Home My Page Projects Code Snippets Project Openings SML/NJ
Summary Activity Forums Tracker Lists Tasks Docs Surveys News SCM Files

SCM Repository

[smlnj] Diff of /sml/trunk/HISTORY
ViewVC logotype

Diff of /sml/trunk/HISTORY

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 578, Tue Mar 14 05:16:29 2000 UTC revision 705, Wed Sep 27 18:44:44 2000 UTC
# Line 11  Line 11 
11  Date:  Date:
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14    
15    ----------------------------------------------------------------------
16    Name: Lal George
17    Date: 2000/09/27 14:42:35 EDT
18    Tag: george-20000927-nodestatus
19    Description:
20    
21    Changed the type of the nodestatus, so that:
22    
23            SPILLED(~1)             is now SPILLED
24            SPILLED(m) where m>=0   is now MEMREG(m)
25            SPILLED(s) where s<~1   is now SPILL_LOC(~s)
26    
27    ----------------------------------------------------------------------
28    Name: Matthias Blume
29    Date: 2000/09/07 14:45:00 JST
30    Tag: blume-20000907-cmerrmsg
31    Description:
32    
33    Small tweak to CM to avoid getting ML syntax error messages twice.
34    
35    ----------------------------------------------------------------------
36    Name: Matthias Blume
37    Date: 2000/08/31 18:00:00 JST
38    Tag: blume-20000831-cvsbootfiles
39    Description:
40    
41    New URL for boot files (because the 110.29 files on the BL server do
42    now work correctly with my updated install scripts for yacc and lex).
43    
44    ----------------------------------------------------------------------
45    Name: Matthias Blume
46    Date: 2000/08/08 12:33:00 JST
47    Tag: blume-20000808-manual
48    Description:
49    
50    Tiny update to CM manual.
51    
52    ----------------------------------------------------------------------
53    Name: Allen Leung
54    Date: 2000/08/7 19:31:00
55    Tag: leunga-20000807-a-whole-bunch-of-stuff
56    
57      Moby, C--, SSA, x86, machine descriptions etc.  Should only affect C--
58    and Mobdy.
59    
60    1.  x86
61    
62       a.  Fixes to peephole module by John and Dan.
63       b.  Assembly fix to SETcc by Allen.
64       c.  Fix to c-call by John.
65       d.  Fix to spilling by John.  (This one deals with the missing FSTPT case)
66       e.  Instruction selection optimization to SETcc as suggested by John.
67    
68           For example,
69    
70            MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
71    
72           should generate:
73    
74            MOVL a, x
75            SUBL b, x
76            SHRL 31, x
77    
78    2.  IR stuff
79    
80         A bunch of new DJ-graph related algorithms added.  These
81         speed up SSA construction.
82    
83    3.  SSA + Scheduling
84    
85         Added code for SSA and scheduling to the repository
86    
87    ----------------------------------------------------------------------
88    Name: Lal George
89    Date: 2000/07/27 11:53:14 EDT
90    
91    Tag: lal-20000727-linux-ppc
92    Description:
93    
94     Made changes to support Linux PPC.
95     p.s. I have confirmation that the 110.29 boot files work fine.
96    
97    ----------------------------------------------------------------------
98    Name: Matthias Blume
99    Date: 2000/07/27 17:40:00 JST
100    Tag: blume-20000727-scripts
101    Description:
102    
103    !!!! WARNING !!!!
104    You must recompile the runtime system!
105    !!!! WARNING !!!!
106    
107    This is basically another round of script-enhancements:
108    
109    1. sml, ml-build, and ml-makedepend accept options -D and -U to define
110       and undefine CM preprocessor symbols.
111    
112    2. ml-build avoids generating a new heap image if it finds that the
113       existing one is still ok.  (The condition is that no ML file had to
114       be recompiled and all ML files are found to be older that the heap
115       file.)
116    
117       To make this work smoothly, I also hacked the runtime system as
118       well as SMLofNJ.SysInfo to get access to the heap image suffix
119       (.sparc-solaris, ...) that is currently being used.
120    
121       Moreover, the signature of CM.mk_standalone has changed.  See the
122       CM manual.
123    
124    3. ml-makedepend accepts additional options -n, -a, and -o.  (See the
125       CM manual for details.)
126    
127    4. More CM manual updates:
128        - all of the above has been documented.
129        - there is now a section describing the (CM-related) command line
130          arguments that are accepted by the "sml" command
131    
132    ----------------------------------------------------------------------
133    Name: Matthias Blume
134    Date: 2000/07/25 16:20:00 JST
135    Tag: blume-20000725-makedepend
136    Description:
137    
138    Added a script called ml-makedepend.  This can be used in makefiles
139    for Unix' make in a way very similar to the "makedepend" command for
140    C.
141    
142    The script internally uses function CM.sources.
143    
144    Synopsis:
145    
146        ml-makedepend [-f makefile] cmfile targetname
147    
148    The default for the makefile is "makefile" (or "Makefile" should
149    "makefile" not exist).
150    
151    ml-makedepend adds a cmfile/targetname-specific section to this
152    makefile (after removing the previous version of this section).  The
153    section contains a single dependency specification with targetname on
154    the LHS (targetname is an arbitrary name), and a list of files derived
155    from the cmfile on the RHS.  Some of the files on the RHS are
156    ARCH/OPSYS-specific.  Therefore, ml-makedepend inserts references to
157    "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
158    path names.  The makefile writer is responsible for making sure that
159    these variables have correct at the time "make" is invoked.
160    
161    ----------------------------------------------------------------------
162    Name: Matthias Blume
163    Date: 2000/07/22 23:30:00 JST
164    Tag: blume-20000722-urlupdate
165    Description:
166    
167    Changed BOOT and config/srcarchiveurl to point to BL server:
168    
169        ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
170    
171    ----------------------------------------------------------------------
172    Name: Matthias Blume
173    Date: 2000/07/18 18:00:00 JST
174    Tag: blume-20000718-Version_110_29
175    Description:
176    
177    1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
178    
179    2. Updated config/version to 110.29
180    
181    3. Updated config/srcarchiveurl
182    
183    3. New boot files!
184       ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
185    
186    ----------------------------------------------------------------------
187    Name: Matthias Blume
188    Date: 2000/07/11 13:58:00 JST
189    Tag: blume-20000711-doctypo
190    Description:
191    
192    Fixed a few typos in CM manual.
193    
194    ----------------------------------------------------------------------
195    Name: Allen Leung
196    Date: 2000/06/15 00:38:00
197    Tag: leunga-20000704-sparc-x86
198    
199    1. x86 peephole improvement sp += k; sp -= k => nop  [from John]
200    2. fix to x86 RET bug [found by Dan Grossman]
201    3. sparc assembly bug fix for ticc instructions [found by Fermin]
202    
203       Affects c-- and moby only
204    
205    ----------------------------------------------------------------------
206    Name: Matthias Blume
207    Date: 2000/07/04 15:26:00
208    Tag: blume-20000704-trigger
209    Description:
210    
211    1. Improvements to CM manual.
212    2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
213       of getting a back-trace.  The function, when called, raises an
214       internal exception which explicitly carries the full back-trace history,
215       so it is unaffected by any intervening handle-raise pairs ("trivial"
216       or not).  The interactive loop will print that history once it arrives
217       at top level.
218       Short of having all exceptions implicitly carry the full history, the
219       recommended way of using this facility is:
220         - compile your program with instrumentation "on"
221         - run it, when it raises an exception, look at the history
222         - if the history is "cut off" because of some handler, go and modify
223           your program so that it explicitly calls BTrace.trigger
224         - recompile (still instrumented), and rerun; look at the full history
225    
226    ----------------------------------------------------------------------
227    Name: Matthias Blume
228    Date: 2000/07/03 15:36:00 JST
229    Tag: blume-20000702-manual
230    Description:
231    
232    Small corrections and updates to CM manual.
233    
234    ----------------------------------------------------------------------
235    Name: Matthias Blume
236    Date: 2000/06/29 16:04:00 JST
237    Tag: blume-20000629-yacctool
238    Description:
239    
240    Changes:
241    
242    1. Class "mlyacc" now takes separate arguments to pass options to
243       generated .sml- and .sig-files independently.
244    2. Corresponding CM manual updates.
245    3. BTrace module now also reports call sites.  (However, for loop clusters
246       it only shows from where the cluster was entered.)  There are associated
247       modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
248    
249    ----------------------------------------------------------------------
250    Name: Matthias Blume
251    Date: 2000/06/27 16:51:00 JST
252    Tag: blume-20000627-noweb
253    Description:
254    
255    Changes:
256    
257     1. Implemented "subdir" and "witness" options for noweb tool.
258        This caused some slight internal changes in CM's tool implementation.
259     2. Fixed bug in "tool plugin" mechanism.  This is essentially cleaning
260        some remaining issues from earlier path anchor changes.
