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 590, Sat Apr 1 02:24:08 2000 UTC revision 1744, Thu Dec 23 23:58:38 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: 2004/12/23 18:00:00 CST
18    Tag: blume-20041223-santa
19    Description:
20    
21    Made ml-build script "smarter" (but only very little).
22    
23    ----------------------------------------------------------------------
24    Name: Matthias Blume (blume (at) tti - c (dot) org)
25    Date: 2004/12/21 15:05:00 CST
26    Tag: blume-20041221-longlong
27    Description:
28    
29    * Implemented access to signed and unsigned long long data in NLFFI.
30    (The parameter-passing part of the picture has not complete. But
31    data structure access seems to work.)
32    
33    * Fixed CM's incorrect assumption that the PPC is little-endian.
34      (On the Mac, it is big-endian.  And that's currently our only
35      PPC platform.)
36    
37    ----------------------------------------------------------------------
38    Name: Matthias Blume (blume (at) tti - c (dot) org)
39    Date: 2004/12/21 12:50:00 CST
40    Tag: blume-20041221-memory
41    Description:
42    
43    Some cleanup in the $c/memory.cm library: separated some concerns by
44    moving allocation code and memory access code each into their own
45    files.
46    
47    ----------------------------------------------------------------------
48    Name: Allen Leung (leunga (at) reservoir (dot) com)
49    Date: 2004/12/17 16:12:00 EST
50    Tag: leunga-20041217-cygwin-smlnj-home
51    Description:
52    
53        The Unix I/O library of SML/NJ on cygwin does not understand
54    Windows style pathname, so problems arise when SMLNJ_HOME is set
55    to a Windows style pathname.  _run-sml now converts SMLNJ_HOME
56    to a POSIX pathname on cygwin.
57    
58    ----------------------------------------------------------------------
59    Name: Matthias Blume (blume (at) tti - c (dot) org)
60    Date: 2004/12/16 13:35:00 CST
61    Tag: Release_110_52
62    Description:
63    
64    Last-minute changes incorporated into 110.52.  Release tag moved.
65    
66    The changes:
67       - HashString.hashString' -> HashString.hashSubstring
68       - bug fix in UnivariateStats
69    
70    ----------------------------------------------------------------------
71    Name: Matthias Blume (blume (at) tti - c (dot) org)
72    Date: 2004/12/15 23:40:00 CST
73    Tag: blume-20041215-hashSubstring
74    Description:
75    
76    - HashString.hashString' -> HashString.hashSubstring
77    - corresponding changes in atom.sml
78    - "de-compressed" (aka. un-obfuscated) code for UnivariateStats and
79      added some comments
80    
81    ----------------------------------------------------------------------
82    Name: Matthias Blume (blume (at) tti - c (dot) org)
83    Date: 2004/12/15 15:30:00 CST
84    Tag: (Release_110_52)
85    Description:
86    
87    New working version (110.52).  NEW BOOTFILES!
88    
89    ----------------------------------------------------------------------
90    Name: Matthias Blume (blume (at) tti - c (dot) org)
91    Date: 2004/12/15 12:45:00 CST
92    Tag: blume-20041215-spaces
93    Description:
94    
95    More on the space problem (this time for Win32).
96    
97    ----------------------------------------------------------------------
98    Name: Matthias Blume (blume (at) tti - c (dot) org)
99    Date: 2004/12/14 17:30:00 CST
100    Tag: blume-20041214-spaces
101    Description:
102    
103    Hacked some of the scripts (in particular: the installer) to cope with
104    spaces in filenames a bit better.  But beware: the current "solution"
105    is likely still full of bugs and inherently incomplete.  (We need to
106    do away with those shell scripts for a comprehensive solution.)
107    
108    ----------------------------------------------------------------------
109    Name: Matthias Blume (blume (at) tti - c (dot) org)
110    Date: 2004/12/13 14:45:00 CST
111    Tag: blume-20041213-ml-makedepend
112    Description:
113    
114    Fixed bug in code for ml-makedepend.
115    
116    ----------------------------------------------------------------------
117    Name: Matthias Blume (blume (at) tti - c (dot) org)
118    Date: 2004/12/09 16:30:00 CST
119    Tag: blume-20041209-statistics
120    Description:
121    
122    Added two simple but potentially useful statistics modules to SML/NJ Library.
123    (See CHANGES file there.)
124    
125    ----------------------------------------------------------------------
126    Name: Matthias Blume (blume (at) tti - c (dot) org)
127    Date: 2004/12/01 16:50:00 CST
128    Tag: blume-20041201-atom
129    Description:
130    
131    smlnj-lib:
132    
133      Added function HashString.hashString' for substrings.
134      Hand-inlined CharVector.foldl into HashString (for speed).
135      Modified implementation of structure Atom to avoid extracting
136      strings from substrings unless necessary.
137      (Also see CHANGES file for smlnj-lib.)
138    
139    ----------------------------------------------------------------------
140    Name: Matthias Blume (blume (at) tti - c (dot) org)
141    Date: 2004/11/24 22:15:00 CST
142    Tag: blume-20041124-cml
143    Description:
144    
145    Made sure CML compiles when Position = Int64.
146    
147    ----------------------------------------------------------------------
148    Name: Matthias Blume (blume (at) tti - c (dot) org)
149    Date: 2004/11/24 14:30:00 CST
150    Tag: blume-20041124-position
151    Description:
152    
153    The compiler can now be compiled in a mode that makes structure Position
154    equal to Int64.  The default, however, is unchanged (Position = Int31)
155    for the time being.
156    
157    To enable 64-bit positions, use the following procedure:
158    
159    1. Start sml
160    2. Autoload $smlnj/cmb.cm (if not already autoloaded)
161    3. Type
162          #set (CMB.symval "USE_64_BIT_POSITIONS") (SOME 1);
163    4. Run CMB.make() as usual.
164    
165    This is barely tested.  The only test so far was a little SML program
166    counting the number of characters in an 8-gigabyte file by
167    reading it character-by-character.  That test was successful.
168    
169    In support of 64-bit positions, a number of new functions have been
170    added to the runtime system.
171    
172    ----------------------------------------------------------------------
173    Name: Matthias Blume (blume (at) tti - c (dot) org)
174    Date: 2004/11/23 14:45:00 CST
175    Tag: blume-20041123-useFile
176    Description:
177    
178    Fixed a problem with unhelpful error messages related to problems with
179    .cm- or .sml files that appear as part of the sml command line.
180    
181    ----------------------------------------------------------------------
182    Name: Matthias Blume (blume (at) tti - c (dot) org)
183    Date: 2004/11/18 15:40:00 CST
184    Tag: Release_110_51
185    Description:
186    
187    New working version (110.51).  NEW BOOTFILES!
188    
189    ----------------------------------------------------------------------
190    Name: Matthias Blume (blume (at) tti - c (dot) org)
191    Date: 2004/11/18 15:35:00 CST
192    Tag: <none>
193    Description:
194    
195    Enabled dlopen and friends for FreeBSD (as recommended by Johannes 5
196    Joemann).
197    
198    ----------------------------------------------------------------------
199    Name: Allen Leung (leunga (at) reservoir (dot) com)
200    Date: 2004/11/17 16:05:21 EST 2004
201    Tag: leunga-20041117-mlrisc-live-kill
202    Description:
203    
204       Added support for MLTree constructs LIVE and KILL
205    to all the architectures.
206    
207    ----------------------------------------------------------------------
208    Name: Matthias Blume (blume (at) tti - c (dot) org)
209    Date: 2004/11/13 00:20:00 CST
210    Tag: blume-20041113-versiontool
211    Description:
212    
213    - Stripped down the versiontool: It now only handles the version number.
214      The date string is generated at bootstrap time (during makeml).
215    
216    - In a previous commit, fixed a minor issue with how polyequal is being
217      translated.  In particular, the code now "looks through" abstractions.
218      This results in slightly fewer polyEqual warnings and hopefully slightly
219      more efficient code.  Important examples for where this matters are
220      the new int64 and word64 types.
221    
222    ----------------------------------------------------------------------
223    Name: Matthias Blume (blume (at) tti - c (dot) org)
224    Date: 2004/11/12 00:30:00 CST
225    Tag: blume-20041112-int64
226    Description:
227    
228    Structure Int64 fully hooked in.  (The implementation is not very
229    efficient, though.)
230    
231    ----------------------------------------------------------------------
232    Name: Matthias Blume (blume (at) tti - c (dot) org)
233    Date: 2004/11/11 17:30:00 CST
234    Tag: blume-20041111-more64
235    Description:
236    
237    All the pieces of Word64 are now there, with the exception of the
238    conversions from and to LargeWord. (Eventually these need to be identities,
239    but for the time being they don't even make sense because LargeWord is
240    32-bit wide.)
241    
242    Also started to add similar support for Int64, but major pieces of that
243    are still missing.
244    
245    ----------------------------------------------------------------------
246    Name: Matthias Blume (blume (at) tti - c (dot) org)
247    Date: 2004/11/11 00:15:00 CST
248    Tag: blume-20041111-word64
249    Description:
250    
251    Structure Word64 is now (almost) complete, word literals and patterns
252    seem to work.  There are a few odd pieces missing.  In particular,
253    I didn't do the {from,to}LargeWord parts because LargeWord is still Word32
254    at the moment.
255    
256    Making Word64 official would mean that LargeWord becomes Word64.  But
257    this requires extreme care because most word-word conversions have to
258    go through LargeWord, so making a mistake means loss of efficiency or
259    worse.  Eventually there will be a solution similar to (but actually
260    simpler than) what I did with IntInf.
261    
262    ----------------------------------------------------------------------
263    Name: Matthias Blume (blume (at) tti - c (dot) org)
264    Date: 2004/11/10 18:12:00 CST
265    Tag: blume-20041110-64bit
266    Description:
267    
268    More 64-bit hacking (but still not even half-way there yet).
269    Also, some assorted improvements to the handling of 8-bit words.
270    
271    ----------------------------------------------------------------------
272    Name: Matthias Blume (blume (at) tti - c (dot) org)
273    Date: 2004/11/09 17:50:00 CST
274    Tag: <none>
275    Description:
276    
277    Started adding some infrastructure for supporting 64-bit int- and
278    word-types.  (Still in its very early stages.)
279    
280    ----------------------------------------------------------------------
281    Name: Matthias Blume (blume (at) tti - c (dot) org)
282    Date: 2004/10/28 10:45:00 CDT
283    Tag: Release_110_50
284    Description:
285    
286    New working version (110.50).  NEW BOOTFILES!
287    
288                 =====================
289    
290    Also:
291    
292      - Changend config/srcarchiveurl from a file just
293        containing the URL string into a file containing
294        shell script code.  The code has access to the $VERSION variable.
295      - Made corresponding changes to config/install.sh and config/unpack.
296      - Default contents of config/srcarchiveurl uses $VERSION and
297        normally does not have to be edited to reflect a version change.
298    
299      (As a result, a version change can be done by just editing
300       config/version, the rest is now automatic.)
301    
302    ----------------------------------------------------------------------
303    Name: Matthias Blume (blume (at) tti - c (dot) org)
304    Date: 2004/10/27 17:50:00 CDT
305    Tag: blume-20041027-btrace-msg
306    Description:
307    
308    BackTrace.monitor now also reports the source of the exception that
309    triggered the trace.
310    
311    ----------------------------------------------------------------------
312    Name: Matthias Blume (blume (at) tti - c (dot) org)
313    Date: 2004/10/27 17:20:00 CDT
314    Tag: blume-20041027-x86-c-calls
315    Description:
316    
317    This is the HISTORY entry for two earlier commits, both concerning
318    the x86 c-calls code in MLRISC:
319    
320      - added a missing LOAD in the code that deals with struct arguments
321      - made sure the caller does not add the wrong number of bytes to the
322        stack pointer after a call of a function returning a struct
323        (the callee already pops the implicit argument which points to
324         the space reserved for the result)
325    
326    ----------------------------------------------------------------------
327    Name: Allen Leung (leunga (at) reservoir (dot) com)
328    Date: 2004/10/24 14:00:00 EST
329    Tag: leunga-20041024-x86-gas-fucomip
330    Description:
331    
332       John discovered a bug in the syntax of fucomip.
333       The opcodes FU?COMIP? have been changed to
334    
335           fu?comip? %st(i), %st
336    
337    ----------------------------------------------------------------------
338    Name: Matthias Blume (blume (at) tti - c (dot) org)
339    Date: 2004/10/20 15:06:00 CDT
340    Tag: blume-20041020-standalone-backtrace
341    Description:
342    
343    Added a mechanism for getting back-trace information from standalone
344    programs.  Here is how it works:
345    
346       1. The part of the program from which you want to get backtrace
347          information (usually the whole program) should be wrapped with
348          BackTrace.monitor.  This is a (unit->'a)->'a function, and your
349          main program could be modified from something like
350    
351             fun main (pgm, args) = ...
