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 572, Thu Mar 9 02:43:06 2000 UTC revision 1728, Wed Dec 15 21:27:01 2004 UTC
# Line 8  Line 8 
8  The form of an entry should be:  The form of an entry should be:
9    
10  Name:  Name:
11  Date:  Date: yyyy/mm/dd
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14    
15  ----------------------------------------------------------------------  ----------------------------------------------------------------------
16    Name: Matthias Blume (blume (at) tti - c (dot) org)
17    Date: 2004/12/15 15:30:00 CST
18    Tag: Release_110_52
19    Description:
20    
21    New working version (110.52).  NEW BOOTFILES!
22    
23    ----------------------------------------------------------------------
24    Name: Matthias Blume (blume (at) tti - c (dot) org)
25    Date: 2004/12/15 12:45:00 CST
26    Tag: blume-20041215-spaces
27    Description:
28    
29    More on the space problem (this time for Win32).
30    
31    ----------------------------------------------------------------------
32    Name: Matthias Blume (blume (at) tti - c (dot) org)
33    Date: 2004/12/14 17:30:00 CST
34    Tag: blume-20041214-spaces
35    Description:
36    
37    Hacked some of the scripts (in particular: the installer) to cope with
38    spaces in filenames a bit better.  But beware: the current "solution"
39    is likely still full of bugs and inherently incomplete.  (We need to
40    do away with those shell scripts for a comprehensive solution.)
41    
42    ----------------------------------------------------------------------
43    Name: Matthias Blume (blume (at) tti - c (dot) org)
44    Date: 2004/12/13 14:45:00 CST
45    Tag: blume-20041213-ml-makedepend
46    Description:
47    
48    Fixed bug in code for ml-makedepend.
49    
50    ----------------------------------------------------------------------
51    Name: Matthias Blume (blume (at) tti - c (dot) org)
52    Date: 2004/12/09 16:30:00 CST
53    Tag: blume-20041209-statistics
54    Description:
55    
56    Added two simple but potentially useful statistics modules to SML/NJ Library.
57    (See CHANGES file there.)
58    
59    ----------------------------------------------------------------------
60    Name: Matthias Blume (blume (at) tti - c (dot) org)
61    Date: 2004/12/01 16:50:00 CST
62    Tag: blume-20041201-atom
63    Description:
64    
65    smlnj-lib:
66    
67      Added function HashString.hashString' for substrings.
68      Hand-inlined CharVector.foldl into HashString (for speed).
69      Modified implementation of structure Atom to avoid extracting
70      strings from substrings unless necessary.
71      (Also see CHANGES file for smlnj-lib.)
72    
73    ----------------------------------------------------------------------
74    Name: Matthias Blume (blume (at) tti - c (dot) org)
75    Date: 2004/11/24 22:15:00 CST
76    Tag: blume-20041124-cml
77    Description:
78    
79    Made sure CML compiles when Position = Int64.
80    
81    ----------------------------------------------------------------------
82    Name: Matthias Blume (blume (at) tti - c (dot) org)
83    Date: 2004/11/24 14:30:00 CST
84    Tag: blume-20041124-position
85    Description:
86    
87    The compiler can now be compiled in a mode that makes structure Position
88    equal to Int64.  The default, however, is unchanged (Position = Int31)
89    for the time being.
90    
91    To enable 64-bit positions, use the following procedure:
92    
93    1. Start sml
94    2. Autoload $smlnj/cmb.cm (if not already autoloaded)
95    3. Type
96          #set (CMB.symval "USE_64_BIT_POSITIONS") (SOME 1);
97    4. Run CMB.make() as usual.
98    
99    This is barely tested.  The only test so far was a little SML program
100    counting the number of characters in an 8-gigabyte file by
101    reading it character-by-character.  That test was successful.
102    
103    In support of 64-bit positions, a number of new functions have been
104    added to the runtime system.
105    
106    ----------------------------------------------------------------------
107    Name: Matthias Blume (blume (at) tti - c (dot) org)
108    Date: 2004/11/23 14:45:00 CST
109    Tag: blume-20041123-useFile
110    Description:
111    
112    Fixed a problem with unhelpful error messages related to problems with
113    .cm- or .sml files that appear as part of the sml command line.
114    
115    ----------------------------------------------------------------------
116    Name: Matthias Blume (blume (at) tti - c (dot) org)
117    Date: 2004/11/18 15:40:00 CST
118    Tag: Release_110_51
119    Description:
120    
121    New working version (110.51).  NEW BOOTFILES!
122    
123    ----------------------------------------------------------------------
124    Name: Matthias Blume (blume (at) tti - c (dot) org)
125    Date: 2004/11/18 15:35:00 CST
126    Tag: <none>
127    Description:
128    
129    Enabled dlopen and friends for FreeBSD (as recommended by Johannes 5
130    Joemann).
131    
132    ----------------------------------------------------------------------
133    Name: Allen Leung (leunga (at) reservoir (dot) com)
134    Date: 2004/11/17 16:05:21 EST 2004
135    Tag: leunga-20041117-mlrisc-live-kill
136    Description:
137    
138       Added support for MLTree constructs LIVE and KILL
139    to all the architectures.
140    
141    ----------------------------------------------------------------------
142    Name: Matthias Blume (blume (at) tti - c (dot) org)
143    Date: 2004/11/13 00:20:00 CST
144    Tag: blume-20041113-versiontool
145    Description:
146    
147    - Stripped down the versiontool: It now only handles the version number.
148      The date string is generated at bootstrap time (during makeml).
149    
150    - In a previous commit, fixed a minor issue with how polyequal is being
151      translated.  In particular, the code now "looks through" abstractions.
152      This results in slightly fewer polyEqual warnings and hopefully slightly
153      more efficient code.  Important examples for where this matters are
154      the new int64 and word64 types.
155    
156    ----------------------------------------------------------------------
157    Name: Matthias Blume (blume (at) tti - c (dot) org)
158    Date: 2004/11/12 00:30:00 CST
159    Tag: blume-20041112-int64
160    Description:
161    
162    Structure Int64 fully hooked in.  (The implementation is not very
163    efficient, though.)
164    
165    ----------------------------------------------------------------------
166    Name: Matthias Blume (blume (at) tti - c (dot) org)
167    Date: 2004/11/11 17:30:00 CST
168    Tag: blume-20041111-more64
169    Description:
170    
171    All the pieces of Word64 are now there, with the exception of the
172    conversions from and to LargeWord. (Eventually these need to be identities,
173    but for the time being they don't even make sense because LargeWord is
174    32-bit wide.)
175    
176    Also started to add similar support for Int64, but major pieces of that
177    are still missing.
178    
179    ----------------------------------------------------------------------
180    Name: Matthias Blume (blume (at) tti - c (dot) org)
181    Date: 2004/11/11 00:15:00 CST
182    Tag: blume-20041111-word64
183    Description:
184    
185    Structure Word64 is now (almost) complete, word literals and patterns
186    seem to work.  There are a few odd pieces missing.  In particular,
187    I didn't do the {from,to}LargeWord parts because LargeWord is still Word32
188    at the moment.
189    
190    Making Word64 official would mean that LargeWord becomes Word64.  But
191    this requires extreme care because most word-word conversions have to
192    go through LargeWord, so making a mistake means loss of efficiency or
193    worse.  Eventually there will be a solution similar to (but actually
194    simpler than) what I did with IntInf.
195    
196    ----------------------------------------------------------------------
197    Name: Matthias Blume (blume (at) tti - c (dot) org)
198    Date: 2004/11/10 18:12:00 CST
199    Tag: blume-20041110-64bit
200    Description:
201    
202    More 64-bit hacking (but still not even half-way there yet).
203    Also, some assorted improvements to the handling of 8-bit words.
204    
205    ----------------------------------------------------------------------
206    Name: Matthias Blume (blume (at) tti - c (dot) org)
207    Date: 2004/11/09 17:50:00 CST
208    Tag: <none>
209    Description:
210    
211    Started adding some infrastructure for supporting 64-bit int- and
212    word-types.  (Still in its very early stages.)
213    
214    ----------------------------------------------------------------------
215    Name: Matthias Blume (blume (at) tti - c (dot) org)
216    Date: 2004/10/28 10:45:00 CDT
217    Tag: Release_110_50
218    Description:
219    
220    New working version (110.50).  NEW BOOTFILES!
221    
222                 =====================
223    
224    Also:
225    
226      - Changend config/srcarchiveurl from a file just
227        containing the URL string into a file containing
228        shell script code.  The code has access to the $VERSION variable.
229      - Made corresponding changes to config/install.sh and config/unpack.
230      - Default contents of config/srcarchiveurl uses $VERSION and
231        normally does not have to be edited to reflect a version change.
232    
233      (As a result, a version change can be done by just editing
234       config/version, the rest is now automatic.)
235    
236    ----------------------------------------------------------------------
237    Name: Matthias Blume (blume (at) tti - c (dot) org)
238    Date: 2004/10/27 17:50:00 CDT
239    Tag: blume-20041027-btrace-msg
240    Description:
241    
242    BackTrace.monitor now also reports the source of the exception that
243    triggered the trace.
244    
245    ----------------------------------------------------------------------
246    Name: Matthias Blume (blume (at) tti - c (dot) org)
247    Date: 2004/10/27 17:20:00 CDT
248    Tag: blume-20041027-x86-c-calls
249    Description:
250    
251    This is the HISTORY entry for two earlier commits, both concerning
252    the x86 c-calls code in MLRISC:
253    
254      - added a missing LOAD in the code that deals with struct arguments
255      - made sure the caller does not add the wrong number of bytes to the
256        stack pointer after a call of a function returning a struct
257        (the callee already pops the implicit argument which points to
258         the space reserved for the result)
259    
260    ----------------------------------------------------------------------
261    Name: Allen Leung (leunga (at) reservoir (dot) com)
262    Date: 2004/10/24 14:00:00 EST
263    Tag: leunga-20041024-x86-gas-fucomip
264    Description:
265    
266       John discovered a bug in the syntax of fucomip.
267       The opcodes FU?COMIP? have been changed to
268    
269           fu?comip? %st(i), %st
270    
271    ----------------------------------------------------------------------
272    Name: Matthias Blume (blume (at) tti - c (dot) org)
273    Date: 2004/10/20 15:06:00 CDT
274    Tag: blume-20041020-standalone-backtrace
275    Description:
276    
277    Added a mechanism for getting back-trace information from standalone
278    programs.  Here is how it works:
279    
280       1. The part of the program from which you want to get backtrace
281          information (usually the whole program) should be wrapped with
282          BackTrace.monitor.  This is a (unit->'a)->'a function, and your
283          main program could be modified from something like
284    
285             fun main (pgm, args) = ...
286    
287         to
288    
289            fun main (pgm, args) = BackTrace.monitor (fn () => ...)
290    
291       2. To be able to access BackTrace.monitor, you have to add
292          library $smlnj-tdp/plugins.cm to the .cm file that contains your
293          main function.
294    
295       3. Remove all compiled code (i.e., all the .cm/ subdirectories that
296          CM might have created in the past for your project).
297    
298       4. Build the system using this command line:
299    
300             ml-build -Ctdp.instrument=true \$smlnj-tdp/back-trace.cm \
301                      myprog.cm MyProg.main myprog
302    
303          instead of the usual
304    
305             ml-build myprog.cm MyProg.main myprog
306    
307    I changed a library name:
308    
309       $/trace-debug-profile.cm --> $smlnj-tdp/plugins.cm
310    
311    New libraries:
312    
313       $smlnj-tdp/back-trace.cm
314          -- when loaded causes the back-trace plugin to be installed
315       $smlnj-tdp/coverage.cm
316          -- when loaded causes the coverage plugin to be installed
317    
318    ----------------------------------------------------------------------
319    Name: Matthias Blume (blume (at) tti - c (dot) org)
320    Date: 2004/10/18 16:45:00 CDT
321    Tag: blume-20041018-groupowner
322    Description:
323    
324    Added an "obsolete" warning for the "group owner" syntax to CM's parser.
325    
326    Eliminated group owner specs from .cm files throughout the source tree.
327    
328    ----------------------------------------------------------------------
329    Name: Matthias Blume (blume (at) tti - c (dot) org)
330    Date: 2004/10/15 15:45:00 CDT
331    Tag: blume-20041015-coverage
332    Description:
333    
334    * Test coverage tool added!
335    * Further reorganization of tracing-, debugging-, and profiling support:
336    
337        - moved original BTImp -- now called BackTrace -- into a separate
338          library called $/trace-debug-profile.cm
339        - eliminated all mentions of BTrace from SMLofNJ.Internals
340        - only the instrumentation mechanism is now left in the compiler proper
341        - BackTrace module is a plugin which is NOT plugged in by default
342        - Coverage module is another such plugin
343    
344    To get the benefits of any of these plugin modules, the code in
345    question must be compiled with tdp instrumentation turned on.  This
346    can be done by setting SMLofNJ.Internals.TDP.mode to true.
347    (The ref cell is also controlled via the -Ctdp.instrument=... switch.)
348    
349    Plugins are selected at link time.  (Pre-compiled instrumented code
350    can be re-loaded with different plugins in effect.)  When an
351    instrumented module is linked, whatever plugins are at that time
352    enabled will come into effect for that module.
353    
354    To enable the back-trace plugin, load library $/trace-debug-profile.cm
355    and invoke BackTrace.install() (e.g., from the interactive prompt).
356    To enable the coverage plugin, load the same library and invoke
357    Coverage.install().
358    
359    Back-traces are generated automatically on uncaught exceptions and
360    when the code in question explicitly invokes BackTrace.trigger().
361    
362    Coverage (and execution frequency-) information must be queried
363    explicitly by calling Coverage.not_covered and Coverage.hot_spots.
364    
365    ----------------------------------------------------------------------
366    Name: Matthias Blume (blume (at) tti - c (dot) org)
367    Date: 2004/10/14 17:40:00 CDT
368    Tag: blume-20041014-tdp-core
369    Description:
370    
371    Snapshot of a significant overhaul of how the trace/debug/profile support
372    is hooked into the system (specifically: Core and SMLofNJ.Internals).
373    
374    ----------------------------------------------------------------------
375    Name: Matthias Blume (blume (at) tti - c (dot) org)
376    Date: 2004/10/13 16:34:00 CDT
377    Tag: blume-20041013-tdp
378    Description:
379    
380    Some rationalization of names:
381    
382       structure BTrace -> structure TDPInstrument
383       etc.
384    
385    This is is preparation of using the original back-trace
386    instrumentation for other purposes.  "TDP" stands for
387    Trace/Debug/Profile.
388    
389    The control flag controlling whether instrumentation is on or off is now
390    registered under a different name, so instead of running sml as
391    
392     sml -Cinstrument.btrace-mode=true
393    
394    one has to say
395    
396     sml -Ctdp.instrument=true
397    
398    ----------------------------------------------------------------------
399    Name: Matthias Blume (blume (at) tti - c (dot) org)
400    Date: 2004/10/11 16:37:00 CDT
401    Tag: blume-20041011-regions
402    Description:
403    
404    Made some minor modifications to elabcore.sml to have source regions
405    be propagated more tightly -- resulting in better (i.e., smaller)
406    regions being reported in error- and debug messages.
407    
408    ----------------------------------------------------------------------
409    Name: Matthias Blume (blume (at) tti - c (dot) org)
410    Date: 2004/10/08 22:50:00 CDT
411    Tag: blume-20041008-cmkw
412    Description:
413    
414    Fixed handling of keywords in .cm files: After seeing "is" the lexer
415    treats subsequent occurrences of "group", "library", "source", "is",
416    "*", and "-" as ordinary identifiers rather than keywords.
417    
418    Most seriously, this fixes a problem with CM's "shell" tool.  The tool
419    is supposed to accept a tool argument called "source", but this did
420    not work because of the clash with the keyword.
421    
422    ----------------------------------------------------------------------
423    Name: Matthias Blume (blume (at) tti - c (dot) org)
424    Date: 2004/10/07 16:00:00 CDT
425    Tag: blume-20041007-cleanup
426    Description:
427    
428    Assorted cleanup work:
429    
430      - got rid of intstrmap in favor of using the library's
431        hash table implementation
432      - threw out most of the pathnames stuff, as it was not used anyway
433      - simplified tokentable implementation
434      - fixed some minor spelling errors
435    
436    ----------------------------------------------------------------------
437    Name: Matthias Blume (blume (at) tti - c (dot) org)
438    Date: 2004/10/06 15:15:15 CDT
439    Tag: blume-20041006-handler
440    Description:
441    
442    Cleaned up the absyn to reflect the invariant that HANDLE always
443    carries a FNexp as part of the type definition.  This eliminates some
444    superfluous sanity checks at runtime down the road.
445    
446    Some minor cleanup of the btrace code.
447    
448    ----------------------------------------------------------------------
449    Name: Matthias Blume (blume (at) tti - c (dot) org)
450    Date: 2004/10/01 10:20:30 CDT
451    Tag: blume-20041001-slave
452    Description:
453    
454    Added hack to make slave mode work in the presence of the version
455    tool.  (Still, since the master does two passes over the code for
456    CMB.make, the release number gets bumped twice when slaves are
457    attached.  I don't know if this is worth fixing...)
458    
459    ----------------------------------------------------------------------
460    Name: Matthias Blume (blume (at) tti - c (dot) org)
461    Date: 2004/09/30 10:55:00 CDT
462    Tag: blume-20040930-version
463    Description:
464    
465    * Moved the "version" magic into its own little library under
466      src/system/smlnj/internal.  This avoids expensive reconstruction of
467      a stable src/compiler/core.cm.
468    
469    * At the same time, structure CompilerVersion is now known as
470      structure SMLNJVersion.
471    
472    * Arranged for the version tool to NOT kick in when rebuilding the system
473      (makeml -rebuild, fixpt).  Otherwise one would never reach a fixpoint.
474      Also, loading the versiontool does not work when rebuilding the system
475      because CM is not properly initialized at that time.