261     3. Updated CM manual accordingly.
262    
263     4. Changed implementation of back-tracing so that I now consider it
264        ready for prime-time.
265    
266        In particular, you don't have to explicitly trigger the back-trace
267        anymore.  Instead, if you are running BTrace-instrumented code and
268        there is an uncaught exception (regardless of whether or not it was
269        raised in instrumented code), the top-level evalloop will print
270        the back-trace.
271    
272        Features:
273    
274          - Instrumented and uninstrumented code work together seemlessly.
275            (Of course, uninstrumented code is never mentioned in actual
276             back-traces.)
277    
278          - Asymptotic time- and space-complexity of instrumented code is
279            equal to that of uninstrumented code.  (This means that
280            tail-recursion is preserved by the instrumentation phase.)
281    
282          - Modules whose code has been instrumented in different sessions
283            work together without problem.
284    
285          - There is no penalty whatsoever on uninstrumented code.
286    
287          - There is no penalty on "raise" expressions, even in
288            instrumented code.
289    
290        A potential bug (or perhaps it is a feature, too):
291    
292          A back-trace reaches no further than the outermost instrumented
293          non-trivial "raise".  Here, a "trivial" raise is one that is the
294          sole RHS of a "handle" rule.  Thus, back-traces reach trough
295    
296               <exp> handle e => raise e
297    
298          and even
299    
300               <exp> handle Foo => raise Bar
301    
302          and, of course, through
303    
304               <exp> handle Foo => ...
305    
306         if the exception was not Foo.
307    
308         Back-traces always reach right through any un-instrumented code
309         including any of its "handle" expressions, trivial or not.
310    
311       To try this out, do the following:
312    
313         - Erase all existing binfiles for your program.
314           (You may keep binfiles for those modules where you think you
315            definitely don't need back-tracing.)
316         - Turn on back-trace instrumentation:
317              SMLofNJ.Internals.BTrace.mode (SOME true);
318         - Recompile your program.  (I.e., run "CM.make" or "use".)
319         - You may now turn instrumentation off again (if you want):
320              SMLofNJ.Internals.BTrace.mode (SOME false);
321         - Run your program as usual.  If it raises an exception that
322           reaches the interactive toplevel, then a back-trace will
323           automatically be printed.  After that, the toplevel loop
324           will print the exception history as usual.
325    
326    ----------------------------------------------------------------------
327    Name: Matthias Blume
328    Date: 2000/06/26 09:56:46 JST
329    Tag: blume-20000626-setup
330    Description:
331    
332    CM: - setup-parameter to "sml" added; this can be used to run arbitrary
333          ML code before and after compiling a file (e.g., to set compiler
334          flags)
335    
336    Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
337              - associated changes to btrace.sml (BTrace instrumentation pass)
338              - cleaner implementation of btimp.sml (BTrace tracing and report
339                module)
340    
341    CM manual: * new path encoding documented
342               * description of setup-parameter to "sml" added
343    
344    The biggest user-visible change to back-tracing is that it is no
345    longer necessary to compile all traced modules within the same
346    session.  (This was a real limitation.)
347    
348    ----------------------------------------------------------------------
349    Name: Matthias Blume
350    Date: 2000/06/24 12:40:00 JST
351    Tag: blume-20000624-startup
352    Description:
353    
354    Fixes startup slowdown problem.  (I was calling SrcPath.sync a _tad_
355    bit too often -- to put it mildly. :)
356    
357    ----------------------------------------------------------------------
358    Name: Matthias Blume
359    Date: 2000/06/23 18:20:00 JST
360    Tag: blume-20000623-btrace
361    Description:
362    
363    This updates adds a backtrace facility to aid programmers in debugging
364    their programs.  This involves the following changes:
365    
366    1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
367       keeping track of the current call stack.  When programs are compiled
368       in a special mode, the compiler will insert calls to these hooks
369       into the user program.
370       "Hook" means that it is possible for different implementations of
371       back-tracing to register themselves (at different times).
372    
373    2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
374       as an Absyn.dec->Absyn.dec rewrite.  Normally this phase is turned off.
375       It can be turned on using this call:
376         SMLofNJ.Internals.BTrace.mode (SOME true);
377       Turning it off again:
378         SMLofNJ.Internals.BTrace.mode (SOME false);
379       Querying the current status:
380         SMLofNJ.Internals.BTrace.mode NONE;
381       Annotated programs are about twice as big as normal ones, and they
382       run a factor of 2 to 4 slower with a dummy back-trace plugin (one
383       where all hooks do nothing).  The slowdown with a plugin that is
384       actually useful (such as the one supplied by default) is even greater,
385       but in the case of the default plugin it is still only an constant
386       factor (amortized).
387    
388    3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
389       with a sub-structure BTrace for controlling back-tracing.  In particular,
390       the above-mentioned function "mode" controls whether the annotation
391       phase is invoked by the compiler.  Another important function is
392       "trigger": when called it aborts the current execution and causes
393       the top-level loop to print a full back-trace.
394    
395    4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
396       for back-tracing.  It keeps track of the dynamic call stack and in
397       addition to that it keeps a partial history at each "level" of that
398       stack.  For example, if a tail-calls b, b tail-calls c, and c tail-calls
399       d and b (at separate times, dynamically), then the report will show:
400    
401       GOTO   d
402             /c
403       GOTO  \b
404       CALL   a
405    
406       This shows that there was an initial non-tail call of a, then a
407       tail-call to b or c, looping behavior in a cluster of functions that
408       consist of b and c, and then a goto from that cluster (i.e., either from
409       b or from c) to d.
410    
411       Note that (depending on the user program) the amount of information
412       that the back-trace module has to keep track of at each level is bounded
413       by a constant.  Thus, the whole implementation has the same asymptotical
414       complexity as the original program (both in space and in time).
415    
416    5. compiler/TopLevel/interact/evalloop.sml has been modified to
417       handle the special exception SMLofNJ.Internals.BTrace.BTrace
418       which is raised by the "trigger" function mentioned above.
419    
420    Notes on usage:
421    
422    - Annotated code works well together with unannotated code:
423    Unannotated calls simply do not show up at all in the backtrace.
424    
425    - It is not a good idea to let modules that were annotated during
426    different sessions run at the same time.  This is because the compiler
427    chooses small integers to identify individual functions, and there
428    will be clashes if different modules were compiled in separate sessions.
429    (Nothing will crash, and you will even be told about the clashes, but
430    back-trace information will in general not be useful.)
431    
432    - Back-tracing can be confused by callcc and capture.
433    
434    - The only way of getting a back-trace right now is to explicitly
435    invoke the "trigger" function from your user program.  Eventually, we
436    should make every exception carry back-trace information (if
437    available).  But since this creates more overhead at "raise"-time
438    (similar to the current exnHistory overhead), I have not yet
439    implemented this.  (The implementation will be rather easy.)  With
440    exceptions carrying back-trace information, this facility will be even
441    more useful because users don't need to modify their programs...
442    
443    - While it is possible to compile the compiler with back-trace
444    annotations turned on (I did it to get some confidence in
445    correctness), you must make absolutely sure that core.sml and
446    btimp.sml are compiled WITHOUT annotation!  (core.sml cannot actually
447    be compiled with annotation because there is no core access yet, but
448    if you compile btimp.sml with annotation, then the system will go into
449    an infinite recursion and crash.)
450    Since CM currently does not know about BTrace, the only way to turn
451    annotations on and off for different modules of the compiler is to
452    interrupt CMB.make, change the settings, and re-invoke it.  Of course,
453    this is awkward and clumsy.
454    
455    Sample sessions:
456    
457    Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
458    - SMLofNJ.Internals.BTrace.mode (SOME true);
459    [autoloading]
460    [autoloading done]
461    val it = false : bool
462    - structure X = struct
463    -     fun main n = let
464    -         fun a (x, 0) = d x
465    -           | a (x, n) = b (x, n - 1)
466    -         and b (x, n) = c (x, n)
467    -         and c (x, n) = a (x, n)
468    -         and d x = e (x, 3)
469    -         and e (x, 0) = f x
470    -           | e (x, n) = e (x, n - 1)
471    -         and f 0 = SMLofNJ.Internals.BTrace.trigger ()
472    -           | f n = n * g (n - 1)
473    -         and g n = a (n, 3)
474    -     in
475    -         f n
476    -     end
477    - end;
478    structure X : sig val main : int -> int end
479    - X.main 3;
480    *** BACK-TRACE ***
481    GOTO   stdIn:4.2-13.20: X.main[2].f
482    GOTO-( stdIn:4.2-13.20: X.main[2].e
483    GOTO   stdIn:4.2-13.20: X.main[2].d
484         / stdIn:4.2-13.20: X.main[2].a
485         | stdIn:4.2-13.20: X.main[2].b
486    GOTO-\ stdIn:4.2-13.20: X.main[2].c
487    CALL   stdIn:4.2-13.20: X.main[2].g
488    GOTO   stdIn:4.2-13.20: X.main[2].f
489    GOTO-( stdIn:4.2-13.20: X.main[2].e
490    GOTO   stdIn:4.2-13.20: X.main[2].d
491         / stdIn:4.2-13.20: X.main[2].a
492         | stdIn:4.2-13.20: X.main[2].b
493    GOTO-\ stdIn:4.2-13.20: X.main[2].c
494    CALL   stdIn:4.2-13.20: X.main[2].g
495    GOTO   stdIn:4.2-13.20: X.main[2].f
496    GOTO-( stdIn:4.2-13.20: X.main[2].e
497    GOTO   stdIn:4.2-13.20: X.main[2].d
498         / stdIn:4.2-13.20: X.main[2].a
499         | stdIn:4.2-13.20: X.main[2].b
500    GOTO-\ stdIn:4.2-13.20: X.main[2].c
501    CALL   stdIn:4.2-13.20: X.main[2].g
502    GOTO   stdIn:4.2-13.20: X.main[2].f
503    CALL   stdIn:2.15-17.4: X.main[2]
504    -
505    
506    (Note that because of a FLINt bug the above code currently does not
507    compile without BTrace turned on.)