352    
353         to
354    
355            fun main (pgm, args) = BackTrace.monitor (fn () => ...)
356    
357       2. To be able to access BackTrace.monitor, you have to add
358          library $smlnj-tdp/plugins.cm to the .cm file that contains your
359          main function.
360    
361       3. Remove all compiled code (i.e., all the .cm/ subdirectories that
362          CM might have created in the past for your project).
363    
364       4. Build the system using this command line:
365    
366             ml-build -Ctdp.instrument=true \$smlnj-tdp/back-trace.cm \
367                      myprog.cm MyProg.main myprog
368    
369          instead of the usual
370    
371             ml-build myprog.cm MyProg.main myprog
372    
373    I changed a library name:
374    
375       $/trace-debug-profile.cm --> $smlnj-tdp/plugins.cm
376    
377    New libraries:
378    
379       $smlnj-tdp/back-trace.cm
380          -- when loaded causes the back-trace plugin to be installed
381       $smlnj-tdp/coverage.cm
382          -- when loaded causes the coverage plugin to be installed
383    
384    ----------------------------------------------------------------------
385    Name: Matthias Blume (blume (at) tti - c (dot) org)
386    Date: 2004/10/18 16:45:00 CDT
387    Tag: blume-20041018-groupowner
388    Description:
389    
390    Added an "obsolete" warning for the "group owner" syntax to CM's parser.
391    
392    Eliminated group owner specs from .cm files throughout the source tree.
393    
394    ----------------------------------------------------------------------
395    Name: Matthias Blume (blume (at) tti - c (dot) org)
396    Date: 2004/10/15 15:45:00 CDT
397    Tag: blume-20041015-coverage
398    Description:
399    
400    * Test coverage tool added!
401    * Further reorganization of tracing-, debugging-, and profiling support:
402    
403        - moved original BTImp -- now called BackTrace -- into a separate
404          library called $/trace-debug-profile.cm
405        - eliminated all mentions of BTrace from SMLofNJ.Internals
406        - only the instrumentation mechanism is now left in the compiler proper
407        - BackTrace module is a plugin which is NOT plugged in by default
408        - Coverage module is another such plugin
409    
410    To get the benefits of any of these plugin modules, the code in
411    question must be compiled with tdp instrumentation turned on.  This
412    can be done by setting SMLofNJ.Internals.TDP.mode to true.
413    (The ref cell is also controlled via the -Ctdp.instrument=... switch.)
414    
415    Plugins are selected at link time.  (Pre-compiled instrumented code
416    can be re-loaded with different plugins in effect.)  When an
417    instrumented module is linked, whatever plugins are at that time
418    enabled will come into effect for that module.
419    
420    To enable the back-trace plugin, load library $/trace-debug-profile.cm
421    and invoke BackTrace.install() (e.g., from the interactive prompt).
422    To enable the coverage plugin, load the same library and invoke
423    Coverage.install().
424    
425    Back-traces are generated automatically on uncaught exceptions and
426    when the code in question explicitly invokes BackTrace.trigger().
427    
428    Coverage (and execution frequency-) information must be queried
429    explicitly by calling Coverage.not_covered and Coverage.hot_spots.
430    
431    ----------------------------------------------------------------------
432    Name: Matthias Blume (blume (at) tti - c (dot) org)
433    Date: 2004/10/14 17:40:00 CDT
434    Tag: blume-20041014-tdp-core
435    Description:
436    
437    Snapshot of a significant overhaul of how the trace/debug/profile support
438    is hooked into the system (specifically: Core and SMLofNJ.Internals).
439    
440    ----------------------------------------------------------------------
441    Name: Matthias Blume (blume (at) tti - c (dot) org)
442    Date: 2004/10/13 16:34:00 CDT
443    Tag: blume-20041013-tdp
444    Description:
445    
446    Some rationalization of names:
447    
448       structure BTrace -> structure TDPInstrument
449       etc.
450    
451    This is is preparation of using the original back-trace
452    instrumentation for other purposes.  "TDP" stands for
453    Trace/Debug/Profile.
454    
455    The control flag controlling whether instrumentation is on or off is now
456    registered under a different name, so instead of running sml as
457    
458     sml -Cinstrument.btrace-mode=true
459    
460    one has to say
461    
462     sml -Ctdp.instrument=true
463    
464    ----------------------------------------------------------------------
465    Name: Matthias Blume (blume (at) tti - c (dot) org)
466    Date: 2004/10/11 16:37:00 CDT
467    Tag: blume-20041011-regions
468    Description:
469    
470    Made some minor modifications to elabcore.sml to have source regions
471    be propagated more tightly -- resulting in better (i.e., smaller)
472    regions being reported in error- and debug messages.
473    
474    ----------------------------------------------------------------------
475    Name: Matthias Blume (blume (at) tti - c (dot) org)
476    Date: 2004/10/08 22:50:00 CDT
477    Tag: blume-20041008-cmkw
478    Description:
479    
480    Fixed handling of keywords in .cm files: After seeing "is" the lexer
481    treats subsequent occurrences of "group", "library", "source", "is",
482    "*", and "-" as ordinary identifiers rather than keywords.
483    
484    Most seriously, this fixes a problem with CM's "shell" tool.  The tool
485    is supposed to accept a tool argument called "source", but this did
486    not work because of the clash with the keyword.
487    
488    ----------------------------------------------------------------------
489    Name: Matthias Blume (blume (at) tti - c (dot) org)
490    Date: 2004/10/07 16:00:00 CDT
491    Tag: blume-20041007-cleanup
492    Description:
493    
494    Assorted cleanup work:
495    
496      - got rid of intstrmap in favor of using the library's
497        hash table implementation
498      - threw out most of the pathnames stuff, as it was not used anyway
499      - simplified tokentable implementation
500      - fixed some minor spelling errors
501    
502    ----------------------------------------------------------------------
503    Name: Matthias Blume (blume (at) tti - c (dot) org)
504    Date: 2004/10/06 15:15:15 CDT
505    Tag: blume-20041006-handler
506    Description:
507    
508    Cleaned up the absyn to reflect the invariant that HANDLE always
509    carries a FNexp as part of the type definition.  This eliminates some
510    superfluous sanity checks at runtime down the road.
511    
512    Some minor cleanup of the btrace code.
513    
514    ----------------------------------------------------------------------
515    Name: Matthias Blume (blume (at) tti - c (dot) org)
516    Date: 2004/10/01 10:20:30 CDT
517    Tag: blume-20041001-slave
518    Description:
519    
520    Added hack to make slave mode work in the presence of the version
521    tool.  (Still, since the master does two passes over the code for
522    CMB.make, the release number gets bumped twice when slaves are
523    attached.  I don't know if this is worth fixing...)
524    
525    ----------------------------------------------------------------------
526    Name: Matthias Blume (blume (at) tti - c (dot) org)
527    Date: 2004/09/30 10:55:00 CDT
528    Tag: blume-20040930-version
529    Description:
530    
531    * Moved the "version" magic into its own little library under
532      src/system/smlnj/internal.  This avoids expensive reconstruction of
533      a stable src/compiler/core.cm.
534    
535    * At the same time, structure CompilerVersion is now known as
536      structure SMLNJVersion.
537    
538    * Arranged for the version tool to NOT kick in when rebuilding the system
539      (makeml -rebuild, fixpt).  Otherwise one would never reach a fixpoint.
540      Also, loading the versiontool does not work when rebuilding the system
541      because CM is not properly initialized at that time.
542    
543    ----------------------------------------------------------------------
544    Name: Matthias Blume (blume (at) tti - c (dot) org)
545    Date: 2004/09/29 14:00:00 CDT
546    Tag: blume-20040929-autoversion
547    Description:
548    
549    Implemented some CM magic to have
550        file src/compiler/TopLevel/main/version.sml
551    generated automagically.
552    The version is taken from two files: config/version and config/release.
553    The first is expected to contain a two-part version number such as 110.49.
554    The second should contain a single number, but it may be missing.
555    
556    If the environment variable VERSIONTOOL_BUMP_RELEASE is defined at the
557    time the version tool is loaded (which is the first time you say CMB.make),
558    then the tool will increment the value stored in config/release every
559    time CMB.make is invoked.
560    
561    The binfile format is now insensitive to anything beyond the first
562    two components of a version number, so bumping the release does not render
563    binfiles incompatible.  Auto-bumping can be used to keep track of versions
564    during development without invalidating existing binfiles.
565    
566    In any case, every CMB.make updates the date information in version.sml.
567    (This is the date that is printed in the banner.)
568    
569    ----------------------------------------------------------------------
570    Name: Matthias Blume (blume (at) tti - c (dot) org)
571    Date: 2004/09/28 10:53:00 CDT
572    Tag: blume-20040928-controls
573    Description:
574    
575    Some cleanup of the controls code.
576    
577    ----------------------------------------------------------------------
578    Name: Matthias Blume (blume (at) tti - c (dot) org)
579    Date: 2004/09/27 22:08:00 CDT
580    Tag: blume-20040927-controls
581    Description:
582    
583    Added two pieces of functionality to the Controls interface:
584    
585    1.   val save'restore: 'a control -> unit -> unit
586    
587       grabs the current value of the control in stage 1 and restores it
588       in stage 2.
589    
590    2.   val set' : 'a control * 'a -> unit -> unit
591    
592       stores the given value into the control in stage 2 (i.e., delayed)
593       but does all error checking in stage 1.
594       (This is for string controls that need to do parse their argument --
595        something that might fail.  In some cases, notably in CM, one
596        already knows the intended argument but wants to delay the actual
597        assignment until a time when error recovery would be more difficult.)
598    
599    Changed the handling of controls in tool arguments to classes "sml" and
600    "lazysml":
601       - use Controls.save'restore as a more robust way of restoring the
602         old value (in particular: without having to re-parse the string)
603       - use controls to handle the "overload" keyword in the init group
604         (I believe this change actually fixes a long-standing obscure bug.)
605    
606    ----------------------------------------------------------------------
607    Name: Matthias Blume (blume (at) tti - c (dot) org)
608    Date: 2004/09/27 17:00:00 CDT
609    Tag: blume-20040927-lazysml
610    Description:
611    
612    Added a new tool class called "lazysml" to CM's tool chest.  The only
613    difference to "sml" is that compilation is done with Control.lazysml
614    set to true.  A source of class "lazysml" is automatically recognized
615    by a file name suffix of ".lml".
616    
617    In addition to the above feature, the original class "sml" now also
618    supports a tool argument "lazy" which has the same effect.  As a
619    result, the following three lines are equivalent:
620    
621        foo.sml : lazysml
622        foo.sml : sml (lazy)
623        foo.sml (lazy)
624    
625    The setting goes into effect both during parsing and during
626    compilation.  The original setting is restored right after parsing and
627    after compilation, respectively.
628    
629    In addition to all the above, there is also a general mechanism to set
630    ANY of the "controls" that are available at the command line via
631    "-C..." on a per-sml-file basis.  The same rules that apply for "lazy"
632    apply as well. (In fact, "lazy" is implemented as a special case of
633    the general mechanism.)
634    
635    The .cm file syntax uses a new keyword tool argument called "with".
636    There are several ways of indicating the desired settings:
637    
638        foo.sml (with:parser.quotations=true)
639        foo.sml (with:(name:parser.quotations value:true))
640        foo.sml (with:(name:name1 value:value1 name:name2 value:value2 ...))
641        foo.sml (with:(name1=value1 name2=value2 ...))
642        foo.sml (with:(name1=value1 name:name2 value:value2 name3=value3 ...))
643    
644        etc.
645    
646    Another possible abbreviation is to leave out the =v or value:v part
647    if the name refers to a boolean control (in which case the value is
648    taken to be true).  Thus, one could get lazy sml also by saying:
649    
650        foo.sml (with:parser.lazy-keyword=true)
651        foo.sml (with:parser.lazy-keyword)
652        foo.sml (with:(name:parser.lazy-keyword value:true))
653        foo.sml (with:(name:parser.lazy-keyword))
654    
655    ----------------------------------------------------------------------
656    Name: Matthias Blume (blume (at) tti - c (dot) org)
657    Date: 2004/09/24 16:22:00 CDT
658    Tag: blume-20040924-ppc-long-branch
659    Description:
660    
661    Turned message about "emiting long form of branch" off by default.
662    Added a control flag to turn it back on when desired.
663    
664    ----------------------------------------------------------------------
665    Name: Matthias Blume (blume (at) tti - c (dot) org)
666    Date: 2004/09/24 16:05:00 CDT
667    Tag: blume-20040924-rounding
668    Description:
669    
670    Applied patch for setting rounding modes under Mac OS X.  Thanks to
671    Melissa O'Neill for providing the code!
672    
673    ----------------------------------------------------------------------
674    Name: Matthias Blume (blume (at) tti - c (dot) org)
675    Date: 2004/09/23 17:30:00 CDT
676    Tag: blume-20040923-envvars
677    Description:
678    
679    1. Changed definition of type ControlRegistry.registry_tree to
680       include control_info (i.e., the name of the controlling environment
681       variable).