476    
477    ----------------------------------------------------------------------
478    Name: Matthias Blume (blume (at) tti - c (dot) org)
479    Date: 2004/09/29 14:00:00 CDT
480    Tag: blume-20040929-autoversion
481    Description:
482    
483    Implemented some CM magic to have
484        file src/compiler/TopLevel/main/version.sml
485    generated automagically.
486    The version is taken from two files: config/version and config/release.
487    The first is expected to contain a two-part version number such as 110.49.
488    The second should contain a single number, but it may be missing.
489    
490    If the environment variable VERSIONTOOL_BUMP_RELEASE is defined at the
491    time the version tool is loaded (which is the first time you say CMB.make),
492    then the tool will increment the value stored in config/release every
493    time CMB.make is invoked.
494    
495    The binfile format is now insensitive to anything beyond the first
496    two components of a version number, so bumping the release does not render
497    binfiles incompatible.  Auto-bumping can be used to keep track of versions
498    during development without invalidating existing binfiles.
499    
500    In any case, every CMB.make updates the date information in version.sml.
501    (This is the date that is printed in the banner.)
502    
503    ----------------------------------------------------------------------
504    Name: Matthias Blume (blume (at) tti - c (dot) org)
505    Date: 2004/09/28 10:53:00 CDT
506    Tag: blume-20040928-controls
507    Description:
508    
509    Some cleanup of the controls code.
510    
511    ----------------------------------------------------------------------
512    Name: Matthias Blume (blume (at) tti - c (dot) org)
513    Date: 2004/09/27 22:08:00 CDT
514    Tag: blume-20040927-controls
515    Description:
516    
517    Added two pieces of functionality to the Controls interface:
518    
519    1.   val save'restore: 'a control -> unit -> unit
520    
521       grabs the current value of the control in stage 1 and restores it
522       in stage 2.
523    
524    2.   val set' : 'a control * 'a -> unit -> unit
525    
526       stores the given value into the control in stage 2 (i.e., delayed)
527       but does all error checking in stage 1.
528       (This is for string controls that need to do parse their argument --
529        something that might fail.  In some cases, notably in CM, one
530        already knows the intended argument but wants to delay the actual
531        assignment until a time when error recovery would be more difficult.)
532    
533    Changed the handling of controls in tool arguments to classes "sml" and
534    "lazysml":
535       - use Controls.save'restore as a more robust way of restoring the
536         old value (in particular: without having to re-parse the string)
537       - use controls to handle the "overload" keyword in the init group
538         (I believe this change actually fixes a long-standing obscure bug.)
539    
540    ----------------------------------------------------------------------
541    Name: Matthias Blume (blume (at) tti - c (dot) org)
542    Date: 2004/09/27 17:00:00 CDT
543    Tag: blume-20040927-lazysml
544    Description:
545    
546    Added a new tool class called "lazysml" to CM's tool chest.  The only
547    difference to "sml" is that compilation is done with Control.lazysml
548    set to true.  A source of class "lazysml" is automatically recognized
549    by a file name suffix of ".lml".
550    
551    In addition to the above feature, the original class "sml" now also
552    supports a tool argument "lazy" which has the same effect.  As a
553    result, the following three lines are equivalent:
554    
555        foo.sml : lazysml
556        foo.sml : sml (lazy)
557        foo.sml (lazy)
558    
559    The setting goes into effect both during parsing and during
560    compilation.  The original setting is restored right after parsing and
561    after compilation, respectively.
562    
563    In addition to all the above, there is also a general mechanism to set
564    ANY of the "controls" that are available at the command line via
565    "-C..." on a per-sml-file basis.  The same rules that apply for "lazy"
566    apply as well. (In fact, "lazy" is implemented as a special case of
567    the general mechanism.)
568    
569    The .cm file syntax uses a new keyword tool argument called "with".
570    There are several ways of indicating the desired settings:
571    
572        foo.sml (with:parser.quotations=true)
573        foo.sml (with:(name:parser.quotations value:true))
574        foo.sml (with:(name:name1 value:value1 name:name2 value:value2 ...))
575        foo.sml (with:(name1=value1 name2=value2 ...))
576        foo.sml (with:(name1=value1 name:name2 value:value2 name3=value3 ...))
577    
578        etc.
579    
580    Another possible abbreviation is to leave out the =v or value:v part
581    if the name refers to a boolean control (in which case the value is
582    taken to be true).  Thus, one could get lazy sml also by saying:
583    
584        foo.sml (with:parser.lazy-keyword=true)
585        foo.sml (with:parser.lazy-keyword)
586        foo.sml (with:(name:parser.lazy-keyword value:true))
587        foo.sml (with:(name:parser.lazy-keyword))
588    
589    ----------------------------------------------------------------------
590    Name: Matthias Blume (blume (at) tti - c (dot) org)
591    Date: 2004/09/24 16:22:00 CDT
592    Tag: blume-20040924-ppc-long-branch
593    Description:
594    
595    Turned message about "emiting long form of branch" off by default.
596    Added a control flag to turn it back on when desired.
597    
598    ----------------------------------------------------------------------
599    Name: Matthias Blume (blume (at) tti - c (dot) org)
600    Date: 2004/09/24 16:05:00 CDT
601    Tag: blume-20040924-rounding
602    Description:
603    
604    Applied patch for setting rounding modes under Mac OS X.  Thanks to
605    Melissa O'Neill for providing the code!
606    
607    ----------------------------------------------------------------------
608    Name: Matthias Blume (blume (at) tti - c (dot) org)
609    Date: 2004/09/23 17:30:00 CDT
610    Tag: blume-20040923-envvars
611    Description:
612    
613    1. Changed definition of type ControlRegistry.registry_tree to
614       include control_info (i.e., the name of the controlling environment
615       variable).
616    
617    2. Added command-line flags -e and -E to print the names of environment
618       variables that can be used to control internal settings.
619       (This uses the new API mentioned in 1.)
620    
621    ----------------------------------------------------------------------
622    Name: Matthias Blume (blume (at) tti - c (dot) org)
623    Date: 2004/09/13 16:50:00 CDT
624    Tag: Release_110_49
625    Description:
626    
627    New working version (110.49).  NEW BOOTFILES!
628    
629    ----------------------------------------------------------------------
630    Name: Matthias Blume
631    Date: 2004/09/13 16:20:00 CDT
632    Tag: blume-20040913-config-mlrisc
633    Description:
634    
635    Put target "mlrisc" back into the default list.
636    (There is no harm in having it, and some users have expressed their
637    wish to have "mlrisc" included by default.)
638    
639    ----------------------------------------------------------------------
640    Name: John Reppy
641    Date: 2004/09/13
642    Tag: jhr-20040913-signals
643    Description:
644    
645    Fixed the signal masking code to properly nest mask/unmask operations
646    on a per-signal basis.
647    
648    ----------------------------------------------------------------------
649    Name: Matthias Blume (blume (at) tti - c (dot) org)
650    Date: 2004/09/08 13:20:00 CDT
651    Tag: blume-20040908-heap-magic
652    Description:
653    
654    Bumped the heap macig number to 0x09082004 to account for the changed
655    layout of the ML frame under MacOS X.
656    
657    ----------------------------------------------------------------------
658    Name: Allen Leung (leunga (at) reservoir (dot) com)
659    Date: 2004/09/03 11:26:00 EST
660    Tag: leunga-20040903-cygwin-install
661    Description:
662    
663       Added a patch to _arch-n-opsys to enable the Cygwin runtime.
664    The Cygwin runtime is turned on by setting the environment
665    variable SMLNJ_CYGWIN_RUNTIME to 1.
666    
667    ----------------------------------------------------------------------
668    Name: Matthias Blume (blume (at) tti - c (dot) org)
669    Date: 2004/08/31 17:15:00 CDT
670    Tag: blume-20040831-core
671    Description:
672    
673    Added some exports to src/compiler/core.cm upon request by J. Joemann.
674    
675    ----------------------------------------------------------------------
676    Name: Matthias Blume (blume (at) tti - c (dot) org)
677    Date: 2004/08/30 17:55:00 CDT
678    Tag: blume-20040830-installer
679    Description:
680    
681    Upon request by Johannes Joemann:
682    
683    - improved ML code of installer to fall back to coping when renaming
684      fails (i.e., when source and target are on different file systems);
685      the code compiles but has yet to be tested in anger
686    - removed mlrisc from list of default targets (config/targets)
687    
688    ----------------------------------------------------------------------
689    Name: Matthias Blume (blume (at) tti - c (dot) org)
690    Date: 2004/08/27 17:20:00 CDT
691    Tag: blume-20040827-ptreql
692    Description:
693    
694    Added ptreql primop to structure InlineT (upon request from Larry
695    Paulson).
696    
697    ----------------------------------------------------------------------
698    Name: Allen Leung (leunga (at) reservoir (dot) com)
699    Date: 2004/08/15 21:21:00 EST
700    Tag: leunga-110_48-udgraph
701    Description:
702    
703    Another bug fix from Carl Hauser:
704    
705    diff /net/niflab/smlnj48/src/MLRISC/graphs/udgraph.sml udgraph.sml
706    > 48c48
707    > <              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then es else
708    > e::L)
709    > ---
710    > >              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then L else e::L)
711    > Without this, any deletion of an edge in an undirected graph does severe
712    > violence to the graph.
713    
714    ----------------------------------------------------------------------
715    Name: Allen Leung (leunga (at) reservoir (dot) com)
716    Date: 2004/08/10 23:35:00 EST
717    Tag: leunga-110_48-ppc
718    Description:
719    
720       The IBM/MacOS syntax switch on PPC was incorrectly swapped.
721    
722    ----------------------------------------------------------------------
723    Name: Matthias Blume (blume (at) tti - c (dot) org)
724    Date: 2004/08/10 12:00:00 CDT
725    Tag: Release_110_48
726    Description:
727    
728    New working version (110.48).  NEW BOOTFILES!
729    
730    ----------------------------------------------------------------------
731    Name: Allen Leung (leunga (at) reservoir (dot) com)
732    Date: 2004/08/09 12:21:00 EST
733    Tag: leunga-110_47-dijsktra
734    Description:
735    
736    Bug fix from Carl Hauser:
737    
738    single_source_shortest_paths in dijkstra.sml was observed to get wrong
739    answers (by comparing to single_source_shortest_paths in bellman-ford.sml).
740    
741    The problem is that following the expression A.update(dist,s,Num.zero)
742    it is necessary to update the priority queue using Q.decreaseWeight(Q,s).
743    
744    ----------------------------------------------------------------------
745    Name: Matthias Blume (blume (at) tti - c (dot) org)
746    Date: 2004/08/06 18:10:00 CDT
747    Tag: blume-20040806-cmdline
748    Description:
749    
750    Fiddled with handling of command-line options:
751    
752      * sml now quits after processing the command line
753        if -H, -S, -h<n>, or -s<n> appears as the last
754        command-line argument
755      * a new option -q terminates the session when encountered on
756        the command line; subsequent arguments will be ignored
757      * bug fixes: short (erroneous) arguments are no longer ignored
758        completely
759    
760    ----------------------------------------------------------------------
761    Name: Allen Leung (leunga (at) reservoir (dot) com)
762    Date: 2004/08/04 18:17:00 EST
763    Tag: leunga-110_47-ppc-ibm-asm
764    Description:
765    
766      - Added minimal IBM assembly syntax support for PowerPC.
767    
768      - Cygwin: manually changed the file cygwin.def.  Some exported symbols have
769        been altered in the runtime.  We need an automatic way to keep the file
770        in sync.
771    
772    ----------------------------------------------------------------------
773    Name: Matthias Blume (blume (at) tti - c (dot) org)
774    Date: 2004/08/04 14:00:00 CDT
775    Tag: Release_110_47
776    Description:
777    
778    New working version (110.47).  NEW BOOTFILES!
779    
780    ----------------------------------------------------------------------
781    Name: Matthias Blume (blume (at) tti - c (dot) org)
782    Date: 2004/08/03 14:25:00 CDT
783    Tag: blume-20040803-callingconv
784    Description:
785    
786    Added low-level support for choosing C calling conventions by
787    twiddling the type of rawccall.  (See
788    src/compiler/Semant/types/cproto.sml for details.)
789    
790    ----------------------------------------------------------------------
791    Name: Matthias Blume (blume (at) tti - c (dot) org)
792    Date: 2004/08/02 15:55:00 CDT
793    Tag: blume-20040802-backout
794    Description:
795    
796    Backed out of change to win32-filesys.c.  The earlier patch to
797    get_file_time caused CM to produce files with the wrong time stamp.
798    
799    ----------------------------------------------------------------------
800    Name: Matthias Blume (blume (at) tti - c (dot) org)
801    Date: 2004/08/02 14:45:00 CDT
802    Tag: blume-20040802-nlffi-win32
803    Description:
804    
805    Added NLFFI support for Win32, adapted from a patch provided by David
806    Hansel.  This is currently completely untested.  Also, the issue
807    concerning stdcall vs. ccall is still unresolved.
808    
809    ----------------------------------------------------------------------
810    Name: Matthias Blume (blume (at) tti - c (dot) org)
811    Date: 2004/07/30 17:55:00 CDT
812    Tag: blume-20040730-various
813    Description:
814    
815    Gearing up towards 110.47...
816    
817    - various minor bugfixes to ml-nlffigen
818    - a beginning of a manual for nlffi
819    
820    - eliminated 'export name=value' in config/install.sh as this does
821      not work with certain versions of /bin/sh
822      (Thanks to David King at Motorola for catching this.)
823    
824    - several bugfixes provided or suggested by David Hansel at Reactive Systems:
825      - added a test for tm==NULL to gmtime.c and localtime.c
826      - applied patch for incorrect GetFileTime under win32
827      - toSeconds -> toMilliseconds in Win32/win32-process.sml
828    
829    ----------------------------------------------------------------------
830    Name: Matthias Blume (blume (at) tti - c (dot) org)
831    Date: 2004/07/21 18:20:00 CDT
832    Tag: blume-20040721-nlffigen
833    Description:
834    
835    - Fixed minor issue in ml-nlffigen: Now generate structure T_foo
836      for a typedef to an incomplete type, but leave out the "typ" member.
837      (This is just for consistency.)
838    - Started to produce what is supposed to become better (i.e., comprehensive)
839      documentation of what ml-nlffigen does and produces.
840    
841    ----------------------------------------------------------------------
842    Name: Matthias Blume (blume (at) tti - c (dot) org)
843    Date: 2004/07/14 16:25:00 CDT
844    Tag: blume-20040714-union
845    Description:
846    
847    Added C_UNION to c-calls/c-types.sml and updated the machinery
848    (ml-nlffigen, cproto.sml) that conveys C function interface
849    information to the code generator.
850    
851    However, the actual architecture-specific implementation of function
852    arguments and results that are C unions is still not implemented.
853    
854    ----------------------------------------------------------------------
855    Name: Allen Leung (leunga (at) reservoir (dot) com)
856    Date: 2004/07/14 14:38:00 EST
857    Tag: leunga-110_46_1-ppc-lwzu
858    Description:
859    
860        Added these instructions to the PowerPC architecture:
861        LBZU(X), LHZU(X), LWZU(X),
862        STWU(X), STFDU, STFSU
863    
864        etc...
865    
866        Note: I haven't added their instruction encoding into the description.
867    
868    ----------------------------------------------------------------------
869    Name: Allen Leung (leunga (at) reservoir (dot) com)
870    Date: 2004/07/13 15:04:00 EST
871    Tag: leunga-110_46_1-ppc-lwarx
872    Description:
873    
874       Added the two instructions LWARX and STWCX to the PowerPC
875    instruction set.
876    
877       A (untested) rewrite of loop-structure.sml.  The old version
878    is completely broken.
879    
880    ----------------------------------------------------------------------
881    Name: Matthias Blume (blume (at) tti - c (dot) org)
882    Date: 2004/07/13 13:50:00 CDT
883    Tag: blume-20040713-nlffi
884    Description:
885    
886    - use paramAlloc to report c-calls with too many arguments
887      (for PPC version where parameter area is pre-allocated)
888    - added ccall_maxargspace to machspec (to implement the above)
889    - made "make" commend in CM's "make" tool configurable
890    - added option (default: on) for passing the name of the SML/NJ's "bin"
891      directory to "make"; the call looks like this:
892    
893         make <options> SMLNJ_BINDIR=<dir> <target>
894    
895      This can be used by the Makefile to, e.g., pick the "right" version
896      of ml-nlffigen.
897    - minor code tweaks
898    
899    ----------------------------------------------------------------------
900    Name: Matthias Blume (blume (at) tti - c (dot) org)
901    Date: 2004/07/12 22:50:00 CDT
902    Tag: blume-110_46_1-macosx-nlffi
903    Description:
904    
905    NLFFI under Mac OS X now working (sort of).  This is largely untested,
906    though.
907    
908    Note:  1. You have to make a new, clean build of the runtime system.
909           2. There are new BOOTFILES, you have to use them!
910              (Doing the bootstrap process yourself would be *very* painful!
911               If you absolutely have to do it, build the system under
912               a different architecture and then cross-compile.)
913    
914    Version bumped to 110.46.1 to account for runtime data format changes.
915    
916    ----------------------------------------------------------------------
917    Name: Matthias Blume (blume (at) tti - c (dot) org)
918    Date: 2004/06/18 14:30:00 CDT
919    Tag: blume-20040618-unix
920    Description:
921    
922    Changed the implementation of structure Unix so that the same stream
923    is returned every time one of the {text,bin}{In,Out}streamOf functions
924    is invoked on the same proc.  This is not what the spec currently
925    says -- although IMO it arguably should.  (See discussion below.)
926    
927    ----------------------------------------------------------------------
928    Name: Matthias Blume (blume (at) tti - c (dot) org)
929    Date: 2004/06/17 18:15:00 CDT
930    Tag: Release_110_46
931    Description:
932    
933    New working version (110.46).  NEW BOOTFILES!
934    
935    ----------------------------------------------------------------------
936    Name: Matthias Blume (blume (at) tti - c (dot) org)
937    Date: 2004/06/17 17:20:00 CDT
938    Tag: blume-20040617-timer-unix
939    Description:
940    
941    Changed the interface of structures Timer and Unix to match the most
942    recent Basis spec.