508    
509    Here is another example, using my modified Tiger compiler:
510    
511    Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
512    - SMLofNJ.Internals.BTrace.mode (SOME true);
513    [autoloading]
514    [autoloading done]
515    val it = false : bool
516    - CM.make "sources.cm";
517    [autoloading]
518    ...
519    [autoloading done]
520    [scanning sources.cm]
521    [parsing (sources.cm):parse.sml]
522    [creating directory CM/SKEL ...]
523    [parsing (sources.cm):tiger.lex.sml]
524    ...
525    [wrote CM/sparc-unix/semant.sml]
526    [compiling (sources.cm):main.sml]
527    [wrote CM/sparc-unix/main.sml]
528    [New bindings added.]
529    val it = true : bool
530    - Main.compile ("../testcases/merge.tig", "foo.out");
531    *** BACK-TRACE ***
532    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
533    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
534    CALL   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
535    GOTO   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
536    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
537    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
538    CALL   lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
539         / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
540    CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
541    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
542    CALL   lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
543    CALL   main.sml:1.18-118.4: Main.compile[2]
544    -
545    
546    ----------------------------------------------------------------------
547    Name: Matthias Blumen
548    Date: 2000/06/21 18:00:00 JST
549    Tag: blume-20000621-manual
550    Description:
551    
552    CM manual update: Path environments documented.
553    
554    ----------------------------------------------------------------------
555    Name: Matthias Blume
556    Date: 2000/06/19 13:40:00
557    Tag: blume-20000619-manual
558    Description:
559    
560    CM manual and system/README update.  This only covers the fact that
561    there are no more implicit anchors.  (Path environments and the "bind"
562    option to "cm" have yet to be documented.)
563    
564    ----------------------------------------------------------------------
565    Name: Matthias Blume
566    Date: 2000/06/19 11:05:00 JST
567    Tag: blume-20000619-chdir-bugfix
568    Description:
569    
570    Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
571    
572    ----------------------------------------------------------------------
573    Name: Matthias Blume
574    Date: 2000/06/18 22:00:10 JST
575    Tag: blume-20000618-implicit-anchors-really-gone
576    Description:
577    
578    I updates the previous HISTORY entry where I forgot to mention that
579    implicit anchors are no longer with us.
580    
581    The current update also gets rid of the (now useless) controller
582    CM.Control.implicit_anchors.
583    
584    ----------------------------------------------------------------------
585    Name: Matthias Blume
586    Date: 2000/06/16 17:30:00 JST
587    Tag: blume-20000616-anchorenv
588    Description:
589    
590    This patch implements the long anticipated (just kidding :) "anchor
591    environment" mechanism.  In the course of doing this, I also
592    re-implemented CM's internal "SrcPath" module from scratch.  The new
593    one should be more robust in certain boundary cases.  In any case, it
594    is a lot cleaner than its predecessor (IMHO).
595    
596    This time, although there is yet another boot file format change, I
597    kept the unpickler backward-compatible.  As a result, no new bootfiles
598    are necessary and bootstrapping is straightforward.  (You cannot read
599    new bootfiles into an old system, but the other way around is no
600    problem.)
601    
602    Visible changes:
603    
604    ** 0. Implicit path anchors (without the leading $-symbol) are no
605    longer recognized at all. This means that such path names are not
606    illegal either.  For example, the name basis.cm simply refers to a
607    local file called "basis.cm" (i.e, the name is an ordinary path
608    relative to .cm-files directory).  Or, to put it differently, only
609    names that start with $ are anchored paths.
610    
611    ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
612    vanished.
613    
614    John (Reppy) had critizised this as soon as I originally proposed and
615    implemented it, but at that time I did not really deeply believe
616    him. :) Now I came full-circle because I need the $<singlearc> syntax
617    in another place where it cannot be seen as an abbreviation for
618    $/<singlearc>.  To avoid the confusion, $<singlearc> now means what it
619    seems to mean (i.e., it "expands" into the corresponding anchor
620    value).
621    
622    However, when paths are used as members in CM description files, it
623    continues to be true that there must be at least another arc after the
624    anchor.  This is now enforced separately during semantic analysis
625    (i.e., from a lexical/syntactical point of view, the notation is ok.)
626    
627    ** 2. The "cm" class now accepts an option "bind".  The option's value
628    is a sub-option list of precisely two items -- one labeled "anchor"
629    and the other one labeled "value".  As you might expect, "anchor" is
630    used to specify an anchor name to be bound, and "value" specifies what
631    the anchor is being bound to.
632    
633    The value must be a directory name and can be given in either standard
634    syntax (including the possibility that it is itself an anchored path)
635    or native syntax.
636    
637    Examples:
638    
639       foo.cm (bind:(anchor:bar value:$mystuff/bar))
640       lib.cm (bind:(anchor:a value:"H:\\x\\y\\z"))  (* only works under windows *)
641    
642    and so on.
643    
644    The meaning of this is that the .cm-file will be processed with an
645    augmented anchor environment where the given anchor(s) is/are bound to
646    the given values(s).
647    
648    The rationale for having this feature is this: Suppose you are trying
649    to use two different (already stable) libraries a.cm and b.cm (that
650    you perhaps didn't write yourself).  Further, suppose each of these
651    two libraries internally uses its own auxiliary library $aux/lib.cm.
652    Normally you would now have a problem because the anchor "lib" can not
653    be bound to more than one value globally.  Therefore, the project that
654    uses both a.cm and b.cm must locally redirect the anchor to some other
655    place:
656    
657       a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
658       b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
659    
660    This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
661    /usr/lib/smlnj/b-stuff/aux.cm, respectively.
662    
663    Hard-wiring path names is a bit inflexible (and CM will verbosely warn
664    you when you do so at the time of CM.stabilize).  Therefore, you can
665    also use an anchored path as the value:
666    
667      a.cm (bind:(anchor:lib value:$a-lib))
668      b.cm (bind:(anchor:lib value:$b-lib))
669    
670    Now you can globally configure (using the usual CM.Anchor.anchor or
671    pathconfig machinery) bindings for "a-lib" and "b-lib".  Since "lib"
672    itself is always locally bound, setting it globally is no longer
673    meaningful or necessary (but it does not hurt either).  In fact, "lib"
674    can still be used as a global anchor for separate purposes.  As a
675    matter of fact, one can locally define "lib" in terms of a global
676    "lib":
677    
678      a.cm (bind:(anchor:lib value:$lib/a))
679      b.cm (bind:(anchor:lib value:$lib/b))
680    
681    ** 3: The encoding of path names has changed.  This affects the way
682    path names are shown in CM's progress report and also the internal
683    protocol encoding used for parallel make.
684    
685    The encoding now uses one or more ':'-separated segments.  Each
686    segments corresponds to a file that has been specified relative to the
687    file given by its preceding segment.  The first segment is either
688    relative to the CWD, absolute, or anchored.  Each segment itself is
689    basically a Unix pathname; all segments but the first are relative.
690    
691    Example:
692    
693       $foo/bar/baz.cm:a/b/c.sml
694    
695    This path denotes the file bar/a/b/c.sml relative to the directory
696    denoted by anchor "foo".  Notice that the encoding also includes
697    baz.cm which is the .cm-file that listed a/b/c.sml.  As usual, such
698    paths are resolved relative to the .cm-files directory, so baz.cm must
699    be ignored to get the "real" pathname.
700    
701    To make this fact more obvious, CM puts the names of such "virtual
702    arcs" into parentheses when they appear in progress reports. (No
703    parentheses will appear in the internal protocol encoding.)  Thus,
704    what you really see is:
705    
706      $foo/bar/(baz.cm):a/b/c.sml
707    
708    I find this notation to be much more informative than before.
709    
710    Another new feature of the encoding is that special characters
711    including parentheses, colons, (back)slashes, and white space are
712    written as \ddd (where ddd is the decimal encoding of the character).