682    
683    2. Added command-line flags -e and -E to print the names of environment
684       variables that can be used to control internal settings.
685       (This uses the new API mentioned in 1.)
686    
687    ----------------------------------------------------------------------
688    Name: Matthias Blume (blume (at) tti - c (dot) org)
689    Date: 2004/09/13 16:50:00 CDT
690    Tag: Release_110_49
691    Description:
692    
693    New working version (110.49).  NEW BOOTFILES!
694    
695    ----------------------------------------------------------------------
696    Name: Matthias Blume
697    Date: 2004/09/13 16:20:00 CDT
698    Tag: blume-20040913-config-mlrisc
699    Description:
700    
701    Put target "mlrisc" back into the default list.
702    (There is no harm in having it, and some users have expressed their
703    wish to have "mlrisc" included by default.)
704    
705    ----------------------------------------------------------------------
706    Name: John Reppy
707    Date: 2004/09/13
708    Tag: jhr-20040913-signals
709    Description:
710    
711    Fixed the signal masking code to properly nest mask/unmask operations
712    on a per-signal basis.
713    
714    ----------------------------------------------------------------------
715    Name: Matthias Blume (blume (at) tti - c (dot) org)
716    Date: 2004/09/08 13:20:00 CDT
717    Tag: blume-20040908-heap-magic
718    Description:
719    
720    Bumped the heap macig number to 0x09082004 to account for the changed
721    layout of the ML frame under MacOS X.
722    
723    ----------------------------------------------------------------------
724    Name: Allen Leung (leunga (at) reservoir (dot) com)
725    Date: 2004/09/03 11:26:00 EST
726    Tag: leunga-20040903-cygwin-install
727    Description:
728    
729       Added a patch to _arch-n-opsys to enable the Cygwin runtime.
730    The Cygwin runtime is turned on by setting the environment
731    variable SMLNJ_CYGWIN_RUNTIME to 1.
732    
733    ----------------------------------------------------------------------
734    Name: Matthias Blume (blume (at) tti - c (dot) org)
735    Date: 2004/08/31 17:15:00 CDT
736    Tag: blume-20040831-core
737    Description:
738    
739    Added some exports to src/compiler/core.cm upon request by J. Joemann.
740    
741    ----------------------------------------------------------------------
742    Name: Matthias Blume (blume (at) tti - c (dot) org)
743    Date: 2004/08/30 17:55:00 CDT
744    Tag: blume-20040830-installer
745    Description:
746    
747    Upon request by Johannes Joemann:
748    
749    - improved ML code of installer to fall back to coping when renaming
750      fails (i.e., when source and target are on different file systems);
751      the code compiles but has yet to be tested in anger
752    - removed mlrisc from list of default targets (config/targets)
753    
754    ----------------------------------------------------------------------
755    Name: Matthias Blume (blume (at) tti - c (dot) org)
756    Date: 2004/08/27 17:20:00 CDT
757    Tag: blume-20040827-ptreql
758    Description:
759    
760    Added ptreql primop to structure InlineT (upon request from Larry
761    Paulson).
762    
763    ----------------------------------------------------------------------
764    Name: Allen Leung (leunga (at) reservoir (dot) com)
765    Date: 2004/08/15 21:21:00 EST
766    Tag: leunga-110_48-udgraph
767    Description:
768    
769    Another bug fix from Carl Hauser:
770    
771    diff /net/niflab/smlnj48/src/MLRISC/graphs/udgraph.sml udgraph.sml
772    > 48c48
773    > <              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then es else
774    > e::L)
775    > ---
776    > >              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then L else e::L)
777    > Without this, any deletion of an edge in an undirected graph does severe
778    > violence to the graph.
779    
780    ----------------------------------------------------------------------
781    Name: Allen Leung (leunga (at) reservoir (dot) com)
782    Date: 2004/08/10 23:35:00 EST
783    Tag: leunga-110_48-ppc
784    Description:
785    
786       The IBM/MacOS syntax switch on PPC was incorrectly swapped.
787    
788    ----------------------------------------------------------------------
789    Name: Matthias Blume (blume (at) tti - c (dot) org)
790    Date: 2004/08/10 12:00:00 CDT
791    Tag: Release_110_48
792    Description:
793    
794    New working version (110.48).  NEW BOOTFILES!
795    
796    ----------------------------------------------------------------------
797    Name: Allen Leung (leunga (at) reservoir (dot) com)
798    Date: 2004/08/09 12:21:00 EST
799    Tag: leunga-110_47-dijsktra
800    Description:
801    
802    Bug fix from Carl Hauser:
803    
804    single_source_shortest_paths in dijkstra.sml was observed to get wrong
805    answers (by comparing to single_source_shortest_paths in bellman-ford.sml).
806    
807    The problem is that following the expression A.update(dist,s,Num.zero)
808    it is necessary to update the priority queue using Q.decreaseWeight(Q,s).
809    
810    ----------------------------------------------------------------------
811    Name: Matthias Blume (blume (at) tti - c (dot) org)
812    Date: 2004/08/06 18:10:00 CDT
813    Tag: blume-20040806-cmdline
814    Description:
815    
816    Fiddled with handling of command-line options:
817    
818      * sml now quits after processing the command line
819        if -H, -S, -h<n>, or -s<n> appears as the last
820        command-line argument
821      * a new option -q terminates the session when encountered on
822        the command line; subsequent arguments will be ignored
823      * bug fixes: short (erroneous) arguments are no longer ignored
824        completely
825    
826    ----------------------------------------------------------------------
827    Name: Allen Leung (leunga (at) reservoir (dot) com)
828    Date: 2004/08/04 18:17:00 EST
829    Tag: leunga-110_47-ppc-ibm-asm
830    Description:
831    
832      - Added minimal IBM assembly syntax support for PowerPC.
833    
834      - Cygwin: manually changed the file cygwin.def.  Some exported symbols have
835        been altered in the runtime.  We need an automatic way to keep the file
836        in sync.
837    
838    ----------------------------------------------------------------------
839    Name: Matthias Blume (blume (at) tti - c (dot) org)
840    Date: 2004/08/04 14:00:00 CDT
841    Tag: Release_110_47
842    Description:
843    
844    New working version (110.47).  NEW BOOTFILES!
845    
846    ----------------------------------------------------------------------
847    Name: Matthias Blume (blume (at) tti - c (dot) org)
848    Date: 2004/08/03 14:25:00 CDT
849    Tag: blume-20040803-callingconv
850    Description:
851    
852    Added low-level support for choosing C calling conventions by
853    twiddling the type of rawccall.  (See
854    src/compiler/Semant/types/cproto.sml for details.)
855    
856    ----------------------------------------------------------------------
857    Name: Matthias Blume (blume (at) tti - c (dot) org)
858    Date: 2004/08/02 15:55:00 CDT
859    Tag: blume-20040802-backout
860    Description:
861    
862    Backed out of change to win32-filesys.c.  The earlier patch to
863    get_file_time caused CM to produce files with the wrong time stamp.
864    
865    ----------------------------------------------------------------------
866    Name: Matthias Blume (blume (at) tti - c (dot) org)
867    Date: 2004/08/02 14:45:00 CDT
868    Tag: blume-20040802-nlffi-win32
869    Description:
870    
871    Added NLFFI support for Win32, adapted from a patch provided by David
872    Hansel.  This is currently completely untested.  Also, the issue
873    concerning stdcall vs. ccall is still unresolved.
874    
875    ----------------------------------------------------------------------
876    Name: Matthias Blume (blume (at) tti - c (dot) org)
877    Date: 2004/07/30 17:55:00 CDT
878    Tag: blume-20040730-various
879    Description:
880    
881    Gearing up towards 110.47...
882    
883    - various minor bugfixes to ml-nlffigen
884    - a beginning of a manual for nlffi
885    
886    - eliminated 'export name=value' in config/install.sh as this does
887      not work with certain versions of /bin/sh
888      (Thanks to David King at Motorola for catching this.)
889    
890    - several bugfixes provided or suggested by David Hansel at Reactive Systems:
891      - added a test for tm==NULL to gmtime.c and localtime.c
892      - applied patch for incorrect GetFileTime under win32
893      - toSeconds -> toMilliseconds in Win32/win32-process.sml
894    
895    ----------------------------------------------------------------------
896    Name: Matthias Blume (blume (at) tti - c (dot) org)
897    Date: 2004/07/21 18:20:00 CDT
898    Tag: blume-20040721-nlffigen
899    Description:
900    
901    - Fixed minor issue in ml-nlffigen: Now generate structure T_foo
902      for a typedef to an incomplete type, but leave out the "typ" member.
903      (This is just for consistency.)
904    - Started to produce what is supposed to become better (i.e., comprehensive)
905      documentation of what ml-nlffigen does and produces.
906    
907    ----------------------------------------------------------------------
908    Name: Matthias Blume (blume (at) tti - c (dot) org)
909    Date: 2004/07/14 16:25:00 CDT
910    Tag: blume-20040714-union
911    Description:
912    
913    Added C_UNION to c-calls/c-types.sml and updated the machinery
914    (ml-nlffigen, cproto.sml) that conveys C function interface
915    information to the code generator.
916    
917    However, the actual architecture-specific implementation of function
918    arguments and results that are C unions is still not implemented.
919    
920    ----------------------------------------------------------------------
921    Name: Allen Leung (leunga (at) reservoir (dot) com)
922    Date: 2004/07/14 14:38:00 EST
923    Tag: leunga-110_46_1-ppc-lwzu
924    Description:
925    
926        Added these instructions to the PowerPC architecture:
927        LBZU(X), LHZU(X), LWZU(X),
928        STWU(X), STFDU, STFSU
929    
930        etc...
931    
932        Note: I haven't added their instruction encoding into the description.
933    
934    ----------------------------------------------------------------------
935    Name: Allen Leung (leunga (at) reservoir (dot) com)
936    Date: 2004/07/13 15:04:00 EST
937    Tag: leunga-110_46_1-ppc-lwarx
938    Description:
939    
940       Added the two instructions LWARX and STWCX to the PowerPC
941    instruction set.
942    
943       A (untested) rewrite of loop-structure.sml.  The old version
944    is completely broken.
945    
946    ----------------------------------------------------------------------
947    Name: Matthias Blume (blume (at) tti - c (dot) org)
948    Date: 2004/07/13 13:50:00 CDT
949    Tag: blume-20040713-nlffi
950    Description:
951    
952    - use paramAlloc to report c-calls with too many arguments
953      (for PPC version where parameter area is pre-allocated)
954    - added ccall_maxargspace to machspec (to implement the above)
955    - made "make" commend in CM's "make" tool configurable
956    - added option (default: on) for passing the name of the SML/NJ's "bin"
957      directory to "make"; the call looks like this:
958    
959         make <options> SMLNJ_BINDIR=<dir> <target>
960    
961      This can be used by the Makefile to, e.g., pick the "right" version
962      of ml-nlffigen.
963    - minor code tweaks
964    
965    ----------------------------------------------------------------------
966    Name: Matthias Blume (blume (at) tti - c (dot) org)
967    Date: 2004/07/12 22:50:00 CDT
968    Tag: blume-110_46_1-macosx-nlffi
969    Description:
970    
971    NLFFI under Mac OS X now working (sort of).  This is largely untested,
972    though.
973    
974    Note:  1. You have to make a new, clean build of the runtime system.
975           2. There are new BOOTFILES, you have to use them!
976              (Doing the bootstrap process yourself would be *very* painful!
977               If you absolutely have to do it, build the system under
978               a different architecture and then cross-compile.)
979    
980    Version bumped to 110.46.1 to account for runtime data format changes.
981    
982    ----------------------------------------------------------------------
983    Name: Matthias Blume (blume (at) tti - c (dot) org)
984    Date: 2004/06/18 14:30:00 CDT
985    Tag: blume-20040618-unix
986    Description:
987    
988    Changed the implementation of structure Unix so that the same stream
989    is returned every time one of the {text,bin}{In,Out}streamOf functions
990    is invoked on the same proc.  This is not what the spec currently
991    says -- although IMO it arguably should.  (See discussion below.)