943    
944    In the case of Unix there still seems to be an open/weird issue:
945    
946       The {text,bin}{In,Out}streamOf functions are supposed to create
947       fresh streams whenever they are called -- as opposed to have them
948       return the same stream every time.  This design is supposed to
949       prevent space leaks caused by proc values hanging on to streams.
950    
951       The reap function, on the other hand, is supposed to close the
952       streams.  This cannot be done without having a handle on the
953       stream in proc after all...
954    
955    I took the liberty to implement the following stopgap solution:
956    
957      The proc value hangs on to the most recently created stream(s).
958      Reap closes those.  If either or both of the two streams hadn't
959      been created at all yet, then reap will close the corresponding
960      file descriptors directly.
961    
962    PS: I don't understand the original space leak argument anymore.  If
963    a proc hangs on to the imperative stream, then I/O operations on those
964    will advance the state of the cached stream and avoid the space leak.
965    
966    ----------------------------------------------------------------------
967    Name: Matthias Blume (blume (at) tti - c (dot) org)
968    Date: 2004/05/28 16:45:00 CDT
969    Tag: blume-20040528-basis
970    Description:
971    
972    Added signature PACK_REAL and exported functor PrimIO.
973    
974    ----------------------------------------------------------------------
975    Name: Matthias Blume (blume (at) tti - c (dot) org)
976    Date: 2004/05/25 16:00:00 CDT
977    Tag: blume-20040525-group-owner
978    Description:
979    
980    CM now ignores (but still accepts) the "owner" information in group
981    descriptions.  The owner of a group is its next enclosing
982    library. Each group must have a unique owner.  (There is a virtual
983    "toplevel" library that owns groups which are not nested within a real
984    library.)  Previously, each group had to explicitly declare its owner,
985    and CM would check that such a declaration is correct.  The new scheme
986    is to have CM check that for each group there is precisely one owning
987    library.
988    
989    The advantage of the new scheme is that the programmer no longer needs
990    to maintain the somewhat annoying owner information.  The downside is
991    that CM cannot enforce the ownership rule across multiple runs of
992    CM.make.  Fortunately, enclosing the same group in two different
993    libraries A and B which are not part of the same program does not
994    cause real problems.
995    
996    ----------------------------------------------------------------------
997    Name: Matthias Blume (blume (at) tti - c (dot) org)
998    Date: 2004/05/20 16:00:00 CDT
999    Tag: blume-20040520-win32
1000    Description:
1001    
1002    Made the win32 version work again.  (Strangely, a misplaced comma had
1003    slipped into win32-process.c which prevented the runtime from being
1004    compiled correctly.)
1005    
1006    Also, included a minor addition to ml-build.bat analogous to what was
1007    done in blume-20040519-ml-build.
1008    
1009    ----------------------------------------------------------------------
1010    Name: Matthias Blume (blume (at) tti - c (dot) org)
1011    Date: 2004/05/19 22:10:00 CDT
1012    Tag: blume-20040519-ml-build
1013    Description:
1014    
1015    Arranged for ml-build to clean up after itself a little bit better.
1016    The script generates a temporary SML source file and compiles it using
1017    CM, so CM generates metadata (GUID, SKEL, objectfile) for it.  It now
1018    gets rid of those at the end, so they don't accumulate under .cm.
1019    
1020    This required a minor change to install.sh because the name of the
1021    metadata directory (default: .cm) is actually configurable at
1022    installation time.
1023    
1024    ----------------------------------------------------------------------
1025    Name: Matthias Blume (blume (at) tti - c (dot) org)
1026    Date: 2004/05/18 15:50:00 CDT
1027    Tag: blume-20040518-mkreader
1028    Description:
1029    
1030    Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
1031    implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
1032    
1033    ----------------------------------------------------------------------
1034    Name: Matthias Blume (blume (at) tti - c (dot) org)
1035    Date: 2004/05/11 14:35:00 CDT
1036    Tag: blume-20040511-win32sock
1037    Description:
1038    
1039    Added previously missing support for many socket-related functions
1040    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
1041    for the voluminous patch!
1042    
1043    (I have not tested this patch under win32 yet.)
1044    
1045    Here is David's e-mail:
1046    
1047    Hi,
1048    
1049    Attached to this email you find a diff against sml/nj 110.45
1050    that will enable socket support under Windows.
1051    
1052    To apply the patch (using unix or cygwin)
1053    1) gunzip runtime.diff.gz
1054    2) "cd" into "src/runtime" in the source tree of a fresh
1055       110.45 installation.
1056    3) patch -p 1 < [your/path/to]runtime.diff
1057    
1058    The code compiles fine but has NOT yet been extensively tested.
1059    I only ran a few tests for basic socket client functionality
1060    (which worked fine).  Especially the functions that use ioctl
1061    are not tested at all and might not work (see below).
1062    
1063    I implemented this since we want to move to a newer version of sml/nj
1064    but need socket support in order to use it.  This is the first time I
1065    even had a look at the sml/nj source,  so please review my changes
1066    before making this part of the distribution!  Here are a few issues
1067    that I think might be better for someone to solve who is more
1068    familiar with the sml/nj source (and socket programming):
1069    
1070    - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
1071      exception since I could not figure out what the windows equivalent
1072      of these functions is
1073    
1074    - In sockets-osdep.h there are a some #include statements that are
1075      only used in a few files that include sockets-osdep.h
1076    
1077    - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
1078      does not process its return value since I don't know how to
1079      report an error upwards.
1080    
1081    - It would probably be good to have a call to WSACleanup() when
1082      the library is unloaded (if there is such a possibility).
1083      Otherwise I think Windows will take care of this automatically
1084      when the process finishes.
1085    
1086    - I used ioctlsocket() as a replacement for ioctl() but I have
1087      no idea if that is actually the proper replacement on Windows.
1088    
1089    - All these issues are marked in the code by "FIXME" comments.
1090    
1091    We use sml/nj extensively in our products and are quite happy
1092    with it.  I hope this contribution will help you.
1093    
1094    Keep up the good work!
1095    
1096    David
1097    
1098    ----------------------------------------------------------------------
1099    Name: Matthias Blume (blume (at) tti - c (dot) org)
1100    Date: 2004/05/11 14:20:00 CDT
1101    Tag: blume-20040511-installml
1102    Description:
1103    
1104    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
1105    the patch.)
1106    
1107    ----------------------------------------------------------------------
1108    Name: Matthias Blume (blume (at) tti - c (dot) org)
1109    Date: 2004/05/11 14:05:00 CDT
1110    Tag: blume-20040511-nlffi-netbsd
1111    Description:
1112    
1113    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
1114    the patch.)
1115    
1116    ----------------------------------------------------------------------
1117    Name: Matthias Blume (blume (at) tti - c (dot) org)
1118    Date: 2004/05/11 12:05:00 CDT
1119    Tag: blume-20040511-exports
1120    Description:
1121    
1122    As per request by Adam Chlipala <adam@hcoop.net>, extended various
1123    export lists in compiler-related .cm-files.
1124    
1125    ----------------------------------------------------------------------
1126    Name: Matthias Blume (blume (at) tti - c (dot) org)
1127    Date: 2004/05/11 11:35:00 CDT
1128    Tag: blume-20040511-allsource
1129    Description:
1130    
1131    The installer now honors the "src-smlnj" target again, although its meaning
1132    has changed from "all sources required for the compiler" to "all sources
1133    the installer knows about".  In other words, if you enable "src-smlnj"
1134    in the "targets" file, then the installer will pull in sources for
1135    everything.  (Notice that this refers to source code only.  Compiled
1136    code is still only installed for modules that were requested explicitly
1137    or which are required for other modules that were requested explicitly.)
1138    
1139    ----------------------------------------------------------------------
1140    Name: Matthias Blume (blume (at) tti - c (dot) org)
1141    Date: 2004/04/23 17:40:00 CDT
1142    Tag: blume-20040423-ieee-scan
1143    Description:
1144    
1145    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
1146    in the exponent calculation we get INF or ZERO (depending on the mantissa
1147    and the sign of the exponent).
1148    
1149    ----------------------------------------------------------------------
1150    Name: Matthias Blume (blume (at) tti - c (dot) org)
1151    Date: 2004/04/23 10:40:00 CDT
1152    Tag: blume-20040423-ml-build
1153    Description:
1154    
1155    The ml-build script now terminates with a non-0 status when something
1156    goes wrong.
1157    
1158    ----------------------------------------------------------------------
1159    Name: Matthias Blume (blume (at) tti - c (dot) org)
1160    Date: 2004/04/22 16:35:00 CDT
1161    Tag: blume-20040422-Option
1162    Description:
1163    
1164    Made exception Option to be the same as exception Option.Option
1165    (as it should be).
1166    
1167    ----------------------------------------------------------------------
1168    Name: Allen Leung (leunga (at) reservoir (dot) com)
1169    Date: 2004/03/19 14:40:00 EST
1170    Tag: leunga-20040319-cygwin-nlffi
1171    Description:
1172    
1173    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
1174    of SML/NJ.  The problem is that
1175    
1176       lib = dlopen(NULL, ...)
1177       f   = dlsym(lib, "malloc");
1178    
1179    does not work on Windows unless we explicitly export symbols
1180    such as 'malloc' during linking.  We fixed this by explicitly
1181    exporting the required symbols with the magic gcc incantation:
1182    
1183        -Wl,--export-all cygwin.def
1184    
1185    where cygwin.def is a file containing all the symbols that
1186    we wish to export.
1187    
1188    I suspect this is a Windows problem and we'll have to
1189    do the same (somehow with windows compilers) when
1190    we build the native win32 version with the system
1191    calls LoadLibrary/GetProcAddress.
1192    
1193    ----------------------------------------------------------------------
1194    Name: Matthias Blume (blume (at) tti - c (dot) org)
1195    Date: 2004/03/04 16:35:00 CST
1196    Tag: blume-20040304-intinf-fmt
1197    Description:
1198    
1199    Fixed problem with IntInf.fmt (sign would show up on the right instead
1200    of on the left for BIN, OCT, and HEX).
1201    
1202    ----------------------------------------------------------------------
1203    Name: Matthias Blume (blume (at) tti - c (dot) org)
1204    Date: 2004/03/04 11:25:00 CST
1205    Tag: blume-20040304-symlinks
1206    Description:
1207    
1208    Fixed problem with installer script (unix only) where bin/ml-yacc and
1209    friends pointed (via symlinks) to absolute locations instead of just
1210    .run-sml.  This was reported by Vesa A Norrman.
1211    
1212    ----------------------------------------------------------------------
1213    Name: Matthias Blume (blume (at) tti - c (dot) org)
1214    Date: 2004/02/13 14:50:00 CST
1215    Tag: Release_110_45
1216    Description:
1217    
1218    New working version (110.45). New bootfiles.
1219    
1220    ----------------------------------------------------------------------
1221    Name: Matthias Blume (blume (at) tti - c (dot) org)
1222    Date: 2004/01/26 15:15:15 CST
1223    Tag: blume-20040126-toplevel
1224    Description:
1225    
1226    Improved handling of exceptions at the interactive toplevel.
1227    
1228    ----------------------------------------------------------------------
1229    Name: Matthias Blume (blume (at) tti - c (dot) org)
1230    Date: 2004/01/26 11:25:00
1231    Tag: blume-20040126-app
1232    Description:
1233    
1234    Type of top-level "app" corrected.
1235    Added code for setting vp_limitPtrMask to Win32-specific runtime.
1236    
1237    ----------------------------------------------------------------------
1238    Name: Matthias Blume (blume (at) tti - c (dot) org)
1239    Date: 2003/11/18 17:10 CST
1240    Tag: blume-20031118-basis-fiddle
1241    Description:
1242    
1243    - changed Timer interface to what might become the spec
1244    - POSIX_FLAGS -> BIT_FLAGS according to spec
1245    - some other minor discrepancies wrt. spec eliminated
1246    
1247    ----------------------------------------------------------------------
1248    Name: Matthias Blume (blume (at) tti - c (dot) org)
1249    Date: 2003/11/06 12:00:00 CST
1250    Tag: Release_110_44
1251    Description:
1252    
1253    New working version (110.44). New bootfiles.
1254    
1255    ----------------------------------------------------------------------
1256    Name: Matthias Blume (blume (at) tti - c (dot) org)
1257    Date: 2003/11/04 11:50:00 CST
1258    Tag: blume-20031104-move-libraries
1259    Description:
1260    
1261    Eliminated the "dont_move_libraries" directive in config/targets.
1262    (The mechanism was broken and could not be fixed easily.  Moreover,
1263    there does not seem to be any reason not to move all libraries into
1264    lib during installation.  I originally implemented this directive as a
1265    backward-compatibility feature when I first introduced the new CM.
1266    Now that things have been stable for a long time and going back to the
1267    old CM is not an option, there is no reason to keep it around.)
1268    
1269    ----------------------------------------------------------------------
1270    Name: Matthias Blume (blume (at) tti - c (dot) org)
1271    Date: 2003/11/03 16:00:00 CST
1272    Tag: blume-20031103-installdir
1273    Description:
1274    
1275    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
1276    Richards for pointing out the problem and providing the solution.)
1277    
1278    ----------------------------------------------------------------------
1279    Name: Matthias Blume (blume (at) tti - c (dot) org)
1280    Date: 2003/10/01 17:05:00 CDT
1281    Tag: blume-20031001-lal-mlrisc
1282    Description:
1283    
1284    MLRISC bug fix from Lal.
1285    
1286    ----------------------------------------------------------------------
1287    Name: Matthias Blume (blume (at) tti - c (dot) org)
1288    Date: 2003/09/30 16:10:00 CDT
1289    Tag: blume-20030930-primio-bat
1290    Description:
1291    
1292    1. Added openVector, nullRd, and nullWr to PRIM_IO.
1293    2. Improved .bat files (for Win32 port) to make things work under Win95.
1294       (thanks to Aaron S. Hawley for this one)
1295    
1296    ----------------------------------------------------------------------
1297    Name: Matthias Blume (blume (at) tti - c (dot) org)
1298    Date: 2003/09/26 16:05:00 CDT
1299    Tag: blume-20030926-wrappriv
1300    Description:
1301    
1302    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
1303    
1304    ----------------------------------------------------------------------
1305    Name: Matthias Blume (blume (at) tti - c (dot) org)
1306    Date: 2003/09/26 15:00:00 CDT
1307    Tag: blume-20030926-110_43_3
1308    Description:
1309    
1310    - additional cleanup
1311    - version number bump, NEW BOOTFILES
1312    
1313    ----------------------------------------------------------------------
1314    Name: Matthias Blume (blume (at) tti - c (dot) org)
1315    Date: 2003/09/26 12:00:00 CDT
1316    Tag: blume-20030926-ppautoload
1317    Description:
1318    
1319    I modified the read-eval-print loop so that the autoloader gets
1320    invoked whenever the prettyprinter tries to look up a symbol that
1321    is not currently defined in the toplevel environment but which
1322    appears in CM's autoload registry.  As a result, we see far fewer of
1323    those ?.Foo.Bar.xxx names in the prettyprinter's output.
1324    
1325    In addition to this I tried to clean up some pieces of the Basis
1326    implementation (e.g., Socket, Word8Array) in order to prevent other
1327    instances of these ?.Foo.Bar.xxx names from being printed.
1328    
1329    The mechanism that picks names for types still needs some work, though.
1330    (Right now it seems that if there is a type A.t which is defined to
1331    be B.u, but B is unavailable at toplevel, then A.t gets printed as
1332    "?.B.u" although the perhaps more sensible solution would be to use
1333    "A.t" in this case.  In other words, the prettyprinter should follow
1334    a chain of DEFtycs not farther than there are corresponding toplevel
1335    names in the current environment.)
1336    
1337    ----------------------------------------------------------------------
1338    Name: Matthias Blume (blume (at) tti - c (dot) org)
1339    Date: 2003/09/24 16:31:00 CDT
1340    Tag: blume-20030924-installer
1341    Description:
1342    
1343    Another installer tweak:  All the ML code for the installer is now
1344    compiled during CMB.make and put into a little library called
1345    $smlnj/installer.cm.  The installation then simply invokes
1346    
1347       sml -m $smlnj/installer.cm
1348    
1349    and everything happens automagically.
1350    
1351    Win32:  ML code senses value of environment variable SMLNJ_HOME.
1352    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
1353            and BINDIR.
1354    
1355    The new scheme guarantees that the ML code responsible for the installation
1356    is in sync with the APIs of the main system.  Also, the installer is
1357    somewhat faster because the installer script is precompiled.
1358    
1359    ----------------------------------------------------------------------
1360    Name: Matthias Blume (blume (at) tti - c (dot) org)
1361    Date: 2003/09/24 15:35:00 CDT
1362    Tag: blume-20030924-synsock
1363    Description:
1364    
1365    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
1366    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
1367    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
1368    implementations of non-blocking operations from CML's Socket
1369    structure.
1370    
1371    ----------------------------------------------------------------------
1372    Name: Matthias Blume (blume (at) tti - c (dot) org)
1373    Date: 2003/09/24 15:10:05 CDT
1374    Tag: blume-20030924-sockets
1375    Description:
1376    
1377    1. Fixed SOCKET API and implementation to match Basis spec.
1378       This required changing the internal representation of sockets to one
1379       that remembers (for each socket file descriptor) whether it is currently
1380       blocking or non-blocking.  This state is maintained lazily (i.e., a system
1381       call is made only if the state actually needs to change).
1382    
1383    2. OS-specific details of sockets were moved into separate files, thus
1384       making it possible to unify the bulk of the socket implementations
1385       between Unix and Win32.
1386    
1387    3. CML's socket API changed accordingly.
1388       (Note that we need to remove non-blocking functions from this API
1389       since they are redundant in the case of CML!)