713    
714    *** The CM manual still needs to be updated.
715    
716    ----------------------------------------------------------------------
717    Name: Allen Leung
718    Date: 2000/06/15 00:38:00
719    Tag: leunga-20000615-x86-peephole
720    
721    x86 Peephole fix by Fermin.  Affects c-- and moby only.
722    
723    ----------------------------------------------------------------------
724    Name: Matthias Blume
725    Date: 2000/06/12 11:40:00
726    Tag: blume-20000612-parmakefix
727    Description:
728    
729    More cleanup after changing the file naming scheme: This time I
730    repaired the parallel make mechanism for CMB.make which I broke earlier.
731    
732    ----------------------------------------------------------------------
733    Name: Allen Leung
734    Date: 2000/06/09 01:25:00
735    Tag: leunga-20000609-various
736    
737    None of these things should affect normal SML/NJ operations
738    
739    1. Peephole improvements provided by Fermin (c--)
740    2. New annotation DEFUSE for adding extra dependence (moby)
741    3. New X86 LOCK instructions (moby)
742    4. New machine description language for reservation tables (scheduling)
743    5. Fixes to various optimization/analysis modules (branch chaining, dominator
744       trees etc.)
745    6. I've changed the CM files so that they can work with versions
746       110.0.6, 110.25 and 110.28
747    
748    ----------------------------------------------------------------------
749    Name: Matthias Blume
750    Date: 2000/06/09 12:40:00
751    Tag: blume-20000609-log
752    Description:
753    
754    - Removed all(?) remaining RCS Log entries from sources.
755    
756    - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
757      anchored paths).
758    
759    ----------------------------------------------------------------------
760    Name: Matthias Blume
761    Date: 2000/06/07 17:00:00 JST
762    Tag: blume-20000607-no-implicit-anchors
763    Description:
764    
765    1. This update changes the default setting for
766    CM.Control.implicit_anchors from true to false.  This means that
767    implicit anchors are no longer permitted by default.  I also tried to
768    make sure that nothing else still relies on implicit anchors.
769    (This is the next step on the schedule towards a CM that does not even
770    have the notion of implicit anchors anymore.)
771    
772    2. More CM manual updates.
773    
774    3. I managed to track down and fix the pickling bug I mentioned last
775    time.  Because of the previously existing workaround, this entails no
776    immediate practical changes.
777    
778    ----------------------------------------------------------------------
779    Name: Matthias Blume
780    Date: 2000/06/06 11:15:00 JST
781    Tag: blume-20000606-lazierpickle
782    Description:
783    
784    !!!! NEW BOOT FILES !!!!
785    
786    * The main purpose of this update is to make library pickles lazier in
787    order to reduce the initial space penalty for autoloading a library.
788    As a result, it is now possible to have $smlnj/compiler.cm
789    pre-registered.  This should take care of the many complaints or
790    inquiries about missing structure Compiler.  This required changes to
791    CM's internal data structures and small tweaks to some algorithms.
792    
793    As a neat additional effect, it is no longer necessary (for the sake
794    of lean heap image files) to distinguish between a "minimal" CM and a
795    "full" CM.  Now, there is only one CM (i.e., the "full" version:
796    $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
797    interactive top level. ($smlnj/cm/minimal.cm is gone.)
798    
799    To make the life of compiler-hackers easier, "makeml" now also
800    pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm).  In other
801    words, after you bootstrap a new sml for the first time, you will not
802    have to autoload $smlnj/cmb.cm again afterwards.  (The first time
803    around you will still have to do it, though.)
804    
805    * A second change consists of major updates to the CM manual.  There
806    are now several appendices with summary information and also a full
807    specification of the CM description file syntax.
808    
809    * In directory src/system I added the script "allcross".  This script
810    invokes sml and cross-compiles the compiler for all supported
811    architectures.  (Useful when providing a new set of boot files.)
812    
813    * There seems to be a latent bug in my "lazy pickles" mechanism.  I
814    added a small tweak to pickle-util.sml to work around this problem,
815    but it is not a proper fix yet.  I will investigate further.  (The
816    effect of the bug was an inflation of library pickle size.)
817    
818    * Version number increased to 110.28.1 (to avoid compatibility problems).
819    
820    ----------------------------------------------------------------------
821    Name: Allen Leung
822    Date: 2000/05/25 17:28 EDT
823    Tag: leunga-20000525-ra
824    Description:
825    
826      Fixed a bug in freezing phase of the register allocator.
827    
828    ----------------------------------------------------------------------
829    Name: Allen Leung
830    Date: 2000/05/15 22:53 EDT
831    Tag: leunga-20000515-alpha-x86-ra
832    Description:
833    
834      1. Alpha
835    
836          Slight cleanup.  Removed the instruction SGNXL
837    
838      2. X86
839    
840          Added the following instructions to the instruction set:
841    
842            ROLx, RORx,
843            BTx, BTSx, BTLx, BTRx,
844            XCHGx, and variants with the LOCK prefix
845    
846      3. Register Allocation
847    
848          The module ra-rewrite-with-renaming has been improved.
849    
850      These have no effect on SML/NJ.
851    
852    ----------------------------------------------------------------------
853    Name: Matthias Blume
854    Date: 2000/05/15 16:20:00 JST
855    Tag: blume-20000515-lightrebuild
856    Description:
857    
858    1. I added an alternative to "-rebuild" to "makeml".  The difference is
859       that prior to calling CMB.make' the CM-variable "LIGHT" will be
860       defined.  In effect, the command will not build any cross-compiler
861       backends and therefore finish more quickly.
862    
863       The "fixpt" script also takes a "-light" switch to be able to use
864       this new facility while compiling for a fixpoint.
865    
866    2. I replaced all mentions of anchored paths in group owner specifications
867       with simple relative paths (usually starting with "..").
868       The rationale is that a library's internal workings should not be
869       compromised by the lack of some anchor.  (An anchor is necessary
870       for someone who wants to refer to the library by an anchored path,
871       but it should not be necessary to build the same library in the first
872       place.)
873    
874    3. I changed the way CM's tool mechanism determines the shell command
875       string used for things like ml-yacc etc. so that it does not break
876       when CM.Control.implicit_anchors is turned off.
877    
878    ----------------------------------------------------------------------
879    Name: Matthias Blume
880    Date: 2000/05/12 18:20:00 JST
881    Tag: blume-20000512-ml-build
882    Description:
883    
884    Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
885    getting installed properly (by config/install.sh).
886    
887    ----------------------------------------------------------------------
888    Name: Matthias Blume
889    Date: 2000/05/12 17:30:00 JST
890    Tag: blume-20000512-anchors
891    Description:
892    
893    !!! NEW BOOT FILES !!!
894    
895    This change is in preparation of fading out support for "implicitly
896    anchored path names".  I went through all sources and used the
897    explicit (and relatively new) $-notation.  See system/README and the
898    CM manual for more info on this.
899    
900    I also modified the anchoring scheme for some things such as "smlnj",
901    "MLRISC", "cm", etc. to take advantage of the fact that explicit
902    anchors are more expressive: anchor name and first arc do not have to
903    coincide.  This entails the following user-visible change:
904    
905    You have to write $smlnj/foo/bar instead of smlnj/foo/bar.  In
906    particular, when you fire up sml with a command-line argument, say,
907    e.g.:
908    
909       sml '$smlnj/cmb.cm'
910    
911    At the ML toplevel prompt:
912    
913       CM.autoload "$smlnj/cmb.cm";
914    
915    There is also a new controller in CM.Control that can be used to turn
916    off all remaining support for implicit anchors by saying:
917    
918        CM.autoload "$smlnj/
919        #set CM.Control.implicit_anchors false;
920    
921    This causes CM to reject implicitly anchored paths.  This is (for the
922    time being) less permissive than the "final" version where there will
923    be no more such implicit anchors and relative paths will be just that:
924    relative.
925    
926    The next step (version after next version?) will be to make the
927    default for CM.Control.implicit_anchors false.  After the dust has
928    settled, I can then produce the "final" version of this...
929    
930    Note: Since bootstrapping is a bit tricky, I provided new boot files.
931    
932    ----------------------------------------------------------------------
933    Name: Matthias Blume
934    Date: 2000/05/11 16:30:00 JST
935    Tag: blume-20000511-sources
936    Description:
937    
938    The main change is that I added function CM.sources as a generalized
939    version of the earlier CM.makedepend.  This entails the following
940    additional changes:
941    
942      - CM.makedepend has been dropped.
943    
944      - CM manual has been updated.
945    
946      - TOOLS signature and API have been changed.