992    
993    ----------------------------------------------------------------------
994    Name: Matthias Blume (blume (at) tti - c (dot) org)
995    Date: 2004/06/17 18:15:00 CDT
996    Tag: Release_110_46
997    Description:
998    
999    New working version (110.46).  NEW BOOTFILES!
1000    
1001    ----------------------------------------------------------------------
1002    Name: Matthias Blume (blume (at) tti - c (dot) org)
1003    Date: 2004/06/17 17:20:00 CDT
1004    Tag: blume-20040617-timer-unix
1005    Description:
1006    
1007    Changed the interface of structures Timer and Unix to match the most
1008    recent Basis spec.
1009    
1010    In the case of Unix there still seems to be an open/weird issue:
1011    
1012       The {text,bin}{In,Out}streamOf functions are supposed to create
1013       fresh streams whenever they are called -- as opposed to have them
1014       return the same stream every time.  This design is supposed to
1015       prevent space leaks caused by proc values hanging on to streams.
1016    
1017       The reap function, on the other hand, is supposed to close the
1018       streams.  This cannot be done without having a handle on the
1019       stream in proc after all...
1020    
1021    I took the liberty to implement the following stopgap solution:
1022    
1023      The proc value hangs on to the most recently created stream(s).
1024      Reap closes those.  If either or both of the two streams hadn't
1025      been created at all yet, then reap will close the corresponding
1026      file descriptors directly.
1027    
1028    PS: I don't understand the original space leak argument anymore.  If
1029    a proc hangs on to the imperative stream, then I/O operations on those
1030    will advance the state of the cached stream and avoid the space leak.
1031    
1032    ----------------------------------------------------------------------
1033    Name: Matthias Blume (blume (at) tti - c (dot) org)
1034    Date: 2004/05/28 16:45:00 CDT
1035    Tag: blume-20040528-basis
1036    Description:
1037    
1038    Added signature PACK_REAL and exported functor PrimIO.
1039    
1040    ----------------------------------------------------------------------
1041    Name: Matthias Blume (blume (at) tti - c (dot) org)
1042    Date: 2004/05/25 16:00:00 CDT
1043    Tag: blume-20040525-group-owner
1044    Description:
1045    
1046    CM now ignores (but still accepts) the "owner" information in group
1047    descriptions.  The owner of a group is its next enclosing
1048    library. Each group must have a unique owner.  (There is a virtual
1049    "toplevel" library that owns groups which are not nested within a real
1050    library.)  Previously, each group had to explicitly declare its owner,
1051    and CM would check that such a declaration is correct.  The new scheme
1052    is to have CM check that for each group there is precisely one owning
1053    library.
1054    
1055    The advantage of the new scheme is that the programmer no longer needs
1056    to maintain the somewhat annoying owner information.  The downside is
1057    that CM cannot enforce the ownership rule across multiple runs of
1058    CM.make.  Fortunately, enclosing the same group in two different
1059    libraries A and B which are not part of the same program does not
1060    cause real problems.
1061    
1062    ----------------------------------------------------------------------
1063    Name: Matthias Blume (blume (at) tti - c (dot) org)
1064    Date: 2004/05/20 16:00:00 CDT
1065    Tag: blume-20040520-win32
1066    Description:
1067    
1068    Made the win32 version work again.  (Strangely, a misplaced comma had
1069    slipped into win32-process.c which prevented the runtime from being
1070    compiled correctly.)
1071    
1072    Also, included a minor addition to ml-build.bat analogous to what was
1073    done in blume-20040519-ml-build.
1074    
1075    ----------------------------------------------------------------------
1076    Name: Matthias Blume (blume (at) tti - c (dot) org)
1077    Date: 2004/05/19 22:10:00 CDT
1078    Tag: blume-20040519-ml-build
1079    Description:
1080    
1081    Arranged for ml-build to clean up after itself a little bit better.
1082    The script generates a temporary SML source file and compiles it using
1083    CM, so CM generates metadata (GUID, SKEL, objectfile) for it.  It now
1084    gets rid of those at the end, so they don't accumulate under .cm.
1085    
1086    This required a minor change to install.sh because the name of the
1087    metadata directory (default: .cm) is actually configurable at
1088    installation time.
1089    
1090    ----------------------------------------------------------------------
1091    Name: Matthias Blume (blume (at) tti - c (dot) org)
1092    Date: 2004/05/18 15:50:00 CDT
1093    Tag: blume-20040518-mkreader
1094    Description:
1095    
1096    Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
1097    implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
1098    
1099    ----------------------------------------------------------------------
1100    Name: Matthias Blume (blume (at) tti - c (dot) org)
1101    Date: 2004/05/11 14:35:00 CDT
1102    Tag: blume-20040511-win32sock
1103    Description:
1104    
1105    Added previously missing support for many socket-related functions
1106    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
1107    for the voluminous patch!
1108    
1109    (I have not tested this patch under win32 yet.)
1110    
1111    Here is David's e-mail:
1112    
1113    Hi,
1114    
1115    Attached to this email you find a diff against sml/nj 110.45
1116    that will enable socket support under Windows.
1117    
1118    To apply the patch (using unix or cygwin)
1119    1) gunzip runtime.diff.gz
1120    2) "cd" into "src/runtime" in the source tree of a fresh
1121       110.45 installation.
1122    3) patch -p 1 < [your/path/to]runtime.diff
1123    
1124    The code compiles fine but has NOT yet been extensively tested.
1125    I only ran a few tests for basic socket client functionality
1126    (which worked fine).  Especially the functions that use ioctl
1127    are not tested at all and might not work (see below).
1128    
1129    I implemented this since we want to move to a newer version of sml/nj
1130    but need socket support in order to use it.  This is the first time I
1131    even had a look at the sml/nj source,  so please review my changes
1132    before making this part of the distribution!  Here are a few issues
1133    that I think might be better for someone to solve who is more
1134    familiar with the sml/nj source (and socket programming):
1135    
1136    - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
1137      exception since I could not figure out what the windows equivalent
1138      of these functions is
1139    
1140    - In sockets-osdep.h there are a some #include statements that are
1141      only used in a few files that include sockets-osdep.h
1142    
1143    - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
1144      does not process its return value since I don't know how to
1145      report an error upwards.
1146    
1147    - It would probably be good to have a call to WSACleanup() when
1148      the library is unloaded (if there is such a possibility).
1149      Otherwise I think Windows will take care of this automatically
1150      when the process finishes.
1151    
1152    - I used ioctlsocket() as a replacement for ioctl() but I have
1153      no idea if that is actually the proper replacement on Windows.
1154    
1155    - All these issues are marked in the code by "FIXME" comments.
1156    
1157    We use sml/nj extensively in our products and are quite happy
1158    with it.  I hope this contribution will help you.
1159    
1160    Keep up the good work!
1161    
1162    David
1163    
1164    ----------------------------------------------------------------------
1165    Name: Matthias Blume (blume (at) tti - c (dot) org)
1166    Date: 2004/05/11 14:20:00 CDT
1167    Tag: blume-20040511-installml
1168    Description:
1169    
1170    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
1171    the patch.)
1172    
1173    ----------------------------------------------------------------------
1174    Name: Matthias Blume (blume (at) tti - c (dot) org)
1175    Date: 2004/05/11 14:05:00 CDT
1176    Tag: blume-20040511-nlffi-netbsd
1177    Description:
1178    
1179    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
1180    the patch.)
1181    
1182    ----------------------------------------------------------------------
1183    Name: Matthias Blume (blume (at) tti - c (dot) org)
1184    Date: 2004/05/11 12:05:00 CDT
1185    Tag: blume-20040511-exports
1186    Description:
1187    
1188    As per request by Adam Chlipala <adam@hcoop.net>, extended various
1189    export lists in compiler-related .cm-files.
1190    
1191    ----------------------------------------------------------------------
1192    Name: Matthias Blume (blume (at) tti - c (dot) org)
1193    Date: 2004/05/11 11:35:00 CDT
1194    Tag: blume-20040511-allsource
1195    Description:
1196    
1197    The installer now honors the "src-smlnj" target again, although its meaning
1198    has changed from "all sources required for the compiler" to "all sources
1199    the installer knows about".  In other words, if you enable "src-smlnj"
1200    in the "targets" file, then the installer will pull in sources for
1201    everything.  (Notice that this refers to source code only.  Compiled
1202    code is still only installed for modules that were requested explicitly
1203    or which are required for other modules that were requested explicitly.)
1204    
1205    ----------------------------------------------------------------------
1206    Name: Matthias Blume (blume (at) tti - c (dot) org)
1207    Date: 2004/04/23 17:40:00 CDT
1208    Tag: blume-20040423-ieee-scan
1209    Description:
1210    
1211    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
1212    in the exponent calculation we get INF or ZERO (depending on the mantissa
1213    and the sign of the exponent).
1214    
1215    ----------------------------------------------------------------------
1216    Name: Matthias Blume (blume (at) tti - c (dot) org)
1217    Date: 2004/04/23 10:40:00 CDT
1218    Tag: blume-20040423-ml-build
1219    Description:
1220    
1221    The ml-build script now terminates with a non-0 status when something
1222    goes wrong.
1223    
1224    ----------------------------------------------------------------------
1225    Name: Matthias Blume (blume (at) tti - c (dot) org)
1226    Date: 2004/04/22 16:35:00 CDT
1227    Tag: blume-20040422-Option
1228    Description:
1229    
1230    Made exception Option to be the same as exception Option.Option
1231    (as it should be).
1232    
1233    ----------------------------------------------------------------------
1234    Name: Allen Leung (leunga (at) reservoir (dot) com)
1235    Date: 2004/03/19 14:40:00 EST
1236    Tag: leunga-20040319-cygwin-nlffi
1237    Description:
1238    
1239    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
1240    of SML/NJ.  The problem is that
1241    
1242       lib = dlopen(NULL, ...)
1243       f   = dlsym(lib, "malloc");
1244    
1245    does not work on Windows unless we explicitly export symbols
1246    such as 'malloc' during linking.  We fixed this by explicitly
1247    exporting the required symbols with the magic gcc incantation:
1248    
1249        -Wl,--export-all cygwin.def
1250    
1251    where cygwin.def is a file containing all the symbols that
1252    we wish to export.
1253    
1254    I suspect this is a Windows problem and we'll have to
1255    do the same (somehow with windows compilers) when
1256    we build the native win32 version with the system
1257    calls LoadLibrary/GetProcAddress.
1258    
1259    ----------------------------------------------------------------------
1260    Name: Matthias Blume (blume (at) tti - c (dot) org)
1261    Date: 2004/03/04 16:35:00 CST
1262    Tag: blume-20040304-intinf-fmt
1263    Description:
1264    
1265    Fixed problem with IntInf.fmt (sign would show up on the right instead
1266    of on the left for BIN, OCT, and HEX).
1267    
1268    ----------------------------------------------------------------------
1269    Name: Matthias Blume (blume (at) tti - c (dot) org)
1270    Date: 2004/03/04 11:25:00 CST
1271    Tag: blume-20040304-symlinks
1272    Description:
1273    
1274    Fixed problem with installer script (unix only) where bin/ml-yacc and
1275    friends pointed (via symlinks) to absolute locations instead of just
1276    .run-sml.  This was reported by Vesa A Norrman.
1277    
1278    ----------------------------------------------------------------------
1279    Name: Matthias Blume (blume (at) tti - c (dot) org)
1280    Date: 2004/02/13 14:50:00 CST
1281    Tag: Release_110_45
1282    Description:
1283    
1284    New working version (110.45). New bootfiles.
1285    
1286    ----------------------------------------------------------------------
1287    Name: Matthias Blume (blume (at) tti - c (dot) org)
1288    Date: 2004/01/26 15:15:15 CST
1289    Tag: blume-20040126-toplevel
1290    Description:
1291    
1292    Improved handling of exceptions at the interactive toplevel.
1293    
1294    ----------------------------------------------------------------------
1295    Name: Matthias Blume (blume (at) tti - c (dot) org)
1296    Date: 2004/01/26 11:25:00
1297    Tag: blume-20040126-app
1298    Description:
1299    
1300    Type of top-level "app" corrected.
1301    Added code for setting vp_limitPtrMask to Win32-specific runtime.
1302    
1303    ----------------------------------------------------------------------
1304    Name: Matthias Blume (blume (at) tti - c (dot) org)
1305    Date: 2003/11/18 17:10 CST
1306    Tag: blume-20031118-basis-fiddle
1307    Description:
1308    
1309    - changed Timer interface to what might become the spec
1310    - POSIX_FLAGS -> BIT_FLAGS according to spec
1311    - some other minor discrepancies wrt. spec eliminated
1312    
1313    ----------------------------------------------------------------------
1314    Name: Matthias Blume (blume (at) tti - c (dot) org)
1315    Date: 2003/11/06 12:00:00 CST
1316    Tag: Release_110_44
1317    Description:
1318    
1319    New working version (110.44). New bootfiles.
1320    
1321    ----------------------------------------------------------------------
1322    Name: Matthias Blume (blume (at) tti - c (dot) org)
1323    Date: 2003/11/04 11:50:00 CST
1324    Tag: blume-20031104-move-libraries
1325    Description:
1326    
1327    Eliminated the "dont_move_libraries" directive in config/targets.
1328    (The mechanism was broken and could not be fixed easily.  Moreover,
1329    there does not seem to be any reason not to move all libraries into
1330    lib during installation.  I originally implemented this directive as a
1331    backward-compatibility feature when I first introduced the new CM.