1390    
1391    4. CML's socket implementation now makes use of non-blocking functions
1392       provided by Basis, thus removing all OS-dependent code from this part
1393       of CML.
1394    
1395    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
1396    
1397    ----------------------------------------------------------------------
1398    Name: Matthias Blume (blume (at) tti - c (dot) org)
1399    Date: 2003/09/22 12:10:00 CDT
1400    Tag: blume-20030922-110_43_2
1401    Description:
1402    
1403    Made a new interim version and bootfiles for developer's bootstrapping
1404    convenience.
1405    
1406       110.43.2 -- NEW BOOTFILES
1407    
1408    ----------------------------------------------------------------------
1409    Name: Matthias Blume (blume (at) tti - c (dot) org)
1410    Date: 2003/09/19 15:55:00 CDT
1411    Tag: blume-20030919-cmdir
1412    Description:
1413    
1414    1. new-install.sh -> install.sh
1415    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
1416    3. tweaked installer so that another name instead of .cm can be chosen
1417       at install time (by setting the CM_DIR_ARC environment variable
1418       during installation); once installation is complete, the name is
1419       fixed
1420    
1421    ----------------------------------------------------------------------
1422    Name: Matthias Blume (blume (at) tti - c (dot) org)
1423    Date: 2003/09/18 16:00:00 CDT
1424    Tag: blume-20030918-110_43_1
1425    Description:
1426    
1427    Made a new interim version and bootfiles for developer's bootstrapping
1428    convenience.
1429    
1430       110.43.1 -- NEW BOOTFILES
1431    
1432    ----------------------------------------------------------------------
1433    Name: Matthias Blume (blume (at) tti - c (dot) org)
1434    Date: 2003/09/18 15:20:00 CDT
1435    Tag: blume-20030918-misc
1436    Description:
1437    
1438    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
1439       this seems to be controversial at the moment) and used those in
1440       Posix.ProcEnv.times.
1441    
1442    2. Added Time.{from,to}Nanoseconds to Time.
1443    
1444    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
1445       For example, fromLargeInt never needs to look at more than 3 "big
1446       digits" to get its 53 bits of precision.
1447    
1448    ----------------------------------------------------------------------
1449    Name: Matthias Blume (blume (at) tti - c (dot) org)
1450    Date: 2003/09/17 16:30:00 CDT
1451    Tag: blume-20030917-real32-slices
1452    Description:
1453    
1454    Added an entry to the primitive environment
1455    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
1456    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
1457    
1458    Removed some of the "magic" constants in real64.sml and replaced them
1459    with code that generates these values from their corresponding
1460    integer counterparts.
1461    
1462    Made all(?) the slice-related changes to the Basis and made everything
1463    compile again...
1464    
1465    ----------------------------------------------------------------------
1466    Name: Matthias Blume (blume (at) tti - c (dot) org)
1467    Date: 2003/09/15 17:45:00 CDT
1468    Tag: blume-20030915-rbase
1469    Description:
1470    
1471    Fixed bug in Real.fromLargeInt.
1472    
1473    ----------------------------------------------------------------------
1474    Name: Matthias Blume (blume (at) tti - c (dot) org)
1475    Date: 2003/09/13 18:11:00 CDT
1476    Tag: blume-20030913-libinstall
1477    Description:
1478    
1479    Minor bugfix in config/libinstall (set anchor with path to
1480    standalone tool after installing it, otherwise libraries that
1481    need ml-lex or ml-yacc won't compile the first time the installer
1482    runs).
1483    
1484    ----------------------------------------------------------------------
1485    Name: Matthias Blume (blume (at) tti - c (dot) org)
1486    Date: 2003/09/12 11:45:00 CDT
1487    Tag: blume-20030912-various
1488    Description:
1489    
1490     - fixed bug in Real.toLargeInt
1491     - fixed bug in Posix.ProcEnv.times
1492     - changed inputLine functions to return an option
1493     - minor installer improvements / bugfixes
1494     - changed default @SMLalloc parameter for x86/celeron to 64k
1495    
1496    ----------------------------------------------------------------------
1497    Name: Matthias Blume (blume (at) tti - c (dot) org)
1498    Date: 2003/09/09 22:00:00 CDT
1499    Tag: Release_110_43
1500    Description:
1501    
1502        New working release 110.43.  New bootfiles.
1503    
1504    ----------------------------------------------------------------------
1505    Name: Matthias Blume (blume (at) tti - c (dot) org)
1506    Date: 2003/09/09 19:20:00 CDT
1507    Tag: blume-20030909-installer
1508    Description:
1509    
1510    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
1511    Modified config/install.bat to take advantage of it.  Also modified
1512    config/install.sh (and called it config/new-install.sh) to take advantage
1513    of it on Unix systems.  (The SML code is (supposed to be) platform-
1514    independent.)
1515    
1516    The installer can now install everything under Win32
1517    as well as under *nix as long as it compiles.
1518    
1519    Other changes:
1520    
1521      - made CML compile again under Win32
1522      - made eXene compile under Win32 (by providing a fake structure UnixSock
1523        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
1524      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
1525        same as type int;  under Win32 it isn't
1526      - fixed some slice-related problems in the win32-specific parts of CML
1527      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
1528        (under Win32, the volume name is case-insensitive, and the
1529        OS.Path code compares volume names for equality)
1530    
1531    ----------------------------------------------------------------------
1532    Name: Matthias Blume (blume (at) tti - c (dot) org)
1533    Date: 2003/09/08 11:55:00 CDT
1534    Tag: blume-20030908-fullpath
1535    Description:
1536    
1537    Made Win32 version of OS.FileSys.fullPath return current directory
1538    when given an empty string.  This is what the spec says, and incidentally,
1539    CM depends on it.  (CM otherwise goes into an infinite loop in certain
1540    cases when presented with the name of a non-existing .cm file.)
1541    
1542    ----------------------------------------------------------------------
1543    Name: Matthias Blume (blume (at) tti - c (dot) org)
1544    Date: 2003/09/04 16:30:00 CDT
1545    Tag: blume-20030905-slices-etc
1546    Description:
1547    
1548      1. Changed interface to vectors and arrays in Basis to match
1549         (draft) Basis spec.
1550      2. Added signatures and implementations of slices according to
1551         Basis spec.
1552      3. Edited source code throughout the system to make it compile again
1553         under 1. and 2.  (In some cases code had to be added to have it
1554         match the new signatures.)
1555      4. MLRISC should be backward-compatible: the copies of the originals
1556         of files that needed to change under 3. were retained, the .cm files
1557         check the compiler version number and use old versions when
1558         appropriate.
1559      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
1560         dirstream -> string option (in accordance with Basis spec).
1561      6. When generating code that counts lines, ml-lex used function
1562         CharVector.foldli, taking advantage of its old interface.
1563         This has been replaced with the corresponding code from
1564         CharVectorSlice.  (html-lex must be re-lexed!)
1565      7. BitArray in smlnj-lib/Util has been extended/modified to match the
1566         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
1567      8. Removed temporary additions (fromInternal, toInternal) from the
1568         (now obsolete) IntInf in smlnj-lib/Util.
1569      9. Cleaned up structure Byte.
1570     10. Added localOffset, scan, and fromString to Date (according to spec).
1571         Cleaned/corrected implementation of Date.
1572         (Still need to check for correctness; implement better canonicalizeDate.)
1573     11. Added "scan" to signature IEEE_REAL.
1574     12. Some improvements to IntInf [in particular: efficiency-hack for
1575         mod and rem when second operand is 2 (for parity checks).]
1576     13. Changed representation of type Time.time, using a single IntInf.int
1577         value counting microseconds.  This considerably simplified the
1578         implementation of structure Time.  We now support negative time
1579         values; scan and fromString handle signs.
1580     14. Functor PrimIO now takes two additional arguments (VectorSlice and
1581         ArraySlice).
1582    
1583    ----------------------------------------------------------------------
1584    Name: Matthias Blume (blume (at) tti - c (dot) org)
1585    Date: 2003/08/28 17:00:00 CDT
1586    Tag: blume-20030828-intinf
1587    Description:
1588    
1589    This is a major update which comes with a version number bump
1590    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
1591    and an implementation of IntInf in the Basis.
1592    
1593    There are a fairly large number of related changes and updates throughout
1594    the system:
1595    
1596    Basis:
1597            - Implemented IntInf.
1598            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
1599            - Added some missing Real64 operations, most notably Real.toLargeInt.
1600            - Added FixedInt as a synonym for Int32.
1601    
1602    compiler:
1603            * Added support for a built-in intinf type.
1604              - literals
1605              - pattern matching
1606              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
1607              - overloading on literals and operations
1608    
1609            This required adding a primitive type intinf, some additional
1610            primops, and implementations for several non-trivial intinf
1611            operations in Core.  (The intinf type is completely abstract
1612            to the compiler; all operations get delegated back to the Core.)
1613    
1614            * Intinf equality is handled by polyequal.  However, the compiler
1615            does not print its usual warning in this case (since polyequal
1616            is the right thing to do there).
1617    
1618            * Improved the organization of structure InlineT.
1619    
1620            * A word about conversion primops:
1621            If conversions involving intinf do not cancel out during
1622            CPS contract, then the compiler must insert calls to Core functions.
1623            Since all core access must be resolved already during the FLINT
1624            translate phase, it would be too late a the time of CPS contract
1625            to add new Core calls.  For this reason, conversion primops
1626            for intinf carry two arguments:  1. the numeric argument that
1627            they are supposed to convert, and 2. the Core function that
1628            can help with this conversion if necessary.  If CPS contract
1629            eliminates a primop, then the associated Core function becomes
1630            dead and goes away.  Intinf conversion primops that do not get
1631            eliminated by CPS contract get rewritten into calls of their
1632            core functions by a separate, new phase.
1633    
1634    interactive system:
1635            - Control.Print.intinfDepth controls max length of intinf constants
1636            being printed. (Analogous to Control.Print.stringDepth.)
1637            - Cleanup in printutil and pputil: got rid of unused stuff and
1638            duplicates; replaced some of the code with code that makes better
1639            use of library functionality.
1640    
1641    CM:
1642            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
1643            are no longer silent.
1644    
1645    CKIT:
1646            Fixed mismatched uses of Int32 and LargeInt.  I always decided
1647            in favor of LargeInt -- which is now the same as IntInf.
1648            CKIT-knowledgable people should check whether this is what's
1649            intended and otherwise change things back to using Int32 or
1650            FixedInt.
1651    
1652    Throughout the code:
1653            Started using IntInf.int literals and built-in operations
1654            (e.g., comparison with 0) where this seems appropriate.
1655    
1656    
1657    ----------------------------------------------------------------------
1658    Name: Dave MacQueen (dbm@cs.uchicago.edu)
1659    Date: 2003/08/13 11:36:00 CDT
1660    Tag: dbm-20030813-mcz-merge1
1661    Description:
1662    
1663    Merging changes from the mcz-branch development branch into trunk.
1664    These changes involve replacement of the emulated old prettyprinter
1665    interface with direct use of the SML/NJ Lib PP library, and fixing
1666    of a couple of bugs (895, 1186) relating to error messages.  A new
1667    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
1668    has been added.
1669    
1670    ----------------------------------------------------------------------
1671    Name: Matthias Blume (blume (at) tti - c (dot) org)
1672    Date: 2003/08/11 15:45:00 CDT
1673    Tag: blume-20030811-windows
1674    Description:
1675    
1676    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
1677    
1678        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
1679    
1680    This patch restores SML/NJ's ability to run under win32.  There are a
1681    number of changes, including fixes for several bugs that had gone
1682    unnoticed until now:
1683    
1684      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
1685        trying to run the win32 version from within cygwin.)
1686    
1687      - There are a number of simple .bat scripts that substitute for their
1688        corresponding Unix shell-scripts.  (See below.)
1689    
1690      - The internals of ml-build have been modified slightly.  The main
1691        difference is that instead of calling ".link-sml" (or link-sml.bat)
1692        using OS.Process.system, the ML process delegates this task back
1693        to the script.  Otherwise problems arise in mixed environments such
1694        as Cygwin where scripts look and work like Unix scripts, but
1695        where OS.Process.system cannot run them.
1696    
1697      - In CM, the srcpath pickler used native pathname syntax -- which
1698        is incorrect in the case of cross-compilation.  The new pickle format
1699        is independent of platform-specific naming conventions.
1700    
1701      - Path configuration files (such as lib/pathconfig) can now choose
1702        between native and standard syntax.  Placing a line of the form
1703    
1704            standard!
1705    
1706        into the file causes all subsequent paths to be interpreted using
1707        CM standard pathname syntax (= Unix conventions); a line
1708    
1709            native!
1710    
1711        switches back to native style.  This was needed so that
1712        path config files can be written portably, see src/system/pathconfig.
1713    
1714      - Runtime system:
1715    
1716          - win32-filesys.c:  get_file_time and set_file_time now
1717            access modification time, not creation time.
1718    
1719          - I/O code made aware of new array representation.
1720    
1721          - Bug fixes in X86.prim.masm.
1722    
1723       - src/system/makeml made aware of win32. (For use under cygwin
1724            and other Unix-environments for windows.)
1725    
1726       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
1727         which caused BinIO.inputAll to fail consistently.
1728    
1729    .bat scripts:
1730    
1731       Windows .bat scripts assume that SMLNJ_HOME is defined.
1732    
1733       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
1734         applications (sml, ml-yacc, ml-lex).
1735       - ml-build.bat: analogous to ml-build.
1736       - config\install.bat:  Analogous to config/install.sh.  This requires
1737         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
1738         (nmake etc. must be on the path, and vcvars32 must have been run.)
1739         Moreover, sources for ml-lex and ml-yacc need to exist under src,
1740         and the bootfile hierarchy must have been unpacked under
1741         sml.boot.x86-win32.
1742         The script is very primitive and does a poor job at error checking.
1743         It only installs the base system, ml-lex, and ml-yacc.  No other
1744         libraries are being installed (i.e., you get only those that
1745         are part of the compiler.)
1746       - link-sml.bat:  analogous to .link-sml, but not currently used
1747    
1748    Unrelated bug fixes:
1749    
1750      - ml-nlffigen now exports structures ST_* corresponding to incomplete
1751        types.
1752      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
1753        otherwise.)
1754    
1755    ----------------------------------------------------------------------
1756    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
1757    Date: 2003/06/17
1758    Tag: macqueen-20030617-bug895
1759    Description:
1760    
1761    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
1762    Tag will be used for new development branch (mcz-branch) for
1763    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
1764    
1765    ----------------------------------------------------------------------
1766    Name: Matthias Blume (blume (at) tti - c (dot) org)
1767    Date: 2003/05/27 16:55:00 CDT
1768    Tag: blume-20030527-polyeq
1769    Description:
1770    
1771    Tried to eliminated most cases of polymorphic equality.
1772    
1773    ----------------------------------------------------------------------
1774    Name: Matthias Blume (blume (at) tti - c (dot) org)
1775    Date: 2003/05/21 17:45:00 CDT
1776    Tag: blume-20030517-complete
1777    Description:
1778    
1779    Two changes:
1780    
1781      1. Added a flag for controlling whether non-exhaustive bindings will
1782         be treated as errors (default is false).
1783      2. Cleaned up the *entire* source tree so that CMB.make goes through
1784         without a single non-exhaustive match- or bind warning.
1785    
1786    ----------------------------------------------------------------------
1787    Name: Matthias Blume (blume (at) tti - c (dot) org)
1788    Date: 2003/05/17 10:20:00 CDT
1789    Tag: blume-20030517-absyn
1790    Description:
1791    
1792    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
1793    
1794       This mainly affects the quality of error messages.  However, some
1795       of the code is now more straightforward than before. (Treatment of
1796       the above four constructs in translate.sml is much simpler than
1797       the "macro-expansion" that was going on before.  Plus, the mach-
1798       compiler no longer gets invoked just to be able to compile an
1799       if-expression.)
1800    
1801    2. The ErrorMsg.Error exception is now caught and absorbed by the
1802       interactive loop.
1803    
1804    ----------------------------------------------------------------------
1805    Name: Allen Leung
1806    Date: 2003/05/16 13:05:00 CDT
1807    Tag: leunga-20030516-cygwin-runtime
1808    Description:
1809    
1810        Ported the runtime system to cygwin, which uses the unix
1811    x86-unix bin files.  Missing/buggy features:
1812    
1813       o getnetbyname, getnetbyaddr: these functions seem to be missing in
1814         the Cygwin library.
1815       o Ctrl-C handling may be flaky.
1816       o Windows system calls and Windows I/O are not supported.
1817    
1818       A new set of binfiles is located at:
1819    
1820         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
1821    
1822       This is only needed for bootstrapping the cygwin version of smlnj.
1823       Other x86 versions can use the existing binfiles.
1824    
1825    ----------------------------------------------------------------------
1826    Name: Matthias Blume
1827    Date: 2003/04/08 15:42:00 CDT
1828    Tag: blume-20030408-listpair
1829    Description:
1830    
1831    1. Added a target 'mlrisc' to installer.
1832    
1833    2. Added missing elements to structure ListPair.
1834    
1835    ----------------------------------------------------------------------
1836    Name: Allen Leung
1837    Date: 2003/01/07 10:40:00 EST
1838    Tag: leunga-20030107-int-rem
1839    Description:
1840    
1841    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
1842    The arguments to the SUBL instruction were swapped.
1843    
1844    ----------------------------------------------------------------------
1845    Name: Matthias Blume
1846    Date: 2002/12/12 16:25:00 EST
1847    Tag: blume-20021212-risc-ra
1848    Description:
1849    
1850    Fixed a serious bug in the rewrite code for FP spilling/reloading that
1851    sent the RA into an infinite loop when floating point registers get
1852    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
1853    110.37 and 110.38.)