947    
948    ----------------------------------------------------------------------
949    Name: Allen Leung
950    Date: 2000/05/10 21:17 EDT
951    Tag: leunga-20000510-moby-c--ssa
952    Description:
953    
954      Various bug fixes and new features for C--, Moby and MLRISC optimizations.
955    None of these affect SML/NJ.
956    
957    1. Register Allocation
958    
959        a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
960           This module tries to remove local (i.e. basic block level) redundancies
961           during spilling.
962    
963        b. A new framework for performing region based register allocation.
964           Not yet entirely functional.
965    
966    2. X86
967    
968       a. DefUse for POP was missing the stack pointer [found by Lal]
969       b. Reload for CALL was incorrect in X86Spill [found by John]
970       c. Various fixes in X86Spill so that it can be used correctly for
971          the new spilling module.
972    
973    3. SSA/IR
974    
975       a. New module ir/dj-dataflow.sml implements elimination based
976          data flow analysis.
977    
978    4. MLRiscGen
979    
980       a. Fix for gc type annotation
981    
982    5. MDGen
983    
984       Various fixes for machine description -> ml code translation.  For ssa
985       only.
986    
987    ----------------------------------------------------------------------
988    Name: Allen Leung
989    Date: 2000/05/08 22:17 EDT
990    Tag: leunga-20000508-labexp
991    Description:
992    
993      Fermin has found a few assembly problems with constant expressions
994      generated in LabelExp.  Mostly, the problems involve extra parentheses,
995      which choke on dumb assemblers.  This is his fix.
996    
997    ----------------------------------------------------------------------
998    Name: Dave MacQueen
999    Date: 2000/04/09 14:00 EDT
1000    Tag: dbm-20000502-Version_110_28
1001    Description:
1002    
1003    1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
1004    
1005    2. Updated config/version to 110.28
1006    
1007    3. Updated config/srcarchiveurl
1008    
1009    3. New boot files!
1010       ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
1011    
1012    ----------------------------------------------------------------------
1013    Name: Matthias Blume
1014    Date: 2000/05/01 19:05:00 JST
1015    Tag: blume-20000501-noweb
1016    Description:
1017    
1018    A new noweb tool has been added.  The existing system is entirely
1019    unaffected by this, but some CM users have asked for renewed noweb
1020    support.  Everything is documented in the CM manual.
1021    
1022    New (plugin) libraries:
1023    
1024       noweb-tool.cm
1025       nw-ext.cm
1026    
1027    ----------------------------------------------------------------------
1028    Name: Dave MacQueen
1029    Date: 2000/04/30 12:40PM EDT
1030    Tag: dbm-20000430-bug_fixes
1031    Description:
1032    
1033    1. Fix for bug 1498
1034       smlnj/src/system/Basis/Implementation/Unsafe/object.sig
1035       smlnj/src/system/Basis/Implementation/Unsafe/object.sml
1036         added toRealArray function
1037       smlnj/src/compiler/MiscUtil/print/ppobj.sml
1038         added check for tag Obj.RealArray to array printing case in ppObj
1039    
1040    2. Fix for bug 1510
1041       smlnj/src/compiler/Semant/types/typesutil.sml
1042         fixed definition of dummyargs (used by equalTycon) so that
1043         dummy args are distinct types
1044    
1045    ----------------------------------------------------------------------
1046    Name: Matthias Blume
1047    Date: 2000/04/30 01:00:00 JST
1048    Tag: blume-20000430-versions
1049    Description:
1050    
1051    1. CM version numbering added.  This is an implementation of Lal's
1052       proposal for adding version numbers and version checking to .cm
1053       files.  Lal said that his proposal was just that -- a proposal.
1054       For the time being I went ahead and implemented it so that people
1055       can comment on it.  Everything is completely backward-compatible
1056       (except for the stable library format, i.e., new bootfiles!).
1057    
1058       As usual, see the CM manual for details.
1059    
1060    2. An alternative syntax for anchored paths has been implemented.
1061       Dave has recently voiced the same concerns that I had when I did
1062       this, so there should be some support.  My take is that eventually
1063       I will let support for the current syntax (where anchors are
1064       "implicit") fade out in favor of the new, explicit syntax.
1065       In order to be backward-compatible, both old and new syntax are
1066       currently supported.
1067    
1068       Again, see the CM manual for details.
1069    
1070    3. Parallel make is trying to be slightly smarter:  When the master
1071       process finds a "bottleneck", i.e., when there is only one
1072       compilation unit that can be compiled and everybody else is
1073       waiting on it, then it will simply compile it directly instead
1074       of clumsily telling one of the slaves to do it.
1075    
1076    4. Support for "unsharing" added.  This is necessary in order to be
1077       able to have two different versions of the same library running
1078       at the same time (e.g., for trying out a new MLRISC while still
1079       having the old MLRISC linked into the current compiler, etc.)
1080       See the CM manual.
1081    
1082    5. Simple "makedepend" functionality added for generating Makefile
1083       dependency information.  (This is rather crude at the moment.
1084       Expect some changes here in the future.)
1085    
1086    6. ".fun" added as a recognized suffix for ML files. Also documented
1087       explicitly in the manual that the fallback behavior (unknown suffix
1088       -> ML file) is not an official feature!
1089    
1090    7. Small changes to the pickler for stable libraries.
1091    
1092    8. Several internal changes to CM (for cleanup/improvement).
1093    
1094    
1095    !!!! NEW BINFILES !!!!
1096    
1097    ----------------------------------------------------------------------
1098    Name: Matthias Blume
1099    Date: 2000/04/28 17:30:00 JST
1100    Tag: blume-20000428-pathconfig
1101    Description:
1102    
1103    1. I changed config/install.sh to remove duplicate entries from the
1104       lib/pathconfig file at the end.  Moreover, the final version of
1105       lib/pathconfig is sorted alphabetically.  The same (sorting) is done
1106       in src/system/installml.
1107    
1108    2. The config/install.sh script now consistently uses relative
1109       pathnames in lib/pathconfig whenever the anchor is in the lib
1110       directory.  (So far this was true for the libraries that come
1111       pre-compiled and bundled as part of the bootfiles but not for
1112       libraries that are compiled by the script itself.)
1113    
1114    ----------------------------------------------------------------------
1115    Name: Matthias Blume
1116    Date: 2000/04/26 13:10:00 JST
1117    Tag: blume-20000426-fun_suffix
1118    Description:
1119    
1120    Added ".fun" as a recognized file name suffix (for ML code).
1121    
1122    ----------------------------------------------------------------------
1123    Name: Allen Leung
1124    Date: 2000/04/25 17:00:00 EST
1125    Tag: leunga-20000425-alpha-ra
1126    Description:
1127    
1128    1. Alpha
1129    
1130        PSEUDOARITH was missing in AlphaRewrite.  This causes an endless loop
1131    in C--.
1132    
1133    2. RA
1134    
1135       Added a flag "ra-dump-size" to print out the size of the flowgraph
1136       and the interference graph.
1137    
1138    ----------------------------------------------------------------------
1139    Name: Dave MacQueen
1140    Date: 2000/04/25/
1141    Tag: dbm-20000425-mlyacc_doc_examples
1142    Description:
1143      Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
1144      Updated all three examples in src/ml-yacc/examples to run
1145      under 110.* using CM.make.
1146    
1147    ----------------------------------------------------------------------
1148    Name: Allen Leung
1149    Date: 2000/04/20 23:04:00 EST
1150    Tag: leunga-20000420-ssa-c---stuff
1151    Description:
1152    
1153      This update synchronizes my repository with Yale's.  Most of these
1154    changes, however, do not affect SML/NJ at all (the RA is an exception).
1155    
1156    1. Register Allocator
1157    
1158       a. An improvement in the interference graph construction:
1159          Given a copy
1160    
1161                s <- t
1162    
1163          no interference edge between s and t is added for this definition of s.
1164    
1165       b. I've added two new spill heuristic modules that Fermin and I developed
1166          (in the new library RA.cm). These are unused in SML/NJ but maybe
1167          useful for others (Moby?)
1168    
1169    2. X86
1170    
1171       a. Various fixes in the backend provided by Fermin [C--] and Lal.
1172    
1173    3. Alpha
1174    
1175       a. Added the BSR instruction and code generation that goes with it [C--]
1176       b. Other fixes too numerous to recount provided by Fermin [C--]
1177    
1178    4. Regmaps
1179    
1180       a. The regmaps are not initialized with the identity physical bindings
1181          at creation time.  This is unneeded.
1182    
1183    5. MLRISC Optimizations
1184    
1185       a. The DJ-Graph module can now compute the iterated dominance frontiers
1186          intersects with liveness incrementally in linear time! Woohoo!
1187          This is now used in my new SSA construction algorithm.
1188    
1189       b. THe branch reorganization module is now smarter about linear chains of
1190          basic blocks.
1191    
1192    
1193    ----------------------------------------------------------------------
1194    Name: Matthias Blume
1195    Date: 2000/04/12 13:52:00 JST
1196    Tag: blume_main_v110p27_1
1197    Description:
1198    
1199    Changed install.sh script to handle archive files without version number
1200    and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
1201    name of the boot file archive.