1332    Now that things have been stable for a long time and going back to the
1333    old CM is not an option, there is no reason to keep it around.)
1334    
1335    ----------------------------------------------------------------------
1336    Name: Matthias Blume (blume (at) tti - c (dot) org)
1337    Date: 2003/11/03 16:00:00 CST
1338    Tag: blume-20031103-installdir
1339    Description:
1340    
1341    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
1342    Richards for pointing out the problem and providing the solution.)
1343    
1344    ----------------------------------------------------------------------
1345    Name: Matthias Blume (blume (at) tti - c (dot) org)
1346    Date: 2003/10/01 17:05:00 CDT
1347    Tag: blume-20031001-lal-mlrisc
1348    Description:
1349    
1350    MLRISC bug fix from Lal.
1351    
1352    ----------------------------------------------------------------------
1353    Name: Matthias Blume (blume (at) tti - c (dot) org)
1354    Date: 2003/09/30 16:10:00 CDT
1355    Tag: blume-20030930-primio-bat
1356    Description:
1357    
1358    1. Added openVector, nullRd, and nullWr to PRIM_IO.
1359    2. Improved .bat files (for Win32 port) to make things work under Win95.
1360       (thanks to Aaron S. Hawley for this one)
1361    
1362    ----------------------------------------------------------------------
1363    Name: Matthias Blume (blume (at) tti - c (dot) org)
1364    Date: 2003/09/26 16:05:00 CDT
1365    Tag: blume-20030926-wrappriv
1366    Description:
1367    
1368    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
1369    
1370    ----------------------------------------------------------------------
1371    Name: Matthias Blume (blume (at) tti - c (dot) org)
1372    Date: 2003/09/26 15:00:00 CDT
1373    Tag: blume-20030926-110_43_3
1374    Description:
1375    
1376    - additional cleanup
1377    - version number bump, NEW BOOTFILES
1378    
1379    ----------------------------------------------------------------------
1380    Name: Matthias Blume (blume (at) tti - c (dot) org)
1381    Date: 2003/09/26 12:00:00 CDT
1382    Tag: blume-20030926-ppautoload
1383    Description:
1384    
1385    I modified the read-eval-print loop so that the autoloader gets
1386    invoked whenever the prettyprinter tries to look up a symbol that
1387    is not currently defined in the toplevel environment but which
1388    appears in CM's autoload registry.  As a result, we see far fewer of
1389    those ?.Foo.Bar.xxx names in the prettyprinter's output.
1390    
1391    In addition to this I tried to clean up some pieces of the Basis
1392    implementation (e.g., Socket, Word8Array) in order to prevent other
1393    instances of these ?.Foo.Bar.xxx names from being printed.
1394    
1395    The mechanism that picks names for types still needs some work, though.
1396    (Right now it seems that if there is a type A.t which is defined to
1397    be B.u, but B is unavailable at toplevel, then A.t gets printed as
1398    "?.B.u" although the perhaps more sensible solution would be to use
1399    "A.t" in this case.  In other words, the prettyprinter should follow
1400    a chain of DEFtycs not farther than there are corresponding toplevel
1401    names in the current environment.)
1402    
1403    ----------------------------------------------------------------------
1404    Name: Matthias Blume (blume (at) tti - c (dot) org)
1405    Date: 2003/09/24 16:31:00 CDT
1406    Tag: blume-20030924-installer
1407    Description:
1408    
1409    Another installer tweak:  All the ML code for the installer is now
1410    compiled during CMB.make and put into a little library called
1411    $smlnj/installer.cm.  The installation then simply invokes
1412    
1413       sml -m $smlnj/installer.cm
1414    
1415    and everything happens automagically.
1416    
1417    Win32:  ML code senses value of environment variable SMLNJ_HOME.
1418    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
1419            and BINDIR.
1420    
1421    The new scheme guarantees that the ML code responsible for the installation
1422    is in sync with the APIs of the main system.  Also, the installer is
1423    somewhat faster because the installer script is precompiled.
1424    
1425    ----------------------------------------------------------------------
1426    Name: Matthias Blume (blume (at) tti - c (dot) org)
1427    Date: 2003/09/24 15:35:00 CDT
1428    Tag: blume-20030924-synsock
1429    Description:
1430    
1431    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
1432    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
1433    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
1434    implementations of non-blocking operations from CML's Socket
1435    structure.
1436    
1437    ----------------------------------------------------------------------
1438    Name: Matthias Blume (blume (at) tti - c (dot) org)
1439    Date: 2003/09/24 15:10:05 CDT
1440    Tag: blume-20030924-sockets
1441    Description:
1442    
1443    1. Fixed SOCKET API and implementation to match Basis spec.
1444       This required changing the internal representation of sockets to one
1445       that remembers (for each socket file descriptor) whether it is currently
1446       blocking or non-blocking.  This state is maintained lazily (i.e., a system
1447       call is made only if the state actually needs to change).
1448    
1449    2. OS-specific details of sockets were moved into separate files, thus
1450       making it possible to unify the bulk of the socket implementations
1451       between Unix and Win32.
1452    
1453    3. CML's socket API changed accordingly.
1454       (Note that we need to remove non-blocking functions from this API
1455       since they are redundant in the case of CML!)
1456    
1457    4. CML's socket implementation now makes use of non-blocking functions
1458       provided by Basis, thus removing all OS-dependent code from this part
1459       of CML.
1460    
1461    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
1462    
1463    ----------------------------------------------------------------------
1464    Name: Matthias Blume (blume (at) tti - c (dot) org)
1465    Date: 2003/09/22 12:10:00 CDT
1466    Tag: blume-20030922-110_43_2
1467    Description:
1468    
1469    Made a new interim version and bootfiles for developer's bootstrapping
1470    convenience.
1471    
1472       110.43.2 -- NEW BOOTFILES
1473    
1474    ----------------------------------------------------------------------
1475    Name: Matthias Blume (blume (at) tti - c (dot) org)
1476    Date: 2003/09/19 15:55:00 CDT
1477    Tag: blume-20030919-cmdir
1478    Description:
1479    
1480    1. new-install.sh -> install.sh
1481    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
1482    3. tweaked installer so that another name instead of .cm can be chosen
1483       at install time (by setting the CM_DIR_ARC environment variable
1484       during installation); once installation is complete, the name is
1485       fixed
1486    
1487    ----------------------------------------------------------------------
1488    Name: Matthias Blume (blume (at) tti - c (dot) org)
1489    Date: 2003/09/18 16:00:00 CDT
1490    Tag: blume-20030918-110_43_1
1491    Description:
1492    
1493    Made a new interim version and bootfiles for developer's bootstrapping
1494    convenience.
1495    
1496       110.43.1 -- NEW BOOTFILES
1497    
1498    ----------------------------------------------------------------------
1499    Name: Matthias Blume (blume (at) tti - c (dot) org)
1500    Date: 2003/09/18 15:20:00 CDT
1501    Tag: blume-20030918-misc
1502    Description:
1503    
1504    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
1505       this seems to be controversial at the moment) and used those in
1506       Posix.ProcEnv.times.
1507    
1508    2. Added Time.{from,to}Nanoseconds to Time.
1509    
1510    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
1511       For example, fromLargeInt never needs to look at more than 3 "big
1512       digits" to get its 53 bits of precision.
1513    
1514    ----------------------------------------------------------------------
1515    Name: Matthias Blume (blume (at) tti - c (dot) org)
1516    Date: 2003/09/17 16:30:00 CDT
1517    Tag: blume-20030917-real32-slices
1518    Description:
1519    
1520    Added an entry to the primitive environment
1521    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
1522    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
1523    
1524    Removed some of the "magic" constants in real64.sml and replaced them
1525    with code that generates these values from their corresponding
1526    integer counterparts.
1527    
1528    Made all(?) the slice-related changes to the Basis and made everything
1529    compile again...
1530    
1531    ----------------------------------------------------------------------
1532    Name: Matthias Blume (blume (at) tti - c (dot) org)
1533    Date: 2003/09/15 17:45:00 CDT
1534    Tag: blume-20030915-rbase
1535    Description:
1536    
1537    Fixed bug in Real.fromLargeInt.
1538    
1539    ----------------------------------------------------------------------
1540    Name: Matthias Blume (blume (at) tti - c (dot) org)
1541    Date: 2003/09/13 18:11:00 CDT
1542    Tag: blume-20030913-libinstall
1543    Description:
1544    
1545    Minor bugfix in config/libinstall (set anchor with path to
1546    standalone tool after installing it, otherwise libraries that
1547    need ml-lex or ml-yacc won't compile the first time the installer
1548    runs).
1549    
1550    ----------------------------------------------------------------------
1551    Name: Matthias Blume (blume (at) tti - c (dot) org)
1552    Date: 2003/09/12 11:45:00 CDT
1553    Tag: blume-20030912-various
1554    Description:
1555    
1556     - fixed bug in Real.toLargeInt
1557     - fixed bug in Posix.ProcEnv.times
1558     - changed inputLine functions to return an option
1559     - minor installer improvements / bugfixes
1560     - changed default @SMLalloc parameter for x86/celeron to 64k
1561    
1562    ----------------------------------------------------------------------
1563    Name: Matthias Blume (blume (at) tti - c (dot) org)
1564    Date: 2003/09/09 22:00:00 CDT
1565    Tag: Release_110_43
1566    Description:
1567    
1568        New working release 110.43.  New bootfiles.
1569    
1570    ----------------------------------------------------------------------
1571    Name: Matthias Blume (blume (at) tti - c (dot) org)
1572    Date: 2003/09/09 19:20:00 CDT
1573    Tag: blume-20030909-installer
1574    Description:
1575    
1576    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
1577    Modified config/install.bat to take advantage of it.  Also modified
1578    config/install.sh (and called it config/new-install.sh) to take advantage
1579    of it on Unix systems.  (The SML code is (supposed to be) platform-
1580    independent.)
1581    
1582    The installer can now install everything under Win32
1583    as well as under *nix as long as it compiles.
1584    
1585    Other changes:
1586    
1587      - made CML compile again under Win32
1588      - made eXene compile under Win32 (by providing a fake structure UnixSock
1589        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
1590      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
1591        same as type int;  under Win32 it isn't
1592      - fixed some slice-related problems in the win32-specific parts of CML
1593      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
1594        (under Win32, the volume name is case-insensitive, and the
1595        OS.Path code compares volume names for equality)
1596    
1597    ----------------------------------------------------------------------
1598    Name: Matthias Blume (blume (at) tti - c (dot) org)
1599    Date: 2003/09/08 11:55:00 CDT
1600    Tag: blume-20030908-fullpath
1601    Description:
1602    
1603    Made Win32 version of OS.FileSys.fullPath return current directory
1604    when given an empty string.  This is what the spec says, and incidentally,
1605    CM depends on it.  (CM otherwise goes into an infinite loop in certain
1606    cases when presented with the name of a non-existing .cm file.)
1607    
1608    ----------------------------------------------------------------------
1609    Name: Matthias Blume (blume (at) tti - c (dot) org)
1610    Date: 2003/09/04 16:30:00 CDT
1611    Tag: blume-20030905-slices-etc
1612    Description:
1613    
1614      1. Changed interface to vectors and arrays in Basis to match
1615         (draft) Basis spec.
1616      2. Added signatures and implementations of slices according to
1617         Basis spec.
1618      3. Edited source code throughout the system to make it compile again
1619         under 1. and 2.  (In some cases code had to be added to have it
1620         match the new signatures.)
1621      4. MLRISC should be backward-compatible: the copies of the originals
1622         of files that needed to change under 3. were retained, the .cm files
1623         check the compiler version number and use old versions when
1624         appropriate.
1625      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
1626         dirstream -> string option (in accordance with Basis spec).
1627      6. When generating code that counts lines, ml-lex used function
1628         CharVector.foldli, taking advantage of its old interface.
1629         This has been replaced with the corresponding code from
1630         CharVectorSlice.  (html-lex must be re-lexed!)
1631      7. BitArray in smlnj-lib/Util has been extended/modified to match the
1632         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
1633      8. Removed temporary additions (fromInternal, toInternal) from the
1634         (now obsolete) IntInf in smlnj-lib/Util.