1854    There was another set of potential problems related to the handling of
1855    MLRISC annotations (but those did not yet cause real problems, apparently).
1856    
1857    ----------------------------------------------------------------------
1858    Name: Matthias Blume
1859    Date: 2002/12/06 22:40:00 EST
1860    Tag: blume-20021206-cm-fileid
1861    Description:
1862    
1863    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
1864    This fixes the problem of CM getting confused by files that suddenly
1865    change their identity (e.g., by getting unlinked and recreated by some
1866    text editor such as vi).  There might be a better/cheaper/cleaner way
1867    of doing this, but for now this will have to do.
1868    
1869    ----------------------------------------------------------------------
1870    Name: Matthias Blume
1871    Date: 2002/10/28 09:50:00 EST
1872    Tag: blume-20021028-typecheck
1873    Description:
1874    
1875    Exported structure Typecheck from $smlnj/viscomp/core.cm.
1876    
1877    ----------------------------------------------------------------------
1878    Name: Matthias Blume
1879    Date: 2002/10/17 09:10:00 EDT
1880    Tag: Release_110_42
1881    Description:
1882    
1883    In good old tradition, there has been a slight hiccup so that we have
1884    to patch 110.42 after the fact.   The old release tag has been replaced
1885    (see below).
1886    
1887    The change solves a problem with two competing approaches the
1888    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
1889    each other's way.
1890    
1891    This change only affects the runtime system code and the installer script.
1892    (No new bootfiles.)
1893    
1894    ----------------------------------------------------------------------
1895    Name: Matthias Blume
1896    Date: 2002/10/16 12:00:00 EDT
1897    Tag: Release_110_42_removed
1898    Description:
1899    
1900    New working release.  New bootfiles.
1901    
1902    ----------------------------------------------------------------------
1903    Name: Matthias Blume
1904    Date: 2002/10/10 13:10:00 EDT
1905    Tag: blume-20021010-ppc-divs
1906    Description:
1907    
1908    The mltree operator DIVS must be implemented with an overflow check on
1909    the PPC because the hardware indicates divide-by-zero using "overflow" as
1910    well.
1911    
1912    ----------------------------------------------------------------------
1913    Name: Matthias Blume
1914    Date: 2002/07/23 11:45:00
1915    Tag: blume-20020723-smlnj-home
1916    Description:
1917    
1918    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
1919    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
1920    CM_PATHCONFIG is also set), the path configuration file is assumed
1921    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
1922    the entire tree to some other place and everything will "just work".
1923    
1924    (Companion commands such as ml-build and ml-makedepend also sense this
1925    variable.)
1926    
1927    ----------------------------------------------------------------------
1928    Name: Matthias Blume
1929    Date: 2002/07/12 21:19:00 EDT
1930    Tag: blume-20020712-liveness
1931    Description:
1932    
1933    Exported two useful "step" functions from liveness module (MLRISC).
1934    
1935    ----------------------------------------------------------------------
1936    Name: Matthias Blume
1937    Date: 2002/07/05 16:00 EDT
1938    Tag: Release_110_41
1939    Description:
1940    
1941    New working release.  New bootfiles.
1942    
1943    ----------------------------------------------------------------------
1944    Name: Matthias Blume
1945    Date: 2002/07/05 10:25:00 EDT
1946    Tag: blume-20020705-btimp
1947    Description:
1948    
1949    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
1950    other clients can set up backtracing support.
1951    
1952    ----------------------------------------------------------------------
1953    Name: Matthias Blume
1954    Date: 2002/06/25 17:23:00 EDT
1955    Tag: blume-20020625-fpmax
1956    Description:
1957    
1958    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
1959    case.  (The sense of the isNaN test was reversed -- which made min and
1960    max always return their first argument.)
1961    
1962    ----------------------------------------------------------------------
1963    Name: Matthias Blume
1964    Date: 2002/06/11
1965    Tag: blume-20020611-unixpath
1966    Description:
1967    
1968    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
1969    
1970    ----------------------------------------------------------------------
1971    Name: Matthias Blume
1972    Date: 2002/06/10 16:35:00 EDT
1973    Tag: blume-20020610-ieeereal
1974    Description:
1975    
1976    I back-ported my implementation of IEEEReal.fromString from the
1977    idlbasis-devel branch so that we can test it.
1978    
1979    Another small change is that ppDec tries to give more information
1980    than just "<sig>" in the case of functors.  However, this code is
1981    broken in some mysterious way if the functor's body's signature
1982    has not been declared by ascription but gets inferred from the
1983    implementation.  This needs fixing...
1984    
1985    ----------------------------------------------------------------------
1986    Name: Matthias Blume
1987    Date: 2002/05/31
1988    Tag: blume-20020531-btrace-mode
1989    Description:
1990    
1991    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
1992    the wayside when I switched over to using Controls everywhere.)
1993    
1994    ----------------------------------------------------------------------
1995    Name: Lal George
1996    Date: 2002/05/23 12:21:40 EDT
1997    Tag: george-20020523-visual-labels
1998    Description:
1999    
2000    Labels are now displayed in the graphical output to make
2001    the fall-through and target blocks obvious.
2002    
2003    ----------------------------------------------------------------------
2004    Name: Matthias Blume
2005    Date: 2002/05/22 11:03:00 EDT
2006    Tag: blume-20020522-shrink
2007    Description:
2008    
2009    John tweaked yesterday's fix for 1131 to handle an out-of-memory
2010    situation that comes up when allocating huge arrays.
2011    
2012    ----------------------------------------------------------------------
2013    Name: Matthias Blume
2014    Date: 2002/05/21 16:00:00 EDT
2015    Tag: Release_110_40
2016    Description:
2017    
2018    New working release (110.40).  New bootfiles.
2019    
2020    [Also: John Reppy fixed GC bug 1131.]
2021    
2022    ----------------------------------------------------------------------
2023    Name: Matthias Blume
2024    Date: 2002/05/21 12:35:00 EDT
2025    Tag: blume-20020521-cmdoc
2026    Description:
2027    
2028    CM documentation update.
2029    
2030    ----------------------------------------------------------------------
2031    Name: Matthias Blume
2032    Date: 2002/05/21 10:55:00 EDT
2033    Tag: blume-20020521-misc
2034    Description:
2035    
2036    - John tweaked runtime to be silent on heap export (except when
2037      GC messages are on).
2038    - I added a few more things (cross-compiling versions of CMB) to
2039      config/preloads (as suggestions).
2040    
2041    ----------------------------------------------------------------------
2042    Name: Matthias Blume
2043    Date: 2002/05/20 22:25:00 EDT
2044    Tag: blume-20020520-controls
2045    Description:
2046    
2047    - Added ControlUtil structure to control-lib.cm.
2048    - Use it throughout.
2049    - Used Controls facility to define MLRISC controls (as opposed to
2050      registering MLRISC control ref cells with Controls after the fact)
2051    - Fixed messed-up controls priorities.
2052    
2053    * Removed again all the stuff from config/preloads that one wouldn't
2054      be able to preload at the time the initial heap image is built.
2055      (Many libraries, e.g., CML, do not exist yet at this time.  The
2056       only libraries that can be preloaded via config/preloads are those
2057       that come bundled with the bootfiles.)
2058    
2059    ----------------------------------------------------------------------
2060    Name: Matthias Blume
2061    Date: 2002/05/20 10:59:00 EDT
2062    Tag: blume-20020520-preloads
2063    Description:
2064    
2065    Added a lot of commented-out suggestions for things to be included
2066    in config/preloads.
2067    
2068    ----------------------------------------------------------------------
2069    Name: Allen Leung
2070    Date: 2002/05/18 14:20:00 EDT
2071    Tag: leunga-20020518-mdl
2072    Description:
2073    
2074    o Made the mdl tool stuff compile and run again.
2075    o I've disabled all the stuff that depends on RTL specifications; they
2076      are all badly broken anyway.
2077    
2078    ----------------------------------------------------------------------
2079    Name: Matthias Blume
2080    Date: 2002/05/17 16:49:00 EDT
2081    Tag: blume-20020517-controls
2082    Description:
2083    
2084    0. John Reppy made several modifications to the SML/NJ library.
2085       In particular, there is a shiny new controls-lib.cm.
2086    
2087    1. Pushed new controls interface through compiler so that everything
2088       compiles again.
2089    
2090    2. Added FormatComb and FORMAT_COMB to the CML version of the
2091       SML/NJ library (so that CML compiles again).
2092    
2093    3. Modified init scripts because XXX_DEFAULT environment variables
2094       are no longer with us.  (Boot-time initialization is now done
2095       using the same environment variables that are also used for
2096       startup-time initialization of controls.)
2097    
2098    ----------------------------------------------------------------------
2099    Name: Lal George
2100    Date: 2002/05/15 09:20:10 EDT
2101    Tag: george-20020515-pseudo-op-decls
2102    Description:
2103    
2104            All pseudo-ops emitted before the first segment declaration
2105    such as TEXT, DATA, and BSS directives are assumed to be global
2106    declarations and are emitted first in the assembly file. This is
2107    useful in a number of situations where one has pseudo-ops that are not
2108    specific to any segment, and also works around the constraint that one
2109    cannot have client pseudo-ops in the TEXT segment.
2110    
2111            Because no segment is associated with these declarations it is
2112    an error to allocate any space or objects before the first segment
2113    directive and an exception will be raised. However, we cannot make
2114    this check for client pseudo-ops.
2115    
2116            These top level declarations are a field in the CFG graph_info.
2117    In theory you can continue to add to this field after the CFG has been
2118    built -- provided you know what you are doing;-)
2119    
2120    ----------------------------------------------------------------------
2121    Name: Matthias Blume
2122    Date: 2002/05/13 16:40:00 EDT
2123    Tag: blume-20020513-pp-etc
2124    Description:
2125    
2126    A few minor bugfixes:
2127    
2128      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
2129        (Bogus printouts for redefined bindings still occur.  Compiler
2130        bug should no longer occur now.  We need to redo the prettyprinter
2131        from scratch.)
2132    
2133      - CM pathname printer now also adds escape sequences for ( and )
2134    
2135      - commend and docu fixes for ml-nlffi
2136    
2137    ----------------------------------------------------------------------
2138    Name: Matthias Blume
2139    Date: 2002/05/10 16:40:00 EDT
2140    Tag: blume-20020510-erg-textio
2141    Description:
2142    
2143    Applied the following bugfix provided by Emden Gansner:
2144    
2145        Output is corrupted when outputSubstr is used rather than output.
2146    
2147        The problem occurs when a substring
2148    
2149            ss = (s, dataStart, dataLen)
2150    
2151        where dataStart > 0, fills a stream buffer with avail bytes left.
2152        avail bytes of s, starting at index dataStart, are copied into the
2153        buffer, the buffer is flushed, and then the remaining dataLen-avail
2154        bytes of ss are copied into the beginning of the buffer. Instead of
2155        starting this copy at index dataStart+avail in s, the current code
2156        starts the copy at index avail.
2157    
2158        Fix:
2159        In text-io-fn.sml, change line 695 from
2160             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
2161        to
2162             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
2163    
2164    ----------------------------------------------------------------------
2165    Name: Matthias Blume
2166    Date: 2002/04/12 13:55:00 EDT
2167    Tag: blume-20020412-assyntax
2168    Description:
2169    
2170    1. Grabbed newer assyntax.h from the XFree86 project.
2171    2. Fiddled with how to compile X86.prim.asm without warnings.
2172    3. (Very) Minor cleanup in CM.
2173    
2174    ----------------------------------------------------------------------
2175    Name: Matthias Blume
2176    Date: 2002/04/01 (no joke!) 17:07:00 EST
2177    Tag: blume-20020401-x86div
2178    Description:
2179    
2180    Added full support for div/mod/rem/quot on the x86, using the machine
2181    instruction's two results (without clumsily recomputing the remainder)
2182    directly where appropriate.
2183    
2184    Some more extensive power-of-two support was added to the x86 instruction
2185    selector (avoiding expensive divs, mods, and muls where they can be
2186    replaced with cheaper shifts and masks).  However, this sort of thing
2187    ought to be done earlier, e.g., within the CPS optimizer so that
2188    all architectures benefit from it.
2189    
2190    The compiler compiles to a fixed point, but changes might be somewhat
2191    fragile nevertheless.  Please, report any strange things that you might
2192    see wrt. div/mod/quot/rem...
2193    
2194    ----------------------------------------------------------------------
2195    Name: Matthias Blume
2196    Date: 2002/03/29 17:22:00
2197    Tag: blume-20020329-div
2198    Description:
2199    
2200    Fixed my broken div/mod logic.  Unfortunately, this means that the
2201    inline code for div/mod now has one more comparison than before.
2202    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
2203    The problem was with quotient = 0, because that alone does not tell
2204    us which way the rounding went.  One then has to look at whether
2205    remainder and divisor have the same sign...  :(
2206    
2207    Anyway, I replaced the bootfiles with fresh ones...
2208    
2209    ----------------------------------------------------------------------
2210    Name: Matthias Blume
2211    Date: 2002/03/29 14:10:00 EST
2212    Tag: blume-20020329-inlprims
2213    Description:
2214    
2215    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
2216    
2217    Primops have changed. This means that the bin/boot-file formats have
2218    changed as well.
2219    
2220    To make sure that there is no confusion, I made a new version.
2221    
2222    
2223    CHANGES:
2224    
2225    * removed REMT from mltree (remainder should never overflow).
2226    
2227    * added primops to deal with divisions of all flavors to the frontend
2228    
2229    * handled these primops all the way through so they map to their respective
2230      MLRISC support
2231    
2232    * used these primops in the implementation of Int, Int32, Word, Word32
2233    
2234    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
2235    
2236    * parameterized INLMIN, INLMAX, and INLABS by a numkind
2237    
2238    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
2239      floating point
2240    
2241    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
2242      and Real (but Real.abs maps to a separate floating-point-only primop)
2243    
2244    
2245    TODO items:
2246    
2247    * Hacked Alpha32 instruction selection, disabling the selection of REMx
2248      instructions because the machine instruction encoder cannot handle
2249      them.  (Hppa, PPC, and Sparc instruction selection did not handle
2250      REM in the first place, and REM is supported by the x86 machine coder.)
2251    
2252    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
2253      selection phase.  (The two can be streamlined because the hardware
2254      delivers both quotient and remainder at the same time anyway.)
2255    
2256    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
2257      (Currently the behavior is inconsistent both across architectures and
2258      wrt. the draft Basis spec.)
2259    
2260    * Word8 should eventually be handled natively, too.
2261    
2262    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
2263      as if there currently is no execution path that could trigger it in
2264      SML/NJ.  (The assumptions underlying functions arith and promotable do not
2265      hold for things like multiplication and division.)
2266    
2267    ----------------------------------------------------------------------
2268    Name: Matthias Blume
2269    Date: 2002/03/27 16:27:00 EST
2270    Tag: blume-20020327-mlrisc-divisions
2271    Description:
2272    
2273    Added support for all four division operations (ML's div, mod, quot,
2274    and rem) to MLRISC.  In the course of doing so, I also rationalized
2275    the naming (no more annoying switch-around of DIV and QUOT), by
2276    parameterizing the operation by div_rounding_mode (which can be either
2277    DIV_TO_ZERO or DIV_TO_NEGINF).
2278    
2279    The generic MLTreeGen functor takes care of compiling all four
2280    operations down to only round-to-zero div.
2281    
2282    Missing pieces:
2283    
2284      * Doing something smarter than relying on MLTreeGen on architectures
2285        like, e.g., the x86 where hardware division delivers both quotient and
2286        remainder at the same time.  With this, the implementation of the
2287        round-to-neginf operations could be further streamlined.
2288    
2289      * Remove inlining support for div/mod/rem from the frontend and replace it
2290        with primops that get carried through to the backend.  Do this for all
2291        int and word types.
2292    
2293    ----------------------------------------------------------------------
2294    Name: Matthias Blume
2295    Date: 2002/03/25 17:25:00 EST
2296    Tag: blume-20020325-divmod
2297    Description:
2298    
2299    I improved (hopefully without breaking them) the implementation of Int.div,
2300    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
2301    advantage of the following observations:
2302    
2303      Let  q = x quot y      r = x rem y
2304           d = x div  y      m = x mod y
2305    
2306    where "quot" is the round-to-zero version of integer division that
2307    hardware usually provides.  Then we have:
2308    
2309         r = x - q * y        where neither the * nor the - will overflow
2310         d = if q >= 0 orelse x = q * y then q else q - 1
2311                              where neither the * nor the - will overflow
2312         m = if q >= 0 orelse r = 0 then r else r + y
2313                              where the + will not overflow
2314    
2315    This results in substantial simplification of the generated code.
2316    The following table shows the number of CFG nodes and edges generated
2317    for
2318            fun f (x, y) = x OPER y
2319            (* with OPER \in div, mod, quot, rem *)
2320    
2321    
2322        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
2323        --------------------------------------------------------
2324         div |         24 |         39 |         12 |         16
2325         mod |         41 |         71 |         12 |         16
2326        quot |          8 |         10 |          8 |         10
2327         rem |         10 |         14 |          8 |         10
2328    
2329    
2330    ----------------------------------------------------------------------
2331    Name: Matthias Blume
2332    Date: 2002/03/25 22:06:00 EST
2333    Tag: blume-20020325-cprotobug
2334    Description:
2335    
2336    Fixed a bug in cproto (c prototype decoder).
2337    
2338    ----------------------------------------------------------------------
2339    Name: Matthias Blume
2340    Date: 2002/03/25 16:00:00 EST
2341    Tag: blume-20020325-raw-primops
2342    Description:
2343    
2344    I did some cleanup to Allen's new primop code and
2345    replaced yesterday's bootfiles with new ones.
2346    (But they are stored in the same place.)