1202    
1203    ----------------------------------------------------------------------
1204    Name: Dave MacQueen
1205    Date: 2000/04/09 14:00 EDT
1206    Tag: dbm-20000410-Version_110_27
1207    Description:
1208    
1209    1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
1210    
1211    2. Updated src/config/version to 110.27
1212    
1213    3. New boot files!
1214    
1215    ----------------------------------------------------------------------
1216    Name: Allen Leung
1217    Date: 2000/04/09 19:09:00 EST
1218    Tag: leunga-20000409-misc
1219    Description:
1220    
1221    1.  Yet another fix for x86 assembly for idivl, imull, mull and friends.
1222    
1223    2.  Miscellaneous improvements to MLRISC (unused in sml/nj)
1224    
1225    ----------------------------------------------------------------------
1226    Name: Stefan
1227    Date: 2000/04/07 10:00:00 EDT
1228    Tag: monnier-20000406-branch-handling
1229    Description:
1230    
1231    Improved handling of branches (mostly those generated from
1232    polymorphic equality), removed switchoff and changed the
1233    default optimization settings (more cpsopt and less flintopt).
1234    
1235    ----------------------------------------------------------------------
1236    Name: Allen Leung
1237    Date: 2000/04/06 01:30:00 EST
1238    Tag: leunga-20000406-peephole-x86-SSA-2
1239    Description:
1240    
1241       Forgot a few files.
1242    
1243    ----------------------------------------------------------------------
1244    Name: Allen Leung
1245    Date: 2000/04/06 00:36:00 EST
1246    Tag: leunga-20000406-peephole-x86-SSA
1247    Description:
1248    
1249    1.  New Peephole code
1250    
1251    2.  Minor improvement to X86 instruction selection
1252    
1253    3.  Various fixes to SSA and machine description -> code translator
1254    
1255    ----------------------------------------------------------------------
1256    Name: Matthias Blume
1257    Date: 2000/04/05 12:30:00 JST
1258    Tag: blume_main_v110p26p2_3
1259    Description:
1260    
1261    This update just merges three minor cosmetic updates to CM's sources
1262    to get ready for the 110.27 code freeze on Friday.  No functionality
1263    has changed.
1264    
1265    ----------------------------------------------------------------------
1266    Name: Allen Leung
1267    Date: 2000/04/04 19:39:00 EST
1268    Tag: leunga-20000404-x86-asm
1269    Description:
1270    
1271    1.  Fixed a problem in X86 assembly.
1272    
1273        Things like
1274    
1275           jmp %eax
1276           jmp (%eax)
1277    
1278        should be output as
1279    
1280           jmp *%eax
1281           jmp *(%eax)
1282    
1283    2.  Assembly output
1284    
1285          Added a new flag
1286    
1287              "asm-indent-copies" (default to false)
1288    
1289          When this flag is on, parallel copies will be indented an extra level.
1290    
1291    ----------------------------------------------------------------------
1292    Name: Allen Leung
1293    Date: 2000/04/04 03:18:00 EST
1294    Tag: leunga-20000404-C--Moby
1295    Description:
1296    
1297        All of these fixes are related to C--, Moby, and my own optimization
1298        stuff; so they shouldn't affect SML/NJ.
1299    
1300    1.  X86
1301    
1302        Various fixes related floating point, and extensions.
1303    
1304    2.  Alpha
1305    
1306        Some extra patterns related to loads with signed/zero extension
1307        provided by Fermin.
1308    
1309    3.  Assembly
1310    
1311        When generating assembly, resolve the value of client defined constants,
1312        instead of generating symbolic values.  This is controlled by the
1313        new flag "asm-resolve-constants", which is default to true.
1314    
1315    4.  Machine Descriptions
1316    
1317        a. The precedence parser was slightly broken when parsing infixr symbols.
1318        b. The type generalizing code had the bound variables reversed, resulting
1319           in a problem during arity raising.
1320        c. Various fixes in machine descriptions.
1321    
1322    ----------------------------------------------------------------------
1323    Name: Matthias Blume
1324    Date: 2000/04/03 16:05:00 JST
1325    Tag: blume_main_v110p26p2_2
1326    Description:
1327    
1328    I eliminated coreEnv from compInfo.  Access to the "Core" structure is
1329    now done via the ordinary static environment that is context to each
1330    compilation unit.
1331    
1332    To this end, I arranged that instead of "structure Core" as "structure
1333    _Core" is bound in the pervasive environment.  Core access is done via
1334    _Core (which can never be accidentally rebound because _Core is not a
1335    legal surface-syntax symbol).
1336    
1337    The current solution is much cleaner because the core environment is
1338    now simply part of the pervasive environment which is part of every
1339    compilation unit's context anyway.  In particular, this eliminates all
1340    special-case handling that was necessary until now in order to deal
1341    with dynamic and symbolic parts of the core environment.
1342    
1343    Remaining hackery (to bind the "magic" symbol _Core) is localized in the
1344    compilation manager's bootstrap compiler (actually: in the "init group"
1345    handling).  See the comments in src/system/smlnj/init/init.cmi for
1346    more details.
1347    
1348    I also tried to track down all mentions of "Core" (as string argument
1349    to Symbol.strSymbol) in the compiler and replaced them with a
1350    reference to the new CoreSym.coreSym.  Seems cleaner since the actual
1351    name appears in one place only.
1352    
1353    Binfile and bootfile format have not changed, but the switchover from
1354    the old "init.cmi" to the new one is a bit tricky, so I supplied new
1355    bootfiles anyway.
1356    
1357    ----------------------------------------------------------------------
1358    Name: Allen Leung
1359    Date: 2000/04/02 21:17:00 EST
1360    Tag: leunga-20000402-mltree
1361    Description:
1362    
1363       1. Renamed the constructor CALL in MLTREE by popular demand.
1364       2. Added a bunch of files from my repository.  These are currently
1365          used by other non-SMLNJ backends.
1366    
1367    ----------------------------------------------------------------------
1368    Name: Allen Leung
1369    Date: 2000/03/31 21:15:00 EST
1370    Tag: leunga-20000331-aliasing
1371    Description:
1372    
1373    This update contains a rewritten (and hopefully more correct) module
1374    for extracting aliasing information from CPS.
1375    
1376       To turn on this feature:
1377    
1378            Compiler.Control.CG.memDisambiguate := true
1379    
1380       To pretty print the region information with assembly
1381    
1382           Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
1383    
1384       To control how many levels of aliasing information are printed, use:
1385    
1386           Compiler.Control.MLRISC.getInt "points-to-show-level" := n
1387    
1388       The default of n is 3.
1389    
1390    ----------------------------------------------------------------------
1391    Name: David MacQueen
1392    Date: 2000/03/31 11:15:00 EST
1393    Tag: dbm-20000331-runtime_fix
1394    Description:
1395    
1396    This update contains:
1397    
1398    1. runtime/c-lib/c-libraries.c
1399       includes added in revision 1.2 caused compilation errors on hppa-hpux
1400    
1401    2. fix for bug 1556
1402       system/Basis/Implementation/NJ/internal-signals.sml
1403    
1404    ----------------------------------------------------------------------
1405    Name: Matthias Blume
1406    Date: 2000/03/31 18:00:00 JST
1407    Tag: blume_main_v110p26p2_1
1408    Description:
1409    
1410    This update contains:
1411    
1412    1. A small change to CM's handling of stable libraries:
1413       CM now maintains one "global" modmap that is used for all stable
1414       libraries.  The use of such a global modmap maximizes sharing and
1415       minimizes the need for re-traversing parts of environments during
1416       modmap construction.  (However, this has minor impact since modmap
1417       construction seems to account for just one percent or less of total
1418       compile time.)
1419    
1420    2. I added a "genmap" phase to the statistics.  This is where I got the
1421       "one percent" number (see above).
1422    
1423    3. CM's new tool parameter mechanism just became _even_ better. :)
1424       - The parser understands named parameters and recursive options.
1425       - The "make" and "shell" tools use these new features.
1426         (This makes it a lot easier to cascade these tools.)
1427       - There is a small syntax change: named parameters use a
1428    
1429           <name> : ( <option> ... )            or
1430           <name> : <string>
1431    
1432         syntax.  Previously, named parameters were implemented in an
1433         ad-hoc fashion by each tool individually (by parsing strings)
1434         and had the form
1435    
1436           <name>=<string>
1437    
1438       See the CM manual for a full description of these issues.
1439    
1440    ----------------------------------------------------------------------
1441    Name: Matthias Blume
1442    Date: 2000/03/30 18:00:00 JST
1443    Tag: blume_main_v110p26p2_0
1444    Description:
1445    
1446    !!!!! WARNING !!!!!!
1447    !!  New binfiles  !!
1448    !!!!!!!!!!!!!!!!!!!!