1635      9. Cleaned up structure Byte.
1636     10. Added localOffset, scan, and fromString to Date (according to spec).
1637         Cleaned/corrected implementation of Date.
1638         (Still need to check for correctness; implement better canonicalizeDate.)
1639     11. Added "scan" to signature IEEE_REAL.
1640     12. Some improvements to IntInf [in particular: efficiency-hack for
1641         mod and rem when second operand is 2 (for parity checks).]
1642     13. Changed representation of type Time.time, using a single IntInf.int
1643         value counting microseconds.  This considerably simplified the
1644         implementation of structure Time.  We now support negative time
1645         values; scan and fromString handle signs.
1646     14. Functor PrimIO now takes two additional arguments (VectorSlice and
1647         ArraySlice).
1648    
1649    ----------------------------------------------------------------------
1650    Name: Matthias Blume (blume (at) tti - c (dot) org)
1651    Date: 2003/08/28 17:00:00 CDT
1652    Tag: blume-20030828-intinf
1653    Description:
1654    
1655    This is a major update which comes with a version number bump
1656    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
1657    and an implementation of IntInf in the Basis.
1658    
1659    There are a fairly large number of related changes and updates throughout
1660    the system:
1661    
1662    Basis:
1663            - Implemented IntInf.
1664            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
1665            - Added some missing Real64 operations, most notably Real.toLargeInt.
1666            - Added FixedInt as a synonym for Int32.
1667    
1668    compiler:
1669            * Added support for a built-in intinf type.
1670              - literals
1671              - pattern matching
1672              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
1673              - overloading on literals and operations
1674    
1675            This required adding a primitive type intinf, some additional
1676            primops, and implementations for several non-trivial intinf
1677            operations in Core.  (The intinf type is completely abstract
1678            to the compiler; all operations get delegated back to the Core.)
1679    
1680            * Intinf equality is handled by polyequal.  However, the compiler
1681            does not print its usual warning in this case (since polyequal
1682            is the right thing to do there).
1683    
1684            * Improved the organization of structure InlineT.
1685    
1686            * A word about conversion primops:
1687            If conversions involving intinf do not cancel out during
1688            CPS contract, then the compiler must insert calls to Core functions.
1689            Since all core access must be resolved already during the FLINT
1690            translate phase, it would be too late a the time of CPS contract
1691            to add new Core calls.  For this reason, conversion primops
1692            for intinf carry two arguments:  1. the numeric argument that
1693            they are supposed to convert, and 2. the Core function that
1694            can help with this conversion if necessary.  If CPS contract
1695            eliminates a primop, then the associated Core function becomes
1696            dead and goes away.  Intinf conversion primops that do not get
1697            eliminated by CPS contract get rewritten into calls of their
1698            core functions by a separate, new phase.
1699    
1700    interactive system:
1701            - Control.Print.intinfDepth controls max length of intinf constants
1702            being printed. (Analogous to Control.Print.stringDepth.)
1703            - Cleanup in printutil and pputil: got rid of unused stuff and
1704            duplicates; replaced some of the code with code that makes better
1705            use of library functionality.
1706    
1707    CM:
1708            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
1709            are no longer silent.
1710    
1711    CKIT:
1712            Fixed mismatched uses of Int32 and LargeInt.  I always decided
1713            in favor of LargeInt -- which is now the same as IntInf.
1714            CKIT-knowledgable people should check whether this is what's
1715            intended and otherwise change things back to using Int32 or
1716            FixedInt.
1717    
1718    Throughout the code:
1719            Started using IntInf.int literals and built-in operations
1720            (e.g., comparison with 0) where this seems appropriate.
1721    
1722    
1723    ----------------------------------------------------------------------
1724    Name: Dave MacQueen (dbm@cs.uchicago.edu)
1725    Date: 2003/08/13 11:36:00 CDT
1726    Tag: dbm-20030813-mcz-merge1
1727    Description:
1728    
1729    Merging changes from the mcz-branch development branch into trunk.
1730    These changes involve replacement of the emulated old prettyprinter
1731    interface with direct use of the SML/NJ Lib PP library, and fixing
1732    of a couple of bugs (895, 1186) relating to error messages.  A new
1733    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
1734    has been added.
1735    
1736    ----------------------------------------------------------------------
1737    Name: Matthias Blume (blume (at) tti - c (dot) org)
1738    Date: 2003/08/11 15:45:00 CDT
1739    Tag: blume-20030811-windows
1740    Description:
1741    
1742    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
1743    
1744        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
1745    
1746    This patch restores SML/NJ's ability to run under win32.  There are a
1747    number of changes, including fixes for several bugs that had gone
1748    unnoticed until now:
1749    
1750      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
1751        trying to run the win32 version from within cygwin.)
1752    
1753      - There are a number of simple .bat scripts that substitute for their
1754        corresponding Unix shell-scripts.  (See below.)
1755    
1756      - The internals of ml-build have been modified slightly.  The main
1757        difference is that instead of calling ".link-sml" (or link-sml.bat)
1758        using OS.Process.system, the ML process delegates this task back
1759        to the script.  Otherwise problems arise in mixed environments such
1760        as Cygwin where scripts look and work like Unix scripts, but
1761        where OS.Process.system cannot run them.
1762    
1763      - In CM, the srcpath pickler used native pathname syntax -- which
1764        is incorrect in the case of cross-compilation.  The new pickle format
1765        is independent of platform-specific naming conventions.
1766    
1767      - Path configuration files (such as lib/pathconfig) can now choose
1768        between native and standard syntax.  Placing a line of the form
1769    
1770            standard!
1771    
1772        into the file causes all subsequent paths to be interpreted using
1773        CM standard pathname syntax (= Unix conventions); a line
1774    
1775            native!
1776    
1777        switches back to native style.  This was needed so that
1778        path config files can be written portably, see src/system/pathconfig.
1779    
1780      - Runtime system:
1781    
1782          - win32-filesys.c:  get_file_time and set_file_time now
1783            access modification time, not creation time.
1784    
1785          - I/O code made aware of new array representation.
1786    
1787          - Bug fixes in X86.prim.masm.
1788    
1789       - src/system/makeml made aware of win32. (For use under cygwin
1790            and other Unix-environments for windows.)
1791    
1792       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
1793         which caused BinIO.inputAll to fail consistently.
1794    
1795    .bat scripts:
1796    
1797       Windows .bat scripts assume that SMLNJ_HOME is defined.
1798    
1799       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
1800         applications (sml, ml-yacc, ml-lex).
1801       - ml-build.bat: analogous to ml-build.
1802       - config\install.bat:  Analogous to config/install.sh.  This requires
1803         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
1804         (nmake etc. must be on the path, and vcvars32 must have been run.)
1805         Moreover, sources for ml-lex and ml-yacc need to exist under src,
1806         and the bootfile hierarchy must have been unpacked under
1807         sml.boot.x86-win32.
1808         The script is very primitive and does a poor job at error checking.
1809         It only installs the base system, ml-lex, and ml-yacc.  No other
1810         libraries are being installed (i.e., you get only those that
1811         are part of the compiler.)
1812       - link-sml.bat:  analogous to .link-sml, but not currently used
1813    
1814    Unrelated bug fixes:
1815    
1816      - ml-nlffigen now exports structures ST_* corresponding to incomplete
1817        types.
1818      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
1819        otherwise.)
1820    
1821    ----------------------------------------------------------------------
1822    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
1823    Date: 2003/06/17
1824    Tag: macqueen-20030617-bug895
1825    Description:
1826    
1827    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
1828    Tag will be used for new development branch (mcz-branch) for
1829    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
1830    
1831    ----------------------------------------------------------------------
1832    Name: Matthias Blume (blume (at) tti - c (dot) org)
1833    Date: 2003/05/27 16:55:00 CDT
1834    Tag: blume-20030527-polyeq
1835    Description:
1836    
1837    Tried to eliminated most cases of polymorphic equality.
1838    
1839    ----------------------------------------------------------------------
1840    Name: Matthias Blume (blume (at) tti - c (dot) org)
1841    Date: 2003/05/21 17:45:00 CDT
1842    Tag: blume-20030517-complete
1843    Description:
1844    
1845    Two changes:
1846    
1847      1. Added a flag for controlling whether non-exhaustive bindings will
1848         be treated as errors (default is false).
1849      2. Cleaned up the *entire* source tree so that CMB.make goes through
1850         without a single non-exhaustive match- or bind warning.
1851    
1852    ----------------------------------------------------------------------
1853    Name: Matthias Blume (blume (at) tti - c (dot) org)
1854    Date: 2003/05/17 10:20:00 CDT
1855    Tag: blume-20030517-absyn
1856    Description:
1857    
1858    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
1859    
1860       This mainly affects the quality of error messages.  However, some
1861       of the code is now more straightforward than before. (Treatment of
1862       the above four constructs in translate.sml is much simpler than
1863       the "macro-expansion" that was going on before.  Plus, the mach-
1864       compiler no longer gets invoked just to be able to compile an
1865       if-expression.)
1866    
1867    2. The ErrorMsg.Error exception is now caught and absorbed by the
1868       interactive loop.
1869    
1870    ----------------------------------------------------------------------
1871    Name: Allen Leung
1872    Date: 2003/05/16 13:05:00 CDT
1873    Tag: leunga-20030516-cygwin-runtime
1874    Description:
1875    
1876        Ported the runtime system to cygwin, which uses the unix
1877    x86-unix bin files.  Missing/buggy features:
1878    
1879       o getnetbyname, getnetbyaddr: these functions seem to be missing in
1880         the Cygwin library.
1881       o Ctrl-C handling may be flaky.
1882       o Windows system calls and Windows I/O are not supported.
1883    
1884       A new set of binfiles is located at:
1885    
1886         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
1887    
1888       This is only needed for bootstrapping the cygwin version of smlnj.
1889       Other x86 versions can use the existing binfiles.
1890    
1891    ----------------------------------------------------------------------
1892    Name: Matthias Blume
1893    Date: 2003/04/08 15:42:00 CDT
1894    Tag: blume-20030408-listpair
1895    Description:
1896    
1897    1. Added a target 'mlrisc' to installer.
1898    
1899    2. Added missing elements to structure ListPair.
1900    
1901    ----------------------------------------------------------------------
1902    Name: Allen Leung
1903    Date: 2003/01/07 10:40:00 EST
1904    Tag: leunga-20030107-int-rem
1905    Description:
1906    
1907    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
1908    The arguments to the SUBL instruction were swapped.
1909    
1910    ----------------------------------------------------------------------
1911    Name: Matthias Blume
1912    Date: 2002/12/12 16:25:00 EST
1913    Tag: blume-20021212-risc-ra
1914    Description:
1915    
1916    Fixed a serious bug in the rewrite code for FP spilling/reloading that
1917    sent the RA into an infinite loop when floating point registers get
1918    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
1919    110.37 and 110.38.)
1920    There was another set of potential problems related to the handling of
1921    MLRISC annotations (but those did not yet cause real problems, apparently).
1922    
1923    ----------------------------------------------------------------------
1924    Name: Matthias Blume
1925    Date: 2002/12/06 22:40:00 EST
1926    Tag: blume-20021206-cm-fileid
1927    Description:
1928    
1929    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
1930    This fixes the problem of CM getting confused by files that suddenly
1931    change their identity (e.g., by getting unlinked and recreated by some
1932    text editor such as vi).  There might be a better/cheaper/cleaner way
1933    of doing this, but for now this will have to do.
1934    
1935    ----------------------------------------------------------------------
1936    Name: Matthias Blume
1937    Date: 2002/10/28 09:50:00 EST
1938    Tag: blume-20021028-typecheck
1939    Description:
1940    
1941    Exported structure Typecheck from $smlnj/viscomp/core.cm.
1942    
1943    ----------------------------------------------------------------------
1944    Name: Matthias Blume
1945    Date: 2002/10/17 09:10:00 EDT
1946    Tag: Release_110_42
1947    Description:
1948    
1949    In good old tradition, there has been a slight hiccup so that we have
1950    to patch 110.42 after the fact.   The old release tag has been replaced
1951    (see below).
1952    
1953    The change solves a problem with two competing approaches the
1954    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
1955    each other's way.
1956    
1957    This change only affects the runtime system code and the installer script.
1958    (No new bootfiles.)
1959    
1960    ----------------------------------------------------------------------
1961    Name: Matthias Blume
1962    Date: 2002/10/16 12:00:00 EDT
1963    Tag: Release_110_42_removed
1964    Description:
1965    
1966    New working release.  New bootfiles.
1967    
1968    ----------------------------------------------------------------------
1969    Name: Matthias Blume
1970    Date: 2002/10/10 13:10:00 EDT
1971    Tag: blume-20021010-ppc-divs
1972    Description:
1973    
1974    The mltree operator DIVS must be implemented with an overflow check on
1975    the PPC because the hardware indicates divide-by-zero using "overflow" as
1976    well.
