
956	Tag: leunga-20000404-x86-asm	Tag: leunga-20000404-x86-asm
#	Line 57	Line 996
996
997	3.  Assembly	3.  Assembly
998
999	When generating assemby, resolve the value of client defined constants,	When generating assembly, resolve the value of client defined constants,
1000	instead of generating symbolic values.  This is controlled by the	instead of generating symbolic values.  This is controlled by the
1001	new flag "asm-resolve-constants", which is default to true.	new flag "asm-resolve-constants", which is default to true.
1002
#	Line 80	Line 1019
1019
1020	To this end, I arranged that instead of "structure Core" as "structure	To this end, I arranged that instead of "structure Core" as "structure
1021	_Core" is bound in the pervasive environment.  Core access is done via	_Core" is bound in the pervasive environment.  Core access is done via
1022	_Core (which can never be accidentially rebound because _Core is not a	_Core (which can never be accidentally rebound because _Core is not a
1023	legal surface-syntax symbol).	legal surface-syntax symbol).
1024
1025	The current solution is much cleaner because the core environment is	The current solution is much cleaner because the core environment is
#	Line 90	Line 1029
1029	with dynamic and symbolic parts of the core environment.	with dynamic and symbolic parts of the core environment.
1030
1031	Remaining hackery (to bind the "magic" symbol _Core) is localized in the	Remaining hackery (to bind the "magic" symbol _Core) is localized in the
1032	compilation mananger's bootstrap compiler (actually: in the "init group"	compilation manager's bootstrap compiler (actually: in the "init group"
1033	handling).  See the comments in src/system/smlnj/init/init.cmi for	handling).  See the comments in src/system/smlnj/init/init.cmi for
1034	more details.	more details.
1035
#	Line 205	Line 1144
1144	(specified in the .cm file at each instance where the tool's class is	(specified in the .cm file at each instance where the tool's class is
1145	used).	used).
1146
1147	This was done to accomodate the new "make" and "shell" tools which	This was done to accommodate the new "make" and "shell" tools which
1148	facilitate fairly seemless hookup to portions of code managed using	facilitate fairly seamless hookup to portions of code managed using
1149	Makefiles or Shell scripts.	Makefiles or Shell scripts.
1150
1151	There are no classes "shared" or "private" anymore.  Instead, the	There are no classes "shared" or "private" anymore.  Instead, the
#	Line 218	Line 1157
1157
1158	All existing tools are described in the CM manual.	All existing tools are described in the CM manual.
1159
1160	- Slightly better error handling.  (CM now surpresses many followup	- Slightly better error handling.  (CM now suppresses many followup
1161	error messages that tended to be more annoying than helpful.)	error messages that tended to be more annoying than helpful.)
1162
1163	2. Major changes to the compiler's static environment data structures.	2. Major changes to the compiler's static environment data structures.
#	Line 352	Line 1291
1291
1292	I've changed andl to testl in the floating point test sequence	I've changed andl to testl in the floating point test sequence
1293	whenever appropriate.  The Intel optimization guide states that	whenever appropriate.  The Intel optimization guide states that
1294	testl is perferable to andl.	testl is preferable to andl.
1295
1296	7. RA (x86 only)	7. RA (x86 only)
1297
#	Line 534	Line 1473
1473
1474	1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an	1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
1475	additional argument called "template" which is an optional string that	additional argument called "template" which is an optional string that
1476	specifiel the layout of the tool command line.  See the CM manual for	specifies the layout of the tool command line.  See the CM manual for
1477	explanation.	explanation.
1478
1479	2. A special-purpose tool can be "regisitered" by simply dropping the	2. A special-purpose tool can be "registered" by simply dropping the
1480	corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same	corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
1481	directory where the .cm file lives that uses this tool.  (The	directory where the .cm file lives that uses this tool.  (The
1482	behavior/misfeature until now was to look for the tool description	behavior/misfeature until now was to look for the tool description
#	Line 581	Line 1520
1520	rebuild.  Having sets of the form <base><k>.{bin,boot}.<arch>-unix for	rebuild.  Having sets of the form <base><k>.{bin,boot}.<arch>-unix for
1521	<k>=1,2,... is normally not a good idea when invoking fixpt.  However,	<k>=1,2,... is normally not a good idea when invoking fixpt.  However,
1522	they might be the result of an earlier partial run of fixpt (which	they might be the result of an earlier partial run of fixpt (which
1523	perhaps got accidentially killed).  In this case, fixpt will quickly	perhaps got accidentally killed).  In this case, fixpt will quickly
1524	move through what exists before continuing where it left off earlier,	move through what exists before continuing where it left off earlier,
1525	and, thus, saves a lot of time.	and, thus, saves a lot of time.
1526
#	Line 631	Line 1570
1570	it from that remote directory.	it from that remote directory.
1571	This should simplify installation further:  For machines that have	This should simplify installation further:  For machines that have
1572	access to the internet, just fetch <version>-config.tgz, unpack it,	access to the internet, just fetch <version>-config.tgz, unpack it,
1573	edit config/targets, and go (run config/install.sh).  The scipt will	edit config/targets, and go (run config/install.sh).  The script will
1574	fetch everything else that it might need all by itself.	fetch everything else that it might need all by itself.
1575
1576	For CVS users, this mechanism is not relevant for source archives, but	For CVS users, this mechanism is not relevant for source archives, but

---

Colored Diff Long Colored Diff Unidiff Context Diff Side by Side

Legend: Removed from v.594   changed lines   Added in v.677

---

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