1449    
1450    This update contains:
1451    
1452    1. Moderate changes to CM:
1453    
1454       - Changes to CM's tools mechanism.  In particular, it is now possible
1455       to have tools that accept additional "command line" parameters
1456       (specified in the .cm file at each instance where the tool's class is
1457       used).
1458    
1459       This was done to accommodate the new "make" and "shell" tools which
1460       facilitate fairly seamless hookup to portions of code managed using
1461       Makefiles or Shell scripts.
1462    
1463       There are no classes "shared" or "private" anymore.  Instead, the
1464       sharing annotation is now a parameter to the "sml" class.
1465    
1466       There is a bit of generic machinery for implementing one's own
1467       tools that accept command-line parameters.  However, I am not yet fully
1468       satisfied with that part, so expect changes here in the future.
1469    
1470       All existing tools are described in the CM manual.
1471    
1472       - Slightly better error handling.  (CM now suppresses many followup
1473       error messages that tended to be more annoying than helpful.)
1474    
1475    2. Major changes to the compiler's static environment data structures.
1476    
1477       - no CMStaticEnv anymore.
1478            - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
1479              but it is called Environment), no conversions between different
1480              kinds of static environments
1481    
1482       - There is still a notion of a "modmap", but such modmaps are generated
1483         on demand at the time when they are needed.  This sounds slow, but I
1484         sped up the code that generates modmaps enough for this not to lead to
1485         a slowdown of the compiler (at least I didn't detect any).
1486    
1487       - To facilitate rapid modmap generation, static environments now
1488         contain an (optional) "modtree" structure.  Modtree annotations are
1489         constructed by the unpickler during unpickling.  (This means that
1490         the elaborator does not have to worry about modtrees at all.)
1491         Modtrees have the advantage that they are compositional in the same
1492         way as the environment data structure itself is compositional.
1493         As a result, modtrees never hang on to parts of an environment that
1494         has already been rendered "stale" by filtering or rebinding.
1495    
1496       - I went through many, many trials and errors before arriving at the
1497         current solution.  (The initial idea of "linkpaths" did not work.)
1498         But the result of all this is that I have touched a lot of files that
1499         depend on the "modules" and "types" data structures (most of the
1500         elaborator). There were a lot of changes during my "linkpath" trials
1501         that could have been reverted to their original state but weren't.
1502         Please, don't be too harsh on me for messing with this code a bit more
1503         than what was strictly necessary...  (I _did_ resist the tempation
1504         of doing any "global reformatting" to avoid an untimely death at
1505         Dave's hands. :)
1506    
1507       - One positive aspect of the previous point:  At least I made sure that
1508         all files that I touched now compile without warnings (other than
1509         "polyEqual").
1510    
1511       - compiler now tends to run "leaner" (i.e., ties up less memory in
1512         redundant modmaps)
1513    
1514    ----------------------------------------------------------------------
1515    Name: Allen Leung
1516    Date: 2000/03/29 18:00:00
1517    Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
1518    Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
1519    Description:
1520    
1521       This update contains *MAJOR* changes to the way code is generated from CPS
1522    in the module mlriscGen, and in various backend modules.
1523    
1524    CHANGES
1525    =======
1526    
1527    1. MLRiscGen: forward propagation fix.
1528    
1529       There was a bug in forward propagation introduced at about the same time
1530       as the MLRISC x86 backend, which prohibits coalescing to be
1531       performed effectively in loops.
1532    
1533       Effect: speed up of loops in RISC architectures.
1534               By itself, this actually slowed down certain benchmarks on the x86.
1535    
1536    2. MLRiscGen:  forward propagating addresses from consing.
1537    
1538       I've changed the way consing code is generated.  Basically I separated
1539       out the initialization part:
1540    
1541            store tag,   offset(allocptr)
1542            store elem1, offset+4(allocptr)
1543            store elem2, offset+8(allocptr)
1544            ...
1545            store elemn, offset+4n(allocptr)
1546    
1547       and the address computation part:
1548    
1549            celladdr <- offset+4+alloctpr
1550    
1551       and move the address computation part
1552    
1553       Effect:  register pressure is generally lower as a result.  This
1554                makes compilation of certain expressions much faster, such as
1555                long lists with non-trivial elements.
1556    
1557                 [(0,0), (0,0), .... (0,0)]
1558    
1559    3. MLRiscGen: base pointer elimination.
1560    
1561        As part of the linkage mechanism, we generate the sequence:
1562    
1563         L:  ...  <- start of the code fragment
1564    
1565         L1:
1566             base pointer <- linkreg - L1 + L
1567    
1568         The base pointer was then used for computing relocatable addresses
1569       in the code fragment.  Frequently (such as in lots of continuations)
1570       this is not needed.  We now eliminate this sequence whenever possible.
1571    
1572         For compile time efficiency, I'm using a very stupid local heuristic.
1573       But in general, this should be done as a control flow analysis.
1574    
1575       Effect:  Smaller code size.  Speed up of most programs.
1576    
1577    4. Hppa back end
1578    
1579         Long jumps in span dependence resolution used to depend on the existence
1580      of the base pointer.
1581    
1582         A jump to a long label L was expanded into the following sequence:
1583    
1584          LDIL %hi(L-8192), %r29
1585          LDO  %lo(L-8192)(%r29), %r29
1586          ADD  %r29, baseptr, %r29
1587          BV,n %r0(%r29)
1588    
1589         In the presence of change (3) above, this will not work.  I've changed
1590       it so that the following sequence of instructions are generated, which
1591       doesn't mention the base pointer at all:
1592    
1593             BL,n  L', %r29           /* branch and link, L' + 4 -> %r29 */
1594        L':  ADDIL L-(L'+4), %r29     /* Compute address of L */
1595             BV,n  %r0(%r29)          /* Jump */
1596    
1597    5. Alpha back end
1598    
1599          New alpha instructions LDB/LDW have been added, as per Fermin's
1600       suggestions.   This is unrelated to all other changes.
1601    
1602    6. X86 back end
1603    
1604         I've changed andl to testl in the floating point test sequence
1605         whenever appropriate.  The Intel optimization guide states that
1606         testl is preferable to andl.
1607    
1608    7. RA (x86 only)
1609    
1610         I've improved the spill propagation algorithm, using an approximation
1611       of maximal weighted independent sets.   This seems to be necessary to
1612       alleviate the negative effect in light of the slow down in (1).
1613    
1614         I'll write down the algorithm one of these days.
1615    
1616    8. MLRiscGen: frequencies
1617    
1618         I've added an annotation that states that all call gc blocks have zero
1619       execution frequencies.  This improves register allocation on the x86.
1620    
1621    BENCHMARKS
1622    ==========
1623    
1624       I've only perform the comparison on 110.25.
1625    
1626       The platforms are:
1627    
1628        HPPA  A four processor HP machine (E9000) with 5G of memory.
1629        X86   A 300Hhz Pentium II with 128M of memory, and
1630        SPARC An Ultra sparc 2 with 512M of memory.
1631    
1632       I used the following parameters for the SML benchmarks:
1633    
1634                 @SMLalloc
1635         HPPA    256k
1636         SPARC   512k
1637         X86     256k
1638    
1639    COMPILATION TIME
1640    ----------------
1641       Here are the numbers comparing the compilation times of the compilers.
1642       I've only compared 110.25 compiling the new sources versus
1643       a fixpoint version of the new compiler compiling the same.
1644    
1645                     110.25                                  New
1646               Total  Time in RA  Spill+Reload   Total  Time In RA Spill+Reload
1647         HPPA   627s    116s        2684+3584     599s    95s       1003+1879
1648         SPARC  892s    173s        2891+3870     708s    116s      1004+1880
1649         X86    999s    315s       94006+130691   987s    296s    108877+141957
1650    
1651                   110.25         New
1652                Code Size      Code Size
1653         HPPA   8596736         8561421
1654         SPARC  8974299         8785143
1655         X86    9029180         8716783
1656    
1657       So in summary, things are at least as good as before.   Dramatic
1658       reduction in compilation is obtained on the Sparc; I can't explain it,
1659       but it is reproducible.  Perhaps someone should try to reproduce this
1660       on their own machines.