2347    
2348    ----------------------------------------------------------------------
2349    Name: Matthias Blume
2350    Date: 2002/03/24 22:40:00 EST
2351    Tag: blume-20020324-bootfiles
2352    Description:
2353    
2354    Made the bootfiles that Allen asked for.
2355    
2356    ----------------------------------------------------------------------
2357    Name: Allen Leung
2358    Date: 2002/03/23 15:50:00 EST
2359    Tag: leunga-20020323-flint-cps-rcc-primops
2360    Description:
2361    
2362      1. Changes to FLINT primops:
2363    
2364        (* make a call to a C-function;
2365         * The primop carries C function prototype information and specifies
2366         * which of its (ML-) arguments are floating point. C prototype
2367         * information is for use by the backend, ML information is for
2368         * use by the CPS converter. *)
2369      | RAW_CCALL of { c_proto: CTypes.c_proto,
2370                       ml_args: ccall_type list,
2371                       ml_res_opt: ccall_type option,
2372                       reentrant : bool
2373                     } option
2374       (* Allocate uninitialized storage on the heap.
2375        * The record is meant to hold short-lived C objects, i.e., they
2376        * are not ML pointers.  With the tag, the representation is
2377        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
2378        *)
2379      | RAW_RECORD of {tag:bool,sz:int}
2380      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
2381    
2382      2.  These CPS primops are now overloaded:
2383    
2384           rawload of {kind:numkind}
2385           rawstore of {kind:numkind}
2386    
2387          The one argument form is:
2388    
2389             rawload {kind} address
2390    
2391          The two argument form is:
2392    
2393             rawload {kind} [ml object, byte-offset]
2394    
2395      3. RAW_CCALL/RCC now takes two extra arguments:
2396    
2397         a. The first is whether the C call is reentrant, i.e., whether
2398            ML state should be saved and restored.
2399         b. The second argument is a string argument specifying the name of
2400            library and the C function.
2401    
2402         These things are currently not handled in the code generator, yet.
2403    
2404      4. In CProto,
2405    
2406         An encoding type of "bool" means "ml object" and is mapped into
2407         C prototype of PTR.  Note that "bool" is different than "string",
2408         even though "string" is also mapped into PTR, because "bool"
2409         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
2410    
2411      5. Pickler/unpicker
2412    
2413         Changed to handle RAW_RECORD and newest RAW_CCALL
2414    
2415      6. MLRiscGen,
2416    
2417         1. Changed to handle the new rawload/rawstore/rawrecord operators.
2418         2. Code for handling C Calls has been moved to a new module CPSCCalls,
2419            in the file CodeGen/cpscompile/cps-c-calls.sml
2420    
2421      7. Added the conditional move operator
2422    
2423             condmove of branch
2424    
2425         to cps.  Generation of this is still buggy so it is currently
2426         disabled.
2427    
2428    ----------------------------------------------------------------------
2429    Name: Lal George
2430    Date: 2002/03/22 14:18:25 EST
2431    Tag: george-20020322-cps-branch-prob
2432    Description:
2433    
2434    Implemented the Ball-Larus branch prediction-heuristics, and
2435    incorporated graphical viewers for control flow graphs.
2436    
2437    Ball-Larus Heuristics:
2438    ---------------------
2439    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
2440    
2441    By design it uses the Dempster-Shafer theory for combining
2442    probabilities.  For example, in the function:
2443    
2444        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
2445    
2446    the ball-larus heuristics predicts that the n=0 is unlikely
2447    (OH-heuristic), and the 'then' branch is unlikely because of the
2448    RH-heuristic -- giving the 'then' branch an even lower combined
2449    probability using the Dempster-Shafer theory.
2450    
2451    Finally, John Reppy's loop analysis in MLRISC, further lowers the
2452    probability of the 'then' branch because of the loop in the else
2453    branch.
2454    
2455    
2456    Graphical Viewing:
2457    ------------------
2458    I merely plugged in Allen's graphical viewers into the compiler. The
2459    additional code is not much. At the top level, saying:
2460    
2461            Control.MLRISC.getFlag "cfg-graphical-view" := true;
2462    
2463    will display the graphical view of the control flow graph just before
2464    back-patching.  daVinci must be in your path for this to work. If
2465    daVinci is not available, then the default viewer can be changed
2466    using:
2467    
2468            Control.MLRISC.getString "viewer"
2469    
2470    which can be set to "dot" or "vcg" for the corresponding viewers. Of
2471    course, these viewers must be in your path.
2472    
2473    The above will display the compilation unit at the level of clusters,
2474    many of which are small, boring, and un-interesting. Also setting:
2475    
2476            Control.MLRISC.getInt "cfg-graphical-view_size"
2477    
2478    will display clusters that are larger than the value set by the above.
2479    
2480    
2481    ----------------------------------------------------------------------
2482    Name: Matthias Blume
2483    Date: 2002/03/21 22:20:00 EST
2484    Tag: blume-20020321-kmp-bugfix
2485    Description:
2486    
2487    Changed the interface to the KMP routine in PreString and fixed
2488    a minor bug in one place where it was used.
2489    
2490    ----------------------------------------------------------------------
2491    Name: Allen Leung
2492    Date: 2002/03/21 20:30:00 EST
2493    Tag: leunga-20020321-cfg
2494    Description:
2495    
2496      Fixed a potential problem in cfg edge splitting.
2497    
2498    ----------------------------------------------------------------------
2499    Name: Allen Leung
2500    Date: 2002/03/21 17:15:00 EST
2501    Tag: leunga-20020321-x86-fp-cfg
2502    Description:
2503    
2504      1. Recoded the buggy parts of x86-fp.
2505    
2506         a. All the block reordering code has been removed.
2507            We now depend on the block placement phases to do this work.
2508    
2509         b. Critical edge splitting code has been simplified and moved into the
2510            CFG modules, as where they belong.
2511    
2512         Both of these were quite buggy and complex.  The code is now much, much
2513         simpler.
2514    
2515      2. X86 backend.
2516    
2517         a. Added instructions for 64-bit support.  Instruction selection for
2518            64-bit has not been committed, however, since that
2519            requires changes to MLTREE which haven't been approved by
2520            Lal and John.
2521    
2522         b. Added support for FUCOMI and FUCOMIP when generating code for
2523            PentiumPro and above.  We only generate these instructions in
2524            the fast-fp mode.
2525    
2526         c. Added cases for JP and JNP in X86FreqProps.
2527    
2528      3. CFG
2529    
2530         CFG now has a bunch of methods for edge splitting and merging.
2531    
2532      4. Machine description.
2533    
2534         John's simplification of MLTREE_BASIS.fcond broke a few machine
2535         description things:
2536    
2537         rtl-build.{sig,sml} and hppa.mdl fixed.
2538    
2539         NOTE: the machine description stuff in the repository is still broken.
2540               Again, I can't put my fixes in because that involves
2541               changes to MLTREE.
2542    
2543    ----------------------------------------------------------------------
2544    Name: Matthias Blume
2545    Date: 2002/03/20 15:55:00 EST
2546    Tag: blume-20020320-kmp
2547    Description:
2548    
2549    Implemented Knuth-Morris-Pratt string matching in PreString and used
2550    it for String.isSubstring, Substring.isSubstring, and
2551    Substring.position.
2552    
2553    (Might need some stress-testing.  Simple examples worked fine.)
2554    
2555    ----------------------------------------------------------------------
2556    Name: Matthias Blume
2557    Date: 2002/03/19 16:37:00 EST
2558    Tag: blume-20020319-witnesses
2559    Description:
2560    
2561    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
2562    
2563    This implements a generic mechanism for changing constness qualifiers
2564    anywhere within big C types without resorting to outright "casts".
2565    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
2566    modify the constness at the outermost level.)
2567    The implementation of "convert" is based on the idea of "witness"
2568    values -- values that are not used by the operation but whose types
2569    "testify" to their applicability.  On the implementation side, "convert"
2570    is simply a projection (returning its second curried argument).  With
2571    cross-module inlining, it should not result in any machine code being
2572    generated.
2573    
2574    ----------------------------------------------------------------------
2575    Name: Matthias Blume
2576    Date: 2002/03/15 16:40:00 EST
2577    Tag: blume-20020315-basis
2578    Description:
2579    
2580    Provided (preliminary?) implementations for
2581    
2582      {String,Substring}.{concatWith,isSuffix,isSubstring}
2583    
2584    and
2585    
2586      Substring.full
2587    
2588    Those are in the Basis spec but they were missing in SML/NJ.
2589    
2590    ----------------------------------------------------------------------
2591    Name: Matthias Blume
2592    Date: 2002/03/14 21:30:00 EST
2593    Tag: blume-20020314-controls
2594    Description:
2595    
2596    Controls:
2597    ---------
2598    
2599    1. Factored out the recently-added Controls : CONTROLS stuff and put
2600       it into its own library $/controls-lib.cm.  The source tree for
2601       this is under src/smlnj-lib/Controls.
2602    
2603    2. Changed the names of types and functions in this interface, so they
2604       make a bit more "sense":
2605    
2606          module -> registry
2607          'a registry -> 'a group
2608    
2609    3. The interface now deals in ref cells only.  The getter/setter interface
2610       is (mostly) gone.
2611    
2612    4. Added a function that lets one register an already-existing ref cell.
2613    
2614    5. Made the corresponding modifications to the rest of the code so that
2615       everything compiles again.
2616    
2617    6. Changed the implementation of Controls.MLRISC back to something closer
2618       to the original.  In particular, this module (and therefore MLRISC)
2619       does not depend on Controls.  There now is some link-time code in
2620       int-sys.sml that registers the MLRISC controls with the Controls
2621       module.
2622    
2623    CM:
2624    ---
2625    
2626      * One can now specify the lambda-split aggressiveness in init.cmi.
2627    
2628    ----------------------------------------------------------------------
2629    Name: Allen Leung
2630    Date: 2002/03/13 17:30:00 EST
2631    Tag: leunga-20020313-x86-fp-unary
2632    Description:
2633    
2634    Bug fix for:
2635    
2636    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
2637    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
2638    > - fun f(x,(y,z)) = Real.~ y;
2639    > [autoloading]
2640    > [autoloading done]
2641    >       fchsl   (%eax), 184(%esp)
2642    > Error: MLRisc bug: X86MCEmitter.emitInstr
2643    >
2644    > uncaught exception Error
2645    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
2646    
2647    The problem was that the code generator did not generate any fp registers
2648    in this case, and the ra didn't know that it needed to run the X86FP phase to
2649    translate the pseudo fp instruction.   This only happened with unary fp
2650    operators in certain situations.
2651    
2652    ----------------------------------------------------------------------
2653    Name: Matthias Blume
2654    Date: 2002/03/13 14:00:00 EST
2655    Tag: blume-20020313-overload-etc
2656    Description:
2657    
2658    1. Added _overload as a synonym for overload for backward compatibility.
2659       (Control.overloadKW must be true for either version to be accepted.)
2660    
2661    2. Fixed bug in install script that caused more things to be installed
2662       than what was requested in config/targets.
2663    
2664    3. Made CM aware of the (_)overload construct so that autoloading
2665       works.
2666    
2667    ----------------------------------------------------------------------
2668    Name: Matthias Blume
2669    Date: 2002/03/12 22:03:00 EST
2670    Tag: blume-20020312-url
2671    Description:
2672    
2673    Forgot to update BOOT and srcarchiveurl.
2674    
2675    ----------------------------------------------------------------------
2676    Name: Matthias Blume
2677    Date: 2002/03/12 17:30:00 EST
2678    Tag: blume-20020312-version110392
2679    Description:
2680    
2681    Yet another version number bump (because of small changes to the
2682    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
2683    
2684    Changes:
2685    
2686      The new pid generation scheme described a few weeks ago was overly
2687      complicated.  I implemented a new mechanism that is simpler and
2688      provides a bit more "stability":  Once CM has seen a compilation
2689      unit, it keeps its identity constant (as long as you do not delete
2690      those crucial CM/GUID/* files).  This means that when you change
2691      an interface, compile, then go back to the old interface, and
2692      compile again, you arrive at the original pid.
2693    
2694      There now also is a mechanism that instructs CM to use the plain
2695      environment hash as a module's pid (effectively making its GUID
2696      the empty string).  For this, "noguid" must be specified as an
2697      option to the .sml file in question within its .cm file.
2698      This is most useful for code that is being generated by tools such
2699      as ml-nlffigen (because during development programmers tend to
2700      erase the tool's entire output directory tree including CM's cached
2701      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
2702      revert to the old, broken behavior of SML/NJ, but in specific cases
2703      where there is no danger of interface confusion, its use is ok
2704      (I think).
2705    
2706      ml-nlffigen by default generates "noguid" annotations.  They can be
2707      turned off by specifying -guid in its command line.
2708    
2709    ----------------------------------------------------------------------
2710    Name: Lal George
2711    Date: 2002/03/12 12 14:42:36 EST
2712    Tag: george-20020312-frequency-computation
2713    Description:
2714    
2715    Integrated jump chaining and static block frequency into the
2716    compiler. More details and numbers later.
2717    
2718    ----------------------------------------------------------------------
2719    Name: Lal George
2720    Date: 2002/03/11 11 22:38:53 EST
2721    Tag: george-20020311-jump-chain-elim
2722    Description:
2723    
2724    Tested the jump chain elimination on all architectures (except the
2725    hppa).  This is on by default right now and is profitable for the
2726    alpha and x86, however, it may not be profitable for the sparc and ppc
2727    when compiling the compiler.
2728    
2729    The gc test will typically jump to a label at the end of the cluster,
2730    where there is another jump to an external cluster containing the actual
2731    code to invoke gc. This is to allow factoring of common gc invocation
2732    sequences. That is to say, we generate:
2733    
2734            f:
2735               testgc
2736               ja   L1      % jump if above to L1
2737    
2738            L1:
2739               jmp L2
2740    
2741    
2742    After jump chain elimination the 'ja L1' instructions is converted to
2743    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
2744    up being implemented in their long form (if L2 is far away) using:
2745    
2746            jbe     L3      % jump if below or equal to L3
2747            jmp     L2
2748         L3:
2749            ...
2750    
2751    
2752    For large compilation units L2  may be far away.
2753    
2754    
2755    ----------------------------------------------------------------------
2756    Name: Matthias Blume
2757    Date: 2002/03/11 13:30:00 EST
2758    Tag: blume-20020311-mltreeeval
2759    Description:
2760    
2761    A functor parameter was missing.
2762    
2763    ----------------------------------------------------------------------
2764    Name: Allen Leung
2765    Date: 2002/03/11 10:30:00 EST
2766    Tag: leunga-20020311-runtime-string0
2767    Description:
2768    
2769       The representation of the empty string now points to a
2770    legal null terminated C string instead of unit.  It is now possible
2771    to convert an ML string into C string with InlineT.CharVector.getData.
2772    This compiles into one single machine instruction.
2773    
2774    ----------------------------------------------------------------------
2775    Name: Allen Leung
2776    Date: 2002/03/10 23:55:00 EST
2777    Tag: leunga-20020310-x86-call
2778    Description:
2779    
2780       Added machine generation for CALL instruction (relative displacement mode)
2781    
2782    ----------------------------------------------------------------------
2783    Name: Matthias Blume
2784    Date: 2002/03/08 16:05:00
2785    Tag: blume-20020308-entrypoints
2786    Description:
2787    
2788    Version number bumped to 110.39.1.  NEW BOOTFILES!
2789    
2790    Entrypoints: non-zero offset into a code object where execution should begin.
2791    
2792    - Added the notion of an entrypoint to CodeObj.
2793    - Added reading/writing of entrypoint info to Binfile.
2794    - Made runtime system bootloader aware of entrypoints.
2795    - Use the address of the label of the first function given to mlriscGen
2796      as the entrypoint.  This address is currently always 0, but it will
2797      not be 0 once we turn on block placement.
2798    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
2799      with entry points) from mlriscGen.
2800    
2801    ----------------------------------------------------------------------
2802    Name: Allen Leung
2803    Date: 2002/03/07 20:45:00 EST
2804    Tag: leunga-20020307-x86-cmov
2805    Description:
2806    
2807       Bug fixes for CMOVcc on x86.
2808    
2809       1. Added machine code generation for CMOVcc
2810       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
2811       3. CMOVcc cannot have an immediate operand as argument.
2812    
2813    ----------------------------------------------------------------------
2814    Name: Matthias Blume
2815    Date: 2002/03/07 16:15:00 EST
2816    Tag: blume-20020307-controls
2817    Description:
2818    
2819    This is a very large but mostly boring patch which makes (almost)
2820    every tuneable compiler knob (i.e., pretty much everything under
2821    Control.* plus a few other things) configurable via both the command
2822    line and environment variables in the style CM did its configuration
2823    until now.
2824    
2825    Try starting sml with '-h' (or, if you are brave, '-H')
2826    
2827    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
2828    implements the underlying generic mechanism.
2829    
2830    The interface to some of the existing such facilities has changed somewhat.
2831    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
2832    (The getFoo interface is still there for backward-compatibility, but its
2833    use is deprecated.)
2834    
2835    The ml-build script passes -Cxxx=yyy command-line arguments through so
2836    that one can now twiddle the compiler settings when using this "batch"
2837    compiler.
2838    
2839    TODO items:
2840    
2841    We should go through and throw out all controls that are no longer
2842    connected to anything.  Moreover, we should go through and provide
2843    meaningful (and correct!) documentation strings for those controls
2844    that still are connected.
2845    
2846    Currently, multiple calls to Controls.new are accepted (only the first
2847    has any effect).  Eventually we should make sure that every control
2848    is being made (via Controls.new) exactly once.  Future access can then
2849    be done using Controls.acc.
2850    
2851    Finally, it would probably be a good idea to use the getter-setter
2852    interface to controls rather than ref cells.  For the time being, both
2853    styles are provided by the Controls module, but getter-setter pairs are
2854    better if thread-safety is of any concern because they can be wrapped.
2855    
2856    *****************************************
2857    
2858    One bug fix: The function blockPlacement in three of the MLRISC
2859    backpatch files used to be hard-wired to one of two possibilities at
2860    link time (according to the value of the placementFlag).  But (I
2861    think) it should rather sense the flag every time.