1977    
1978    ----------------------------------------------------------------------
1979    Name: Matthias Blume
1980    Date: 2002/07/23 11:45:00
1981    Tag: blume-20020723-smlnj-home
1982    Description:
1983    
1984    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
1985    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
1986    CM_PATHCONFIG is also set), the path configuration file is assumed
1987    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
1988    the entire tree to some other place and everything will "just work".
1989    
1990    (Companion commands such as ml-build and ml-makedepend also sense this
1991    variable.)
1992    
1993    ----------------------------------------------------------------------
1994    Name: Matthias Blume
1995    Date: 2002/07/12 21:19:00 EDT
1996    Tag: blume-20020712-liveness
1997    Description:
1998    
1999    Exported two useful "step" functions from liveness module (MLRISC).
2000    
2001    ----------------------------------------------------------------------
2002    Name: Matthias Blume
2003    Date: 2002/07/05 16:00 EDT
2004    Tag: Release_110_41
2005    Description:
2006    
2007    New working release.  New bootfiles.
2008    
2009    ----------------------------------------------------------------------
2010    Name: Matthias Blume
2011    Date: 2002/07/05 10:25:00 EDT
2012    Tag: blume-20020705-btimp
2013    Description:
2014    
2015    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
2016    other clients can set up backtracing support.
2017    
2018    ----------------------------------------------------------------------
2019    Name: Matthias Blume
2020    Date: 2002/06/25 17:23:00 EDT
2021    Tag: blume-20020625-fpmax
2022    Description:
2023    
2024    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
2025    case.  (The sense of the isNaN test was reversed -- which made min and
2026    max always return their first argument.)
2027    
2028    ----------------------------------------------------------------------
2029    Name: Matthias Blume
2030    Date: 2002/06/11
2031    Tag: blume-20020611-unixpath
2032    Description:
2033    
2034    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
2035    
2036    ----------------------------------------------------------------------
2037    Name: Matthias Blume
2038    Date: 2002/06/10 16:35:00 EDT
2039    Tag: blume-20020610-ieeereal
2040    Description:
2041    
2042    I back-ported my implementation of IEEEReal.fromString from the
2043    idlbasis-devel branch so that we can test it.
2044    
2045    Another small change is that ppDec tries to give more information
2046    than just "<sig>" in the case of functors.  However, this code is
2047    broken in some mysterious way if the functor's body's signature
2048    has not been declared by ascription but gets inferred from the
2049    implementation.  This needs fixing...
2050    
2051    ----------------------------------------------------------------------
2052    Name: Matthias Blume
2053    Date: 2002/05/31
2054    Tag: blume-20020531-btrace-mode
2055    Description:
2056    
2057    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
2058    the wayside when I switched over to using Controls everywhere.)
2059    
2060    ----------------------------------------------------------------------
2061    Name: Lal George
2062    Date: 2002/05/23 12:21:40 EDT
2063    Tag: george-20020523-visual-labels
2064    Description:
2065    
2066    Labels are now displayed in the graphical output to make
2067    the fall-through and target blocks obvious.
2068    
2069    ----------------------------------------------------------------------
2070    Name: Matthias Blume
2071    Date: 2002/05/22 11:03:00 EDT
2072    Tag: blume-20020522-shrink
2073    Description:
2074    
2075    John tweaked yesterday's fix for 1131 to handle an out-of-memory
2076    situation that comes up when allocating huge arrays.
2077    
2078    ----------------------------------------------------------------------
2079    Name: Matthias Blume
2080    Date: 2002/05/21 16:00:00 EDT
2081    Tag: Release_110_40
2082    Description:
2083    
2084    New working release (110.40).  New bootfiles.
2085    
2086    [Also: John Reppy fixed GC bug 1131.]
2087    
2088    ----------------------------------------------------------------------
2089    Name: Matthias Blume
2090    Date: 2002/05/21 12:35:00 EDT
2091    Tag: blume-20020521-cmdoc
2092    Description:
2093    
2094    CM documentation update.
2095    
2096    ----------------------------------------------------------------------
2097    Name: Matthias Blume
2098    Date: 2002/05/21 10:55:00 EDT
2099    Tag: blume-20020521-misc
2100    Description:
2101    
2102    - John tweaked runtime to be silent on heap export (except when
2103      GC messages are on).
2104    - I added a few more things (cross-compiling versions of CMB) to
2105      config/preloads (as suggestions).
2106    
2107    ----------------------------------------------------------------------
2108    Name: Matthias Blume
2109    Date: 2002/05/20 22:25:00 EDT
2110    Tag: blume-20020520-controls
2111    Description:
2112    
2113    - Added ControlUtil structure to control-lib.cm.
2114    - Use it throughout.
2115    - Used Controls facility to define MLRISC controls (as opposed to
2116      registering MLRISC control ref cells with Controls after the fact)
2117    - Fixed messed-up controls priorities.
2118    
2119    * Removed again all the stuff from config/preloads that one wouldn't
2120      be able to preload at the time the initial heap image is built.
2121      (Many libraries, e.g., CML, do not exist yet at this time.  The
2122       only libraries that can be preloaded via config/preloads are those
2123       that come bundled with the bootfiles.)
2124    
2125    ----------------------------------------------------------------------
2126    Name: Matthias Blume
2127    Date: 2002/05/20 10:59:00 EDT
2128    Tag: blume-20020520-preloads
2129    Description:
2130    
2131    Added a lot of commented-out suggestions for things to be included
2132    in config/preloads.
2133    
2134    ----------------------------------------------------------------------
2135    Name: Allen Leung
2136    Date: 2002/05/18 14:20:00 EDT
2137    Tag: leunga-20020518-mdl
2138    Description:
2139    
2140    o Made the mdl tool stuff compile and run again.
2141    o I've disabled all the stuff that depends on RTL specifications; they
2142      are all badly broken anyway.
2143    
2144    ----------------------------------------------------------------------
2145    Name: Matthias Blume
2146    Date: 2002/05/17 16:49:00 EDT
2147    Tag: blume-20020517-controls
2148    Description:
2149    
2150    0. John Reppy made several modifications to the SML/NJ library.
2151       In particular, there is a shiny new controls-lib.cm.
2152    
2153    1. Pushed new controls interface through compiler so that everything
2154       compiles again.
2155    
2156    2. Added FormatComb and FORMAT_COMB to the CML version of the
2157       SML/NJ library (so that CML compiles again).
2158    
2159    3. Modified init scripts because XXX_DEFAULT environment variables
2160       are no longer with us.  (Boot-time initialization is now done
2161       using the same environment variables that are also used for
2162       startup-time initialization of controls.)
2163    
2164    ----------------------------------------------------------------------
2165    Name: Lal George
2166    Date: 2002/05/15 09:20:10 EDT
2167    Tag: george-20020515-pseudo-op-decls
2168    Description:
2169    
2170            All pseudo-ops emitted before the first segment declaration
2171    such as TEXT, DATA, and BSS directives are assumed to be global
2172    declarations and are emitted first in the assembly file. This is
2173    useful in a number of situations where one has pseudo-ops that are not
2174    specific to any segment, and also works around the constraint that one
2175    cannot have client pseudo-ops in the TEXT segment.
2176    
2177            Because no segment is associated with these declarations it is
2178    an error to allocate any space or objects before the first segment
2179    directive and an exception will be raised. However, we cannot make
2180    this check for client pseudo-ops.
2181    
2182            These top level declarations are a field in the CFG graph_info.
2183    In theory you can continue to add to this field after the CFG has been
2184    built -- provided you know what you are doing;-)
2185    
2186    ----------------------------------------------------------------------
2187    Name: Matthias Blume
2188    Date: 2002/05/13 16:40:00 EDT
2189    Tag: blume-20020513-pp-etc
2190    Description:
2191    
2192    A few minor bugfixes:
2193    
2194      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
2195        (Bogus printouts for redefined bindings still occur.  Compiler
2196        bug should no longer occur now.  We need to redo the prettyprinter
2197        from scratch.)
2198    
2199      - CM pathname printer now also adds escape sequences for ( and )
2200    
2201      - commend and docu fixes for ml-nlffi
2202    
2203    ----------------------------------------------------------------------
2204    Name: Matthias Blume
2205    Date: 2002/05/10 16:40:00 EDT
2206    Tag: blume-20020510-erg-textio
2207    Description:
2208    
2209    Applied the following bugfix provided by Emden Gansner:
2210    
2211        Output is corrupted when outputSubstr is used rather than output.
2212    
2213        The problem occurs when a substring
2214    
2215            ss = (s, dataStart, dataLen)
2216    
2217        where dataStart > 0, fills a stream buffer with avail bytes left.
2218        avail bytes of s, starting at index dataStart, are copied into the
2219        buffer, the buffer is flushed, and then the remaining dataLen-avail
2220        bytes of ss are copied into the beginning of the buffer. Instead of
2221        starting this copy at index dataStart+avail in s, the current code
2222        starts the copy at index avail.
2223    
2224        Fix:
2225        In text-io-fn.sml, change line 695 from
2226             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
2227        to
2228             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
2229    
2230    ----------------------------------------------------------------------
2231    Name: Matthias Blume
2232    Date: 2002/04/12 13:55:00 EDT
2233    Tag: blume-20020412-assyntax
2234    Description:
2235    
2236    1. Grabbed newer assyntax.h from the XFree86 project.
2237    2. Fiddled with how to compile X86.prim.asm without warnings.
2238    3. (Very) Minor cleanup in CM.
2239    
2240    ----------------------------------------------------------------------
2241    Name: Matthias Blume
2242    Date: 2002/04/01 (no joke!) 17:07:00 EST
2243    Tag: blume-20020401-x86div
2244    Description:
2245    
2246    Added full support for div/mod/rem/quot on the x86, using the machine
2247    instruction's two results (without clumsily recomputing the remainder)
2248    directly where appropriate.
2249    
2250    Some more extensive power-of-two support was added to the x86 instruction
2251    selector (avoiding expensive divs, mods, and muls where they can be
2252    replaced with cheaper shifts and masks).  However, this sort of thing
2253    ought to be done earlier, e.g., within the CPS optimizer so that
2254    all architectures benefit from it.
2255    
2256    The compiler compiles to a fixed point, but changes might be somewhat
2257    fragile nevertheless.  Please, report any strange things that you might
2258    see wrt. div/mod/quot/rem...
2259    
2260    ----------------------------------------------------------------------
2261    Name: Matthias Blume
2262    Date: 2002/03/29 17:22:00
2263    Tag: blume-20020329-div
2264    Description:
2265    
2266    Fixed my broken div/mod logic.  Unfortunately, this means that the
2267    inline code for div/mod now has one more comparison than before.
2268    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
2269    The problem was with quotient = 0, because that alone does not tell
2270    us which way the rounding went.  One then has to look at whether
2271    remainder and divisor have the same sign...  :(
2272    
2273    Anyway, I replaced the bootfiles with fresh ones...
2274    
2275    ----------------------------------------------------------------------
2276    Name: Matthias Blume
2277    Date: 2002/03/29 14:10:00 EST
2278    Tag: blume-20020329-inlprims
2279    Description:
2280    
2281    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
2282    
2283    Primops have changed. This means that the bin/boot-file formats have
2284    changed as well.
2285    
2286    To make sure that there is no confusion, I made a new version.
2287    
2288    
2289    CHANGES:
2290    
2291    * removed REMT from mltree (remainder should never overflow).
2292    
2293    * added primops to deal with divisions of all flavors to the frontend
2294    
2295    * handled these primops all the way through so they map to their respective
2296      MLRISC support
2297    
2298    * used these primops in the implementation of Int, Int32, Word, Word32
2299    
2300    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
2301    
2302    * parameterized INLMIN, INLMAX, and INLABS by a numkind
2303    
2304    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
2305      floating point
2306    
2307    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
2308      and Real (but Real.abs maps to a separate floating-point-only primop)
2309    
2310    
2311    TODO items:
2312    
2313    * Hacked Alpha32 instruction selection, disabling the selection of REMx
2314      instructions because the machine instruction encoder cannot handle
2315      them.  (Hppa, PPC, and Sparc instruction selection did not handle
2316      REM in the first place, and REM is supported by the x86 machine coder.)
2317    
2318    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
2319      selection phase.  (The two can be streamlined because the hardware
2320      delivers both quotient and remainder at the same time anyway.)
2321    
2322    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
2323      (Currently the behavior is inconsistent both across architectures and
2324      wrt. the draft Basis spec.)