1661    
1662    SML BENCHMARKS
1663    --------------
1664    
1665        On the average, all benchmarks perform at least as well as before.
1666    
1667          HPPA         Compilation Time     Spill+Reload      Run Time
1668                     110.25  New            110.25    New   110.25  New
1669    
1670          barnesHut  3.158  3.015  4.75%    1+1       0+0   2.980  2.922   2.00%
1671              boyer  6.152  5.708  7.77%    0+0       0+0   0.218  0.213   2.34%
1672       count-graphs  1.168  1.120  4.32%    0+0       0+0  22.705 23.073  -1.60%
1673                fft  0.877  0.792 10.74%    1+3       1+3   0.602  0.587   2.56%
1674        knuthBendix  3.180  2.857 11.32%    0+0       0+0   0.675  0.662   2.02%
1675             lexgen  6.190  5.290 17.01%    0+0       0+0   0.913  0.788  15.86%
1676               life  0.803  0.703 14.22%   25+25      0+0   0.153  0.140   9.52%
1677              logic  2.048  2.007  2.08%    6+6       1+1   4.133  4.008   3.12%
1678         mandelbrot  0.077  0.080 -4.17%    0+0       0+0   0.765  0.712   7.49%
1679             mlyacc 22.932 20.937  9.53%  154+181    32+57  0.468  0.430   8.91%
1680            nucleic  5.183  5.060  2.44%    2+2       0+0   0.125  0.120   4.17%
1681      ratio-regions  3.357  3.142  6.84%    0+0       0+0  116.225 113.173 2.70%
1682                ray  1.283  1.290 -0.52%    0+0       0+0   2.887  2.855   1.11%
1683             simple  6.307  6.032  4.56%   28+30      5+7   3.705  3.658   1.28%
1684                tsp  0.888  0.862  3.09%    0+0       0+0   7.040  6.893   2.13%
1685               vliw 24.378 23.455  3.94%  106+127    25+45  2.758  2.707   1.91%
1686      --------------------------------------------------------------------------
1687       Average                     6.12%                                   4.09%
1688    
1689          SPARC        Compilation Time     Spill+Reload      Run Time
1690                     110.25  New            110.25    New   110.25  New
1691    
1692          barnesHut  3.778  3.592  5.20%    2+2       0+0   3.648  3.453    5.65%
1693              boyer  6.632  6.110  8.54%    0+0       0+0   0.258  0.242    6.90%
1694       count-graphs  1.435  1.325  8.30%    0+0       0+0  33.672 34.737   -3.07%
1695                fft  0.980  0.940  4.26%    3+9       2+6   0.838  0.827    1.41%
1696        knuthBendix  3.590  3.138 14.39%    0+0       0+0   0.962  0.967   -0.52%
1697             lexgen  6.593  6.072  8.59%    1+1       0+0   1.077  1.078   -0.15%
1698               life  0.972  0.868 11.90%   26+26      0+0   0.143  0.140    2.38%
1699              logic  2.525  2.387  5.80%    7+7       1+1   5.625  5.158    9.05%
1700         mandelbrot  0.090  0.093 -3.57%    0+0       0+0   0.855  0.728   17.39%
1701             mlyacc 26.732 23.827 12.19%  162+189    32+57  0.550  0.560   -1.79%
1702            nucleic  6.233  6.197  0.59%    3+3       0+0   0.163  0.173   -5.77%
1703      ratio-regions  3.780  3.507  7.79%    0+0       0+0 133.993 131.035   2.26%
1704                ray  1.595  1.550  2.90%    1+1       0+0   3.440  3.418    0.63%
1705             simple  6.972  6.487  7.48%   29+32      5+7   3.523  3.525   -0.05%
1706                tsp  1.115  1.063  4.86%    0+0       0+0   7.393  7.265    1.77%
1707               vliw 27.765 24.818 11.87%  110+135    25+45  2.265  2.135    6.09%
1708      ----------------------------------------------------------------------------
1709       Average                     6.94%                                    2.64%
1710    
1711          X86          Compilation Time     Spill+Reload      Run Time
1712                     110.25  New            110.25    New   110.25  New
1713    
1714          barnesHut  5.530  5.420  2.03%  593+893   597+915   3.532  3.440   2.66%
1715              boyer  8.768  7.747 13.19%  493+199   301+289   0.327  0.297  10.11%
1716       count-graphs  2.040  2.010  1.49%  298+394   315+457  26.578 28.660  -7.26%
1717                fft  1.327  1.302  1.92%  112+209   115+210   1.055  0.962   9.71%
1718        knuthBendix  5.218  5.475 -4.69%  451+598   510+650   0.928  0.932  -0.36%
1719             lexgen  9.970  9.623  3.60% 1014+841  1157+885   0.947  0.928   1.97%
1720               life  1.183  1.183  0.00%  162+182   145+148   0.127  0.103  22.58%
1721              logic  3.285  3.512 -6.45%  514+684   591+836   5.682  5.577   1.88%
1722         mandelbrot  0.147  0.143  2.33%   38+41     33+54    0.703  0.690   1.93%
1723             mlyacc 35.457 32.763  8.22% 3496+4564 3611+4860  0.552  0.550   0.30%
1724            nucleic  7.100  6.888  3.07%  239+168   201+158   0.175  0.173   0.96%
1725      ratio-regions  6.388  6.843 -6.65% 1182+257   981+300  120.142 120.345 -0.17%
1726                ray  2.332  2.338 -0.29%  346+398   402+494   3.593  3.540   1.51%
1727             simple  9.912  9.903  0.08% 1475+941  1579+1168  3.057  3.178  -3.83%
1728                tsp  1.623  1.532  5.98%  266+200   250+211   8.045  7.878   2.12%
1729               vliw 33.947 35.470 -4.29% 2629+2774 2877+3171  2.072  1.890   9.61%
1730      ----------------------------------------------------------------------------
1731       Average                     1.22%                                     3.36%
1732    
1733    ----------------------------------------------------------------------
1734    Name: Allen Leung
1735    Date: 2000/03/23 16:25:00
1736    Tag: leunga-20000323-fix_x86_alpha
1737    Description:
1738    
1739    1. X86 fixes/changes
1740    
1741       a.  The old code generated for SETcc was completely wrong.
1742           The Intel optimization guide is VERY misleading.
1743    
1744    2. ALPHA fixes/changes
1745    
1746       a.  Added the instructions LDBU, LDWU, STB, STW as per Fermin's suggestion.
1747       b.  Added a new mode byteWordLoadStores to the functor parameter to Alpha()
1748       c.  Added reassociation code for address computation.
1749    
1750    ----------------------------------------------------------------------
1751    Name: Allen Leung
1752    Date: 2000/03/22 01:23:00
1753    Tag: leunga-20000322-fix_x86_hppa_ra
1754    Description:
1755    
1756    1. X86 fixes/changes
1757    
1758       a.  x86Rewrite bug with MUL3 (found by Lal)
1759       b.  Added the instructions FSTS, FSTL
1760    
1761    2. PA-RISC fixes/changes
1762    
1763       a.  B label should not be a delay slot candidate!  Why did this work?
1764       b.  ADDT(32, REG(32, r), LI n) now generates one instruction instead of two,
1765           as it should be.
1766       c.  The assembly syntax for fstds and fstdd was wrong.
1767       d.  Added the composite instruction COMICLR/LDO, which is the immediate
1768           operand variant of COMCLR/LDO.
1769    
1770    3. Generic MLRISC
1771    
1772       a.  shuffle.sml rewritten to be slightly more efficient
1773       b.  DIV bug in mltree-simplify fixed (found by Fermin)
1774    
1775    4. Register Allocator
1776    
1777       a.  I now release the interference graph earlier during spilling.
1778           May improve memory usage.
1779    
1780  ----------------------------------------------------------------------  ----------------------------------------------------------------------
1781  Name: Matthias Blume  Name: Matthias Blume
1782  Date: 2000/03/14 14:15:32  Date: 2000/03/14 14:15:32
# Line 19  Line 1785 
1785    
1786  1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an  1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
1787  additional argument called "template" which is an optional string that  additional argument called "template" which is an optional string that
1788  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
1789  explanation.  explanation.
1790    
1791  2. A special-purpose tool can be "regisitered" by simply dropping the  2. A special-purpose tool can be "registered" by simply dropping the
1792  corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same  corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
1793  directory where the .cm file lives that uses this tool.  (The  directory where the .cm file lives that uses this tool.  (The
1794  behavior/misfeature until now was to look for the tool description  behavior/misfeature until now was to look for the tool description
# Line 66  Line 1832 
1832  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
1833  <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,
1834  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
1835  perhaps got accidentially killed).  In this case, fixpt will quickly  perhaps got accidentally killed).  In this case, fixpt will quickly
1836  move through what exists before continuing where it left off earlier,  move through what exists before continuing where it left off earlier,
1837  and, thus, saves a lot of time.  and, thus, saves a lot of time.
1838    
# Line 116  Line 1882 
1882    it from that remote directory.    it from that remote directory.
1883    This should simplify installation further:  For machines that have    This should simplify installation further:  For machines that have
1884    access to the internet, just fetch <version>-config.tgz, unpack it,    access to the internet, just fetch <version>-config.tgz, unpack it,
1885    edit config/targets, and go (run config/install.sh).  The scipt will    edit config/targets, and go (run config/install.sh).  The script will
1886    fetch everything else that it might need all by itself.    fetch everything else that it might need all by itself.
1887    
1888    For CVS users, this mechanism is not relevant for source archives, but    For CVS users, this mechanism is not relevant for source archives, but

Legend:
Removed from v.578  
changed lines
  Added in v.705

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