2862    
2863    *****************************************
2864    
2865    Other assorted changes (by other people who did not supply a HISTORY entry):
2866    
2867    1. the cross-module inliner now works much better (Monnier)
2868    2. representation of weights, frequencies, and probabilities in MLRISC
2869       changed in preparation of using those for weighted block placement
2870       (Reppy, George)
2871    
2872    ----------------------------------------------------------------------
2873    Name: Lal George
2874    Date: 2002/03/07 14:44:24 EST 2002
2875    Tag: george-20020307-weighted-block-placement
2876    
2877    Tested the weighted block placement optimization on all architectures
2878    (except the hppa) using AMPL to generate the block and edge frequencies.
2879    Changes were required in the machine properties to correctly
2880    categorize trap instructions. There is an MLRISC flag
2881    "weighted-block-placement" that can be used to enable weighted block
2882    placement, but this will be ineffective without block/edge
2883    frequencies (coming soon).
2884    
2885    
2886    ----------------------------------------------------------------------
2887    Name: Lal George
2888    Date: 2002/03/05 17:24:48 EST
2889    Tag: george-20020305-linkage-cluster
2890    
2891    In order to support the block placement optimization, a new cluster
2892    is generated as the very first cluster (called the linkage cluster).
2893    It contains a single jump to the 'real' entry point for the compilation
2894    unit. Block placement has no effect on the linkage cluster itself, but
2895    all the other clusters  have full freedom in the manner in which they
2896    reorder blocks or functions.
2897    
2898    On the x86 the typical linkage code that is generated is:
2899       ----------------------
2900            .align 2
2901       L0:
2902            addl    $L1-L0, 72(%esp)
2903            jmp     L1
2904    
2905    
2906            .align  2
2907       L1:
2908       ----------------------
2909    
2910    72(%esp) is the memory location for the stdlink register. This
2911    must contain the address of the CPS function being called. In the
2912    above example, it contains the address of  L0; before
2913    calling L1 (the real entry point for the compilation unit), it
2914    must contain the address for L1, and hence
2915    
2916            addl $L1-L0, 72(%esp)
2917    
2918    I have tested this on all architectures except the hppa.The increase
2919    in code size is of course negligible
2920    
2921    ----------------------------------------------------------------------
2922    Name: Allen Leung
2923    Date: 2002/03/03 13:20:00 EST
2924    Tag: leunga-20020303-mlrisc-tools
2925    
2926      Added #[ ... ] expressions to mlrisc tools
2927    
2928    ----------------------------------------------------------------------
2929    Name: Matthias Blume
2930    Date: 2002/02/27 12:29:00 EST
2931    Tag: blume-20020227-cdebug
2932    Description:
2933    
2934    - made types in structure C and C_Debug to be equal
2935    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
2936    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
2937    
2938    ----------------------------------------------------------------------
2939    Name: Matthias Blume
2940    Date: 2002/02/26 12:00:00 EST
2941    Tag: blume-20020226-ffi
2942    Description:
2943    
2944    1. Fixed a minor bug in CM's "noweb" tool:
2945       If numbering is turned off, then truly don't number (i.e., do not
2946       supply the -L option to noweb).  The previous behavior was to supply
2947       -L'' -- which caused noweb to use the "default" line numbering scheme.
2948       Thanks to Chris Richards for pointing this out (and supplying the fix).
2949    
2950    2. Once again, I reworked some aspects of the FFI:
2951    
2952       A. The incomplete/complete type business:
2953    
2954       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
2955         gone!
2956       - ML types representing an incomplete type are now *equal* to
2957         ML types representing their corresponding complete types (just like
2958         in C).  This is still safe because ml-nlffigen will not generate
2959         RTTI for incomplete types, nor will it generate functions that
2960         require access to such RTTI.   But when ML code generated from both
2961         incomplete and complete versions of the C type meet, the ML types
2962         are trivially interoperable.
2963    
2964         NOTE:  These changes restore the full generality of the translation
2965         (which was previously lost when I eliminated functorization)!
2966    
2967       B. Enum types:
2968    
2969       - Structure C now has a type constructor "enum" that is similar to
2970         how the "su" constructor works.  However, "enum" is not a phantom
2971         type because each "T enum" has values (and is isomorphic to
2972         MLRep.Signed.int).
2973       - There are generic access operations for enum objects (using
2974         MLRep.Signed.int).
2975       - ml-nlffigen will generate a structure E_foo for each "enum foo".
2976         * The structure contains the definition of type "mlrep" (the ML-side
2977         representation type of the enum).  Normally, mlrep is the same
2978         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
2979         then mlrep will be defined as a datatype -- thus facilitating
2980         pattern matching on mlrep values.
2981         ("-ec" will be suppressed if there are duplicate values in an
2982          enumeration.)
2983         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
2984         will be generated for each C enum constant xxx.
2985         * Conversion functions m2i and i2m convert between mlrep and
2986         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
2987         * Coversion functions c and ml convert between mlrep and "tag enum".
2988         * Access functions (get/set) fetch and store mlrep values.
2989       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
2990         enumerations are merged into one single enumeration represented by
2991         structure E_'.
2992    
2993    ----------------------------------------------------------------------
2994    Name: Allen Leung
2995    Date: 2002/02/25 04:45:00 EST
2996    Tag: leunga-20020225-cps-spill
2997    
2998    This is a new implementation of the CPS spill phase.
2999    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
3000    In case of problems, replace it with the old file spill.sml
3001    
3002    The current compiler runs into some serious performance problems when
3003    constructing a large record.  This can happen when we try to compile a
3004    structure with many items.  Even a very simple structure like the following
3005    makes the compiler slow down.
3006    
3007        structure Foo = struct
3008           val x_1 = 0w1 : Word32.int
3009           val x_2 = 0w2 : Word32.int
3010           val x_3 = 0w3 : Word32.int
3011           ...
3012           val x_N = 0wN : Word32.int
3013        end
3014    
3015    The following table shows the compile time, from N=1000 to N=4000,
3016    with the old compiler:
3017    
3018    N
3019    1000   CPS 100 spill                           0.04u  0.00s  0.00g
3020           MLRISC ra                               0.06u  0.00s  0.05g
3021              (spills = 0 reloads = 0)
3022           TOTAL                                   0.63u  0.07s  0.21g
3023    
3024    1100   CPS 100 spill                           8.25u  0.32s  0.64g
3025           MLRISC ra                               5.68u  0.59s  3.93g
3026              (spills = 0 reloads = 0)
3027           TOTAL                                   14.71u  0.99s  4.81g
3028    
3029    1500   CPS 100 spill                           58.55u  2.34s  1.74g
3030           MLRISC ra                               5.54u  0.65s  3.91g
3031              (spills = 543 reloads = 1082)
3032           TOTAL                                   65.40u  3.13s  6.00g
3033    
3034    2000   CPS 100 spill                           126.69u  4.84s  3.08g
3035           MLRISC ra                               0.80u  0.10s  0.55g
3036              (spills = 42 reloads = 84)
3037           TOTAL                                   129.42u  5.10s  4.13g
3038    
3039    3000   CPS 100 spill                           675.59u  19.03s  11.64g
3040           MLRISC ra                               2.69u  0.27s  1.38g
3041              (spills = 62 reloads = 124)
3042           TOTAL                                   682.48u  19.61s  13.99g
3043    
3044    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
3045           MLRISC ra                               4.96u  0.27s  2.72g
3046              (spills = 85 reloads = 170)
3047           TOTAL                                   2375.26u  57.21s  48.00g
3048    
3049    As you can see the old cps spill module suffers from some serious
3050    performance problem.  But since I cannot decipher the old code fully,
3051    instead of patching the problems up, I'm reimplementing it
3052    with a different algorithm.  The new code is more modular,
3053    smaller when compiled, and substantially faster
3054    (O(n log n) time and O(n) space).  Timing of the new spill module:
3055    
3056    4000  CPS 100 spill                           0.02u  0.00s  0.00g
3057          MLRISC ra                               0.25u  0.02s  0.15g
3058             (spills=1 reloads=3)
3059          TOTAL                                   7.74u  0.34s  1.62g
3060    
3061    Implementation details:
3062    
3063    As far as I can tell, the purpose of the CPS spill module is to make sure the
3064    number of live variables at any program point (the bandwidth)
3065    does not exceed a certain limit, which is determined by the
3066    size of the spill area.
3067    
3068    When the bandwidth is too large, we decrease the register pressure by
3069    packing live variables into spill records.  How we achieve this is
3070    completely different than what we did in the old code.
3071    
3072    First, there is something about the MLRiscGen code generator
3073    that we should be aware of:
3074    
3075    o MLRiscGen performs code motion!
3076    
3077       In particular, it will move floating point computations and
3078       address computations involving only the heap pointer to
3079       their use sites (if there is only a single use).
3080       What this means is that if we have a CPS record construction
3081       statement
3082    
3083           RECORD(k,vl,w,e)
3084    
3085       we should never count the new record address w as live if w
3086       has only one use (which is often the case).
3087    
3088       We should do something similar to floating point, but the transformation
3089       there is much more complex, so I won't deal with that.
3090    
3091    Secondly, there are now two new cps primops at our disposal:
3092    
3093     1. rawrecord of record_kind option
3094        This pure operator allocates some uninitialized storage from the heap.
3095        There are two forms:
3096    
3097         rawrecord NONE [INT n]  allocates a tagless record of length n
3098         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
3099                                     and initializes the tag.
3100    
3101     2. rawupdate of cty
3102          rawupdate cty (v,i,x)
3103          Assigns to x to the ith component of record v.
3104          The storelist is not updated.
3105    
3106    We use these new primops for both spilling and increment record construction.
3107    
3108     1. Spilling.
3109    
3110        This is implemented with a linear scan algorithm (but generalized
3111        to trees).  The algorithm will create a single spill record at the
3112        beginning of the cps function and use rawupdate to spill to it,
3113        and SELECT or SELp to reload from it.  So both spills and reloads
3114        are fine-grain operations.  In contrast, in the old algorithm
3115        "spills" have to be bundled together in records.
3116    
3117        Ideally, we should sink the spill record construction to where
3118        it is needed.  We can even split the spill record into multiple ones
3119        at the places where they are needed.  But CPS is not a good
3120        representation for global code motion, so I'll keep it simple and
3121        am not attempting this.
3122    
3123     2. Incremental record construction (aka record splitting).
3124    
3125        Long records with many component values which are simulatenously live
3126        (recall that single use record addresses are not considered to
3127         be live) are constructed with rawrecord and rawupdate.
3128        We allocate space on the heap with rawrecord first, then gradually
3129        fill it in with rawupdate.  This is the technique suggested to me
3130        by Matthias.
3131    
3132        Some restrictions on when this is applicable:
3133        1. It is not a VECTOR record.  The code generator currently does not handle
3134           this case. VECTOR record uses double indirection like arrays.
3135        2. All the record component values are defined in the same "basic block"
3136           as the record constructor.  This is to prevent speculative
3137           record construction.
3138    
3139    ----------------------------------------------------------------------
3140    Name: Allen Leung
3141    Date: 2002/02/22 01:02:00 EST
3142    Tag: leunga-20020222-mlrisc-tools
3143    
3144    Minor bug fixes in the parser and rewriter
3145    
3146    ----------------------------------------------------------------------
3147    Name: Allen Leung
3148    Date: 2002/02/21 20:20:00 EST
3149    Tag: leunga-20020221-peephole
3150    
3151    Regenerated the peephole files.  Some contained typos in the specification
3152    and some didn't compile because of pretty printing bugs in the old version
3153    of 'nowhere'.
3154    
3155    ----------------------------------------------------------------------
3156    Name: Allen Leung
3157    Date: 2002/02/19 20:20:00 EST
3158    Tag: leunga-20020219-mlrisc-tools
3159    Description:
3160    
3161       Minor bug fixes to the mlrisc-tools library:
3162    
3163       1.  Fixed up parsing colon suffixed keywords
3164       2.  Added the ability to shut the error messages up
3165       3.  Reimplemented the pretty printer and fixed up/improved
3166           the pretty printing of handle and -> types.
3167       4.  Fixed up generation of literal symbols in the nowhere tool.
3168       5.  Added some SML keywords to to sml.sty
3169    
3170    ----------------------------------------------------------------------
3171    Name: Matthias Blume
3172    Date: 2002/02/19 16:20:00 EST
3173    Tag: blume-20020219-cmffi
3174    Description:
3175    
3176    A wild mix of changes, some minor, some major:
3177    
3178    * All C FFI-related libraries are now anchored under $c:
3179        $/c.cm      --> $c/c.cm
3180        $/c-int.cm  --> $c/internals/c-int.cm
3181        $/memory.cm --> $c/memory/memory.cm
3182    
3183    * "make" tool (in CM) now treats its argument pathname slightly
3184      differently:
3185        1. If the native expansion is an absolute name, then before invoking
3186           the "make" command on it, CM will apply OS.Path.mkRelative
3187           (with relativeTo = OS.FileSys.getDir()) to it.
3188        2. The argument will be passed through to subsequent phases of CM
3189           processing without "going native".  In particular, if the argument
3190           was an anchored path, then "make" will not lose track of that anchor.
3191    
3192    * Compiler backends now "know" their respective C calling conventions
3193      instead of having to be told about it by ml-nlffigen.  This relieves
3194      ml-nlffigen from one of its burdens.
3195    
3196    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
3197    
3198    * Export C_DEBUG and C_Debug from $c/c.cm.
3199    
3200    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
3201      subtyping relationship between incomplete pointers and their complete
3202      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
3203      with the convention of instantiating 'o with ('t, 'c) obj whenever
3204      the pointer target type is complete.  In the incomplete case, 'o
3205      will be instantiated with some "'c iobj" -- a type obtained by
3206      using one of the functors PointerToIncompleteType or PointerToCompleteType.
3207    
3208      Operations that work on both incomplete and complete pointer types are
3209      typed as taking an 'o ptr while operations that require the target to
3210      be known are typed as taking some ('t, 'c) obj ptr.
3211    
3212      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
3213      where void is an eqtype without any values.  This makes it possible
3214      to work on voidptr values using functions meant to operate on light
3215      incomplete pointers.
3216    
3217    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
3218      been vastly simplified.
3219    
3220    ----------------------------------------------------------------------
3221    Name: Matthias Blume
3222    Date: 2002/02/19 10:48:00 EST
3223    Tag: blume-20020219-pqfix
3224    Description:
3225    
3226    Applied Chris Okasaki's bug fix for priority queues.
3227    
3228    ----------------------------------------------------------------------
3229    Name: Matthias Blume
3230    Date: 2002/02/15 17:05:00
3231    Tag: Release_110_39
3232    Description:
3233    
3234    Last-minute retagging is becoming a tradition... :-(
3235    
3236    This is the working release 110.39.
3237    
3238    ----------------------------------------------------------------------
3239    Name: Matthias Blume
3240    Date: 2002/02/15 16:00:00 EST
3241    Tag: Release_110_39-orig
3242    Description:
3243    
3244    Working release 110.39.  New bootfiles.
3245    
3246    (Update: There was a small bug in the installer so it wouldn't work
3247    with all shells.  So I retagged. -Matthias)
3248    
3249    ----------------------------------------------------------------------
3250    Name: Matthias Blume
3251    Date: 2002/02/15 14:17:00 EST
3252    Tag: blume-20020215-showbindings
3253    Description:
3254    
3255    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
3256    the latter can be useful for exploring what bindings are available at
3257    the interactive prompt.  (The first function returns only the list
3258    of symbols that are really bound, the second prints those but also the
3259    ones that CM's autoloading mechanism knows about.)
3260    
3261    ----------------------------------------------------------------------
3262    Name: Matthias Blume
3263    Date: 2002/02/15 12:08:00 EST
3264    Tag: blume-20020215-iptrs
3265    Description:
3266    
3267    Two improvements to ml-nlffigen:
3268    
3269      1. Write files only if they do not exist or if their current contents
3270         do not coincide with what's being written.  (That is, avoid messing
3271         with the time stamps unless absolutely necessary.)
3272    
3273      2. Implement a "repository" mechanism for generated files related
3274         to "incomplete pointer types".   See the README file for details.
3275    
3276    ----------------------------------------------------------------------
3277    Name: Matthias Blume
3278    Date: 2002/02/14 11:50:00 EST
3279    Tag: blume-20020214-quote
3280    Description:
3281    
3282    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
3283    because of the new and improved tag generation scheme.  (Thanks to Allen
3284    Leung for pointing it out.)
3285    
3286    ----------------------------------------------------------------------
3287    Name: Lal George
3288    Date: 2002/02/14 09:55:27 EST 2002
3289    Tag: george-20020214-isabelle-bug
3290    Description:
3291    
3292    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
3293    of Isabelle on the x86.
3294    
3295    From Allen:
3296    -----------
3297     I've found the problem:
3298    
3299         in ra-core.sml, I use the counter "blocked" to keep track of the
3300         true number of elements in the freeze queue.  When the counter goes
3301         to zero, I skip examining the queue.  But I've messed up the
3302         bookkeeping in combine():
3303    
3304             else ();
3305             case !ucol of
3306               PSEUDO => (if !cntv > 0 then
3307                     (if !cntu > 0 then blocked := !blocked - 1 else ();
3308                                        ^^^^^^^^^^^^^^^^^^^^^^^
3309                      moveu := mergeMoveList(!movev, !moveu)
3310                     )
3311                  else ();
3312    
3313         combine() is called to coalesce two nodes u and v.
3314         I think I was thinking that if the move counts of u and v are both
3315         greater than zero then after they are coalesced then one node is
3316         removed from the freeze queue.  Apparently I was thinking that
3317         both u and v are of low degree, but that's clearly not necessarily true.
3318    
3319    
3320    02/12/2002:
3321        Here's the patch.  HOL now compiles.
3322    
3323        I don't know how this impact on performance (compile
3324        time or runtime).  This bug caused the RA (especially on the x86)
3325        to go thru the potential spill phase when there are still nodes on the
3326        freeze queue.