2325    
2326    * Word8 should eventually be handled natively, too.
2327    
2328    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
2329      as if there currently is no execution path that could trigger it in
2330      SML/NJ.  (The assumptions underlying functions arith and promotable do not
2331      hold for things like multiplication and division.)
2332    
2333    ----------------------------------------------------------------------
2334    Name: Matthias Blume
2335    Date: 2002/03/27 16:27:00 EST
2336    Tag: blume-20020327-mlrisc-divisions
2337    Description:
2338    
2339    Added support for all four division operations (ML's div, mod, quot,
2340    and rem) to MLRISC.  In the course of doing so, I also rationalized
2341    the naming (no more annoying switch-around of DIV and QUOT), by
2342    parameterizing the operation by div_rounding_mode (which can be either
2343    DIV_TO_ZERO or DIV_TO_NEGINF).
2344    
2345    The generic MLTreeGen functor takes care of compiling all four
2346    operations down to only round-to-zero div.
2347    
2348    Missing pieces:
2349    
2350      * Doing something smarter than relying on MLTreeGen on architectures
2351        like, e.g., the x86 where hardware division delivers both quotient and
2352        remainder at the same time.  With this, the implementation of the
2353        round-to-neginf operations could be further streamlined.
2354    
2355      * Remove inlining support for div/mod/rem from the frontend and replace it
2356        with primops that get carried through to the backend.  Do this for all
2357        int and word types.
2358    
2359    ----------------------------------------------------------------------
2360    Name: Matthias Blume
2361    Date: 2002/03/25 17:25:00 EST
2362    Tag: blume-20020325-divmod
2363    Description:
2364    
2365    I improved (hopefully without breaking them) the implementation of Int.div,
2366    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
2367    advantage of the following observations:
2368    
2369      Let  q = x quot y      r = x rem y
2370           d = x div  y      m = x mod y
2371    
2372    where "quot" is the round-to-zero version of integer division that
2373    hardware usually provides.  Then we have:
2374    
2375         r = x - q * y        where neither the * nor the - will overflow
2376         d = if q >= 0 orelse x = q * y then q else q - 1
2377                              where neither the * nor the - will overflow
2378         m = if q >= 0 orelse r = 0 then r else r + y
2379                              where the + will not overflow
2380    
2381    This results in substantial simplification of the generated code.
2382    The following table shows the number of CFG nodes and edges generated
2383    for
2384            fun f (x, y) = x OPER y
2385            (* with OPER \in div, mod, quot, rem *)
2386    
2387    
2388        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
2389        --------------------------------------------------------
2390         div |         24 |         39 |         12 |         16
2391         mod |         41 |         71 |         12 |         16
2392        quot |          8 |         10 |          8 |         10
2393         rem |         10 |         14 |          8 |         10
2394    
2395    
2396    ----------------------------------------------------------------------
2397    Name: Matthias Blume
2398    Date: 2002/03/25 22:06:00 EST
2399    Tag: blume-20020325-cprotobug
2400    Description:
2401    
2402    Fixed a bug in cproto (c prototype decoder).
2403    
2404    ----------------------------------------------------------------------
2405    Name: Matthias Blume
2406    Date: 2002/03/25 16:00:00 EST
2407    Tag: blume-20020325-raw-primops
2408    Description:
2409    
2410    I did some cleanup to Allen's new primop code and
2411    replaced yesterday's bootfiles with new ones.
2412    (But they are stored in the same place.)
2413    
2414    ----------------------------------------------------------------------
2415    Name: Matthias Blume
2416    Date: 2002/03/24 22:40:00 EST
2417    Tag: blume-20020324-bootfiles
2418    Description:
2419    
2420    Made the bootfiles that Allen asked for.
2421    
2422    ----------------------------------------------------------------------
2423    Name: Allen Leung
2424    Date: 2002/03/23 15:50:00 EST
2425    Tag: leunga-20020323-flint-cps-rcc-primops
2426    Description:
2427    
2428      1. Changes to FLINT primops:
2429    
2430        (* make a call to a C-function;
2431         * The primop carries C function prototype information and specifies
2432         * which of its (ML-) arguments are floating point. C prototype
2433         * information is for use by the backend, ML information is for
2434         * use by the CPS converter. *)
2435      | RAW_CCALL of { c_proto: CTypes.c_proto,
2436                       ml_args: ccall_type list,
2437                       ml_res_opt: ccall_type option,
2438                       reentrant : bool
2439                     } option
2440       (* Allocate uninitialized storage on the heap.
2441        * The record is meant to hold short-lived C objects, i.e., they
2442        * are not ML pointers.  With the tag, the representation is
2443        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
2444        *)
2445      | RAW_RECORD of {tag:bool,sz:int}
2446      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
2447    
2448      2.  These CPS primops are now overloaded:
2449    
2450           rawload of {kind:numkind}
2451           rawstore of {kind:numkind}
2452    
2453          The one argument form is:
2454    
2455             rawload {kind} address
2456    
2457          The two argument form is:
2458    
2459             rawload {kind} [ml object, byte-offset]
2460    
2461      3. RAW_CCALL/RCC now takes two extra arguments:
2462    
2463         a. The first is whether the C call is reentrant, i.e., whether
2464            ML state should be saved and restored.
2465         b. The second argument is a string argument specifying the name of
2466            library and the C function.
2467    
2468         These things are currently not handled in the code generator, yet.
2469    
2470      4. In CProto,
2471    
2472         An encoding type of "bool" means "ml object" and is mapped into
2473         C prototype of PTR.  Note that "bool" is different than "string",
2474         even though "string" is also mapped into PTR, because "bool"
2475         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
2476    
2477      5. Pickler/unpicker
2478    
2479         Changed to handle RAW_RECORD and newest RAW_CCALL
2480    
2481      6. MLRiscGen,
2482    
2483         1. Changed to handle the new rawload/rawstore/rawrecord operators.
2484         2. Code for handling C Calls has been moved to a new module CPSCCalls,
2485            in the file CodeGen/cpscompile/cps-c-calls.sml
2486    
2487      7. Added the conditional move operator
2488    
2489             condmove of branch
2490    
2491         to cps.  Generation of this is still buggy so it is currently
2492         disabled.
2493    
2494    ----------------------------------------------------------------------
2495    Name: Lal George
2496    Date: 2002/03/22 14:18:25 EST
2497    Tag: george-20020322-cps-branch-prob
2498    Description:
2499    
2500    Implemented the Ball-Larus branch prediction-heuristics, and
2501    incorporated graphical viewers for control flow graphs.
2502    
2503    Ball-Larus Heuristics:
2504    ---------------------
2505    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
2506    
2507    By design it uses the Dempster-Shafer theory for combining
2508    probabilities.  For example, in the function:
2509    
2510        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
2511    
2512    the ball-larus heuristics predicts that the n=0 is unlikely
2513    (OH-heuristic), and the 'then' branch is unlikely because of the
2514    RH-heuristic -- giving the 'then' branch an even lower combined
2515    probability using the Dempster-Shafer theory.
2516    
2517    Finally, John Reppy's loop analysis in MLRISC, further lowers the
2518    probability of the 'then' branch because of the loop in the else
2519    branch.
2520    
2521    
2522    Graphical Viewing:
2523    ------------------
2524    I merely plugged in Allen's graphical viewers into the compiler. The
2525    additional code is not much. At the top level, saying:
2526    
2527            Control.MLRISC.getFlag "cfg-graphical-view" := true;
2528    
2529    will display the graphical view of the control flow graph just before
2530    back-patching.  daVinci must be in your path for this to work. If
2531    daVinci is not available, then the default viewer can be changed
2532    using:
2533    
2534            Control.MLRISC.getString "viewer"
2535    
2536    which can be set to "dot" or "vcg" for the corresponding viewers. Of
2537    course, these viewers must be in your path.
2538    
2539    The above will display the compilation unit at the level of clusters,
2540    many of which are small, boring, and un-interesting. Also setting:
2541    
2542            Control.MLRISC.getInt "cfg-graphical-view_size"
2543    
2544    will display clusters that are larger than the value set by the above.
2545    
2546    
2547    ----------------------------------------------------------------------
2548    Name: Matthias Blume
2549    Date: 2002/03/21 22:20:00 EST
2550    Tag: blume-20020321-kmp-bugfix
2551    Description:
2552    
2553    Changed the interface to the KMP routine in PreString and fixed
2554    a minor bug in one place where it was used.
2555    
2556    ----------------------------------------------------------------------
2557    Name: Allen Leung
2558    Date: 2002/03/21 20:30:00 EST
2559    Tag: leunga-20020321-cfg
2560    Description:
2561    
2562      Fixed a potential problem in cfg edge splitting.
2563    
2564    ----------------------------------------------------------------------
2565    Name: Allen Leung
2566    Date: 2002/03/21 17:15:00 EST
2567    Tag: leunga-20020321-x86-fp-cfg
2568    Description:
2569    
2570      1. Recoded the buggy parts of x86-fp.
2571    
2572         a. All the block reordering code has been removed.
2573            We now depend on the block placement phases to do this work.
2574    
2575         b. Critical edge splitting code has been simplified and moved into the
2576            CFG modules, as where they belong.
2577    
2578         Both of these were quite buggy and complex.  The code is now much, much
2579         simpler.
2580    
2581      2. X86 backend.
2582    
2583         a. Added instructions for 64-bit support.  Instruction selection for
2584            64-bit has not been committed, however, since that
2585            requires changes to MLTREE which haven't been approved by
2586            Lal and John.
2587    
2588         b. Added support for FUCOMI and FUCOMIP when generating code for
2589            PentiumPro and above.  We only generate these instructions in
2590            the fast-fp mode.
2591    
2592         c. Added cases for JP and JNP in X86FreqProps.
2593    
2594      3. CFG
2595    
2596         CFG now has a bunch of methods for edge splitting and merging.
2597    
2598      4. Machine description.
2599    
2600         John's simplification of MLTREE_BASIS.fcond broke a few machine
2601         description things:
2602    
2603         rtl-build.{sig,sml} and hppa.mdl fixed.
2604    
2605         NOTE: the machine description stuff in the repository is still broken.
2606               Again, I can't put my fixes in because that involves
2607               changes to MLTREE.
2608    
2609    ----------------------------------------------------------------------
2610    Name: Matthias Blume
2611    Date: 2002/03/20 15:55:00 EST
2612    Tag: blume-20020320-kmp
2613    Description:
2614    
2615    Implemented Knuth-Morris-Pratt string matching in PreString and used
2616    it for String.isSubstring, Substring.isSubstring, and
2617    Substring.position.
2618    
2619    (Might need some stress-testing.  Simple examples worked fine.)
2620    
2621    ----------------------------------------------------------------------
2622    Name: Matthias Blume
2623    Date: 2002/03/19 16:37:00 EST
2624    Tag: blume-20020319-witnesses
2625    Description:
2626    
2627    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
2628    
2629    This implements a generic mechanism for changing constness qualifiers
2630    anywhere within big C types without resorting to outright "casts".
2631    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
2632    modify the constness at the outermost level.)
2633    The implementation of "convert" is based on the idea of "witness"
2634    values -- values that are not used by the operation but whose types
2635    "testify" to their applicability.  On the implementation side, "convert"
2636    is simply a projection (returning its second curried argument).  With
2637    cross-module inlining, it should not result in any machine code being
2638    generated.
2639    
2640    ----------------------------------------------------------------------
2641    Name: Matthias Blume
2642    Date: 2002/03/15 16:40:00 EST
2643    Tag: blume-20020315-basis
2644    Description:
2645    
2646    Provided (preliminary?) implementations for
2647    
2648      {String,Substring}.{concatWith,isSuffix,isSubstring}
2649    
2650    and
2651    
2652      Substring.full
2653    
2654    Those are in the Basis spec but they were missing in SML/NJ.
2655    
2656    ----------------------------------------------------------------------
2657    Name: Matthias Blume
2658    Date: 2002/03/14 21:30:00 EST
2659    Tag: blume-20020314-controls
2660    Description:
2661    
2662    Controls:
2663    ---------
2664    
2665    1. Factored out the recently-added Controls : CONTROLS stuff and put
2666       it into its own library $/controls-lib.cm.  The source tree for
2667       this is under src/smlnj-lib/Controls.
2668    
2669    2. Changed the names of types and functions in this interface, so they
2670       make a bit more "sense":
2671    
2672          module -> registry
2673          'a registry -> 'a group
2674    
2675    3. The interface now deals in ref cells only.  The getter/setter interface
2676       is (mostly) gone.
2677    
2678    4. Added a function that lets one register an already-existing ref cell.
2679    
2680    5. Made the corresponding modifications to the rest of the code so that
2681       everything compiles again.