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

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