3327    
3328    
3329    
3330    
3331    ----------------------------------------------------------------------
3332    Name: Matthias Blume
3333    Date: 2002/02/13 22:40:00 EST
3334    Tag: blume-20020213-fptr-rtti
3335    Description:
3336    
3337    Fixed a bug in ml-nlffigen that was introduced with one of the previous
3338    updates.
3339    
3340    ----------------------------------------------------------------------
3341    Name: Matthias Blume
3342    Date: 2002/02/13 16:41:00 EST
3343    Tag: blume-20020213-cmlpq
3344    Description:
3345    
3346    Added new priority queue export symbols (which have just been added to
3347    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
3348    not compile and the installer would choke.)
3349    
3350    ----------------------------------------------------------------------
3351    Name: Matthias Blume
3352    Date: 2002/02/13 16:15:00 EST
3353    Tag: blume-20020213-various
3354    Description:
3355    
3356    1. More tweaks to ml-nlffigen:
3357    
3358       - better internal datastructures (resulting in slight speedup)
3359       - "-match" option requires exact match
3360       - "localized" gensym counters (untagged structs/unions nested within
3361         other structs/unions or within typedefs get a fresh counter; their
3362         tag will be prefixed by a concatenation of their parents' tags)
3363       - bug fixes (related to calculation of transitive closure of types
3364         to be included in the output)
3365    
3366    2. Minor Basis updates:
3367    
3368       - added implementations for List.collate and Option.app
3369    
3370    ----------------------------------------------------------------------
3371    Name: Matthias Blume
3372    Date: 2002/02/11 15:55:00 EST
3373    Tag: blume-20020211-gensym
3374    Description:
3375    
3376    Added a "-gensym" option to command line of ml-nlffigen.  This can be
3377    used to specify a "stem" -- a string that is inserted in all "gensym'd"
3378    names (ML structure names that correspond to unnamed C structs, unions,
3379    and enums), so that separate runs of ml-nlffigen do not clash.
3380    
3381    ----------------------------------------------------------------------
3382    Name: Matthias Blume
3383    Date: 2002/02/11 12:05:00 EST
3384    Tag: blume-20020211-gensml
3385    Description:
3386    
3387    A quick fix for a problem with GenSML (in the pgraph-util library):
3388    Make generation of toplevel "local" optional.  (Strictly speaking,
3389    signature definitions within "local" are not legal SML.)
3390    
3391    Other than that: updates to INSTALL and cm/TODO.
3392    
3393    ----------------------------------------------------------------------
3394    Name: Matthias Blume
3395    Date: 2002/02/08 15:00:00 EST
3396    Tag: blume-20020208-uniquepid
3397    Description:
3398    
3399    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
3400    
3401    1. The installer (config/install.sh) has gotten smarter:
3402    
3403         - Configuration options are a bit easier to specify now
3404           (in config/targets).
3405         - Bug in recognizing .tar.bz2 files fixed.
3406         - Installer automatically resolves dependencies between
3407           configuration options (e.g., if you ask for eXene, you will
3408           also get cml -- regardless whether you asked for it or not).
3409         - Installer can run in "quieter mode" by setting the environment
3410           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
3411           "completely silent", though.
3412         - Build HashCons library as part of smlnj-lib.
3413    
3414    2. A new scheme for assigning persistent identifiers to compilation
3415       units (and, by extension, to types etc.) has been put into place.
3416       This fixes a long-standing bug where types and even dynamic values
3417       can get internally confused, thereby compromising type safety
3418       (abstraction) and dynamic correctness.  See
3419    
3420         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
3421    
3422       for an example of how things could go wrong until now.
3423    
3424       The downside of the new scheme is that pids are not quite as
3425       persistent as they used to be: CM will generate a fresh pid
3426       for every compilation unit that it thinks it sees for the first
3427       time.  That means that if you compile starting from a clean, fresh
3428       source tree at two different times, you end up with different
3429       binaries.
3430    
3431       Cutoff recompilation, however, has not been compromised because
3432       CM keeps pid information in special caches between runs.
3433    
3434    ----------------------------------------------------------------------
3435    Name: Lal George
3436    Date: 2002/02/07 15:34:13 EST 2002
3437    Tag: <none>
3438    Description:
3439    
3440    Compilers that generate assembly code may produce  global labels
3441    whose value is resolved at link time. The various peephole optimization
3442    modules did not take this in account.
3443    
3444    TODO. The Labels.addrOf function should really return an option
3445    type so that clients are forced to deal with this issue, rather
3446    than an exception being raised.
3447    
3448    ----------------------------------------------------------------------
3449    Name: Lal George
3450    Date: 2002/02/06 13:55:02 EST
3451    Tag: george-20020206-ra-breakup
3452    Description:
3453    
3454    1. A bug fix from Allen.
3455    
3456        A typo causes extra fstp %st(0)'s to be generated at compensation
3457        edges, which might cause stack underflow traps at runtime.  This
3458        occurs in fft where there are extraneous fstps right before the 'into'
3459        trap instruction (in this case they are harmless since none of the
3460        integers overflow.)
3461    
3462    2. Pulled out various utility modules that were embedded in the modules
3463       of the register allocator. I need these modules for other purposes, but
3464       they are not complete enough to put into a library (just yet).
3465    ----------------------------------------------------------------------
3466    Name: Matthias Blume
3467    Date: 2002/01/31 16:05:00 EST
3468    Tag: blume-20020131-sparc-ccalls
3469    Description:
3470    
3471    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
3472       of extra stack space by mistake.  Fixed.
3473    
3474    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
3475    
3476    ----------------------------------------------------------------------
3477    Name: Allen Leung
3478    Date: 2002/01/30
3479    Tag: leunga-20020130-nowhere-bug-fix
3480    Description:
3481    
3482       MLRISC bug fixes:
3483       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
3484       2. MachineInt.fromString was negating its value.
3485    
3486    ----------------------------------------------------------------------
3487    Name: Matthias Blume
3488    Date: 2002/01/29
3489    Tag: blume-20020129-INSTALL
3490    Description:
3491    
3492    - Added somewhat detailed installation instructions (file INSTALL).
3493    - Fixed curl-detection bug in config/install.sh.
3494    - It is now possible to select the URL getter using the URLGETTER
3495      environment variable:
3496    
3497          not set / "unknown"      --> automatic detection (script tries wget,
3498                                       curl, and lynx)
3499          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
3500                                       how to properly invoke them)
3501          other                    --> use $URLGETTER directly, it must take
3502                                       precisely two command-line arguments
3503                                       (source URL and destination file name)
3504    
3505    ----------------------------------------------------------------------
3506    Name: Matthias Blume
3507    Date: 2002/01/28
3508    Tag: blume-20020128-sparc-ccalls
3509    Description:
3510    
3511    - Fixed problem with calculation of "used" registers in sparc-c-calls.
3512    - Make use of the allocParam argument in sparc-c-calls.
3513    
3514    ----------------------------------------------------------------------
3515    Name: Matthias Blume
3516    Date: 2002/01/28
3517    Tag: blume-20020128-allocParam
3518    Description:
3519    
3520    John Reppy:  Changes c-calls API to accept client-callback for
3521    allocating extra stack space.
3522    me: Corresponding changes to mlriscGen (using a dummy argument that
3523        does not change the current behavior).
3524    
3525    ----------------------------------------------------------------------
3526    Name: Matthias Blume
3527    Date: 2002/01/28 12:00:00
3528    Tag: Release_110_38
3529    Description:
3530    
3531    This time for real!!!
3532    
3533    ----------------------------------------------------------------------
3534    Name: Matthias Blume
3535    Date: 2002/01/28 10:56:00 EST
3536    Tag: blume-20020128-retraction
3537    Description:
3538    
3539    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
3540       with blume-Release_110_38-retracted.)
3541    
3542    1. Fixed a problem with incorrect rounding modes in real64.sml.
3543       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
3544    
3545    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
3546       and enums fixed.  The naming of corresponding ML identifiers should
3547       now be consistent again.
3548    
3549    ----------------------------------------------------------------------
3550    Name: Allen Leung
3551    Date: 2002/01/27
3552    Tag: leunga-20020127-nowhere
3553    Description:
3554    
3555       Added a target called nowhere in the configuration scripts.
3556       Enabling this will build the MLRISC 'nowhere' tool (for translating
3557       programs with where-clauses into legal SML code) during installation.
3558    
3559    ----------------------------------------------------------------------
3560    Name: Matthias Blume
3561    Date: 2002/01/25 21:27:00 EST
3562    Tag: blume-Release_110_38-retracted
3563    Description:
3564    
3565    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
3566    
3567    README will be added later.
3568    
3569    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
3570    (Release_110_38) removed.  Reason: Last-minute bug fixes.
3571    
3572    ----------------------------------------------------------------------
3573    Name: Matthias Blume
3574    Date: 2002/01/25
3575    Tag: blume-20020125-ffi
3576    Description:
3577    
3578    A large number of tweaks and improvements to ml-nlffi-lib and
3579    ml-nlffigen:
3580    
3581       - ML represenation types have been streamlined
3582       - getter and setter functions work with concrete values, not abstract
3583         ones where possible
3584       - ml-nlffigen command line more flexible (see README file there)
3585       - some bugs have been fixed (hopefully)
3586    
3587    ----------------------------------------------------------------------
3588    Name: Lal George
3589    Date: 2002/01/24
3590    Tag: george-20020124-risc-ra-interface
3591    Description:
3592    
3593       There is a dramatic simplification in the interface to the
3594       register allocator for RISC architectures as a result of making
3595       parallel copy instructions explicit.
3596    
3597    ----------------------------------------------------------------------
3598    Name: Matthias Blume
3599    Date: 2002/01/22
3600    Tag: blume-20020122-x86-ccalls
3601    Description:
3602    
3603    Bug fix for c-calls on x86 (having to do with how char- and
3604    short-arguments are being handled).
3605    
3606    ----------------------------------------------------------------------
3607    Name: Matthias Blume
3608    Date: 2002/01/21
3609    Tag: blume-20020121-ff
3610    Description:
3611    
3612    Another day of fiddling with the FFI...
3613    
3614    1. Bug fix/workaround:  CKIT does not complain about negative array
3615       dimensions, so ml-nlffigen has to guard itself against this possibility.
3616       (Otherwise a negative dimension would send it into an infinite loop.)
3617    
3618    2. Some of the abstract types (light objects, light pointers, most "base"
3619       types) in structure C are now eqtypes.
3620    
3621    3. Added constructors and test functions for NULL function pointers.
3622    
3623    ----------------------------------------------------------------------
3624    Name: Matthias Blume
3625    Date: 2002/01/18
3626    Tag: blume-20020118-ready-for-new-release
3627    Description:
3628    
3629    Made config/srcarchiveurl point to a new place.  (Will provide boot
3630    files shortly.)
3631    
3632    Maybe we christen this to be 110.38?
3633    
3634    ----------------------------------------------------------------------
3635    Name: Matthias Blume
3636    Date: 2002/01/18
3637    Tag: blume-20020118-more-ffifiddle
3638    Description:
3639    
3640    Today's FFI fiddling:
3641    
3642      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
3643        C.Set.* that use concrete (MLRep.*) arguments and results instead
3644        of abstract ones.
3645    
3646      - Provided word-style bit operations etc. for "int" representation
3647        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
3648        and Long.
3649    
3650    ----------------------------------------------------------------------
3651    Name: Matthias Blume
3652    Date: 2002/01/18
3653    Tag: blume-20020118-use-x86-fp
3654    Description:
3655    
3656    Now that x86-fast-fp seems to be working, I turned it back on again
3657    by default.  (Seems to work fine now, even with the FFI.)
3658    
3659    Other than that, I added some documentation about the FFI to
3660    src/ml-nlffigen/README and updated the FFI test examples in
3661    src/ml-nlffi-lib/Tests/*.
3662    
3663    ----------------------------------------------------------------------
3664    Name: Allen Leung
3665    Date: 2002/01/17
3666    Tag: leunga-20020117-x86-fast-fp-call
3667    Description:
3668    
3669       1. Fixed a problem with handling return fp values when x86's fast fp
3670          mode is turned on.
3671    
3672       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
3673          of %f32.
3674    
3675       3. Added a constructor INT32lit to the ast of MLRISC tools.
3676    
3677    ----------------------------------------------------------------------
3678    Name: Matthias Blume
3679    Date: 2002/01/16
3680    Tag: blume-20020116-ffifiddle
3681    Description:
3682    
3683    More fiddling with the FFI interface:
3684    
3685     - Make constness 'c instead of rw wherever possible.  This eliminates
3686       the need for certain explicit coercions.  (However, due to ML's
3687       value polymorphism, there will still be many cases where explicit
3688       coercions are necessary.  Phantom types are not the whole answer
3689       to modeling a subtyping relationship in ML.)
3690    
3691     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
3692    
3693     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
3694       clumsy workaround for varargs problem.)
3695    
3696    ----------------------------------------------------------------------
3697    Name: Lal George
3698    Date: 2002/01/15
3699    Tag: <none>
3700    Description:
3701    
3702    1. Since COPY instructions are no longer native to the architecture,
3703       a generic functor can be used to implement the expandCopies function.
3704    
3705    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
3706       TEXT segment.
3707    
3708    ----------------------------------------------------------------------
3709    Name: Matthias Blume
3710    Date: 2002/01/15
3711    Tag: blume-20020115-ffiupdates
3712    Description:
3713    
3714    1. Fix for bug resulting in single-precision float values being returned
3715       incorrectly from FFI calls.
3716    
3717    2. Small modifications to C FFI API:
3718    
3719        - memory-allocation routines return straight objects (no options)
3720          and raise an exception in out-of-memory situations
3721        - unsafe extensions to cast between function pointers and pointers
3722          from/to ints
3723        - added structure C_Debug as an alternative to structure C where
3724          pointer-dereferencing (|*| and |*!) always check for null-pointers
3725        - added open_lib' to DynLinkage;  open_lib' works like open_lib
3726          but also takes a (possibly empty) list of existing library handles
3727          that the current library depends on
3728    
3729    ----------------------------------------------------------------------
3730    Name: Matthias Blume
3731    Date: 2002/01/10
3732    Tag: blume-20020110-newffigen
3733    Description:
3734    
3735    1. Updates to portable graph code.
3736    
3737    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
3738       more scalable now so that even huge interfaces such as the one
3739       for GTK compile in finite time and space. :-)
3740       See src/ml-nlffigen/README for details on what's new.
3741    
3742    ----------------------------------------------------------------------
3743    Name: Lal George
3744    Date: 2001/01/09 14:31:35 EST 2002
3745    Tag: george-20011206-rm-native-copy
3746    Description:
3747    
3748            Removed the native COPY and FCOPY instructions
3749            from all the architectures and replaced it with the
3750            explicit COPY instruction from the previous commit.
3751    
3752            It is now possible to simplify many of the optimizations
3753            modules that manipulate copies. This has not been
3754            done in this change.
3755    
3756    ----------------------------------------------------------------------
3757    Name: Lal George
3758    Date: 2001/12/06 16:50:13 EST 2001
3759    Tag: george-20011206-mlrisc-instruction
3760    Description:
3761    
3762    Changed the representation of instructions from being fully abstract
3763    to being partially concrete. That is to say:
3764    
3765      from
3766            type instruction
3767    
3768      to
3769            type instr                              (* machine instruction *)
3770    
3771            datatype instruction =
3772                LIVE of {regs: C.cellset, spilled: C.cellset}
3773              | KILL of {regs: C.cellset, spilled: C.cellset}
3774              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
3775              | ANNOTATION of {i: instruction, a: Annotations.annotation}
3776              | INSTR of instr
3777    
3778    This makes the handling of certain special instructions that appear on
3779    all architectures easier and uniform.
3780    
3781    LIVE and KILL say that a list of registers are live or killed at the
3782    program point where they appear. No spill code is generated when an
3783    element of the 'regs' field is spilled, but the register is moved to
3784    the 'spilled' (which is present, more for debugging than anything else).
3785    
3786    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
3787    We used to generate:
3788    
3789            DEFFREG f1
3790            f1 := f2 + f3
3791            trapb
3792    
3793    but now generate:
3794    
3795            f1 := f2 + f3
3796            trapb
3797            LIVE {regs=[f1,f2,f3], spilled=[]}
3798    
3799    Furthermore, the DEFFREG (hack) required that all floating point instruction
3800    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
3801    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
3802    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
3803    this architecture).
3804    
3805    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
3806    all the architectures. This will result in further simplification of the
3807    register allocator that must be aware of them for coalescing purposes, and
3808    will also simplify certain aspects of the machine description that provides
3809    callbacks related to parallel copies.
3810    
3811    ANNOTATION should be obvious, and now INSTR represents the honest to God
3812    machine instruction set!
3813    
3814    The <arch>/instructions/<arch>Instr.sml files define certain utility
3815    functions for making porting easier -- essentially converting upper case
3816    to lower case. All machine instructions (of type instr) are in upper case,
3817    and the lower case form generates an MLRISC instruction. For example on
3818    the alpha we have:
3819    
3820      datatype instr =
3821         LDA of {r:cell, b:cell, d:operand}
3822       | ...
3823    
3824      val lda : {r:cell, b:cell, d:operand} -> instruction
3825        ...
3826    
3827    where lda is just (INSTR o LDA), etc.
3828    
3829    ----------------------------------------------------------------------
3830    Name: Matthias Blume
3831    Date: 2001/11/22 21:40:00 EST
3832    Tag: Release_110_37
3833    Description:
3834    
3835    Release 110.37.  This time for real.
3836    
3837    ----------------------------------------------------------------------
3838    Name: Matthias Blume
3839    Date: 2001/11/21 16:35:00 EST
3840    Tag: blume-20011121-foot-in-mouth
3841    Description:
3842    
3843    Removed the "Release_110_37" tag because of a serious bug.
3844    This will be re-tagged once the bug is fixed.
3845    
3846    ----------------------------------------------------------------------
3847    Name: Matthias Blume
3848    Date: 2001/11/21 16:14:00 EST
3849    Tag: blume-20011121-forgottenfile
3850    Description:
3851    
3852 &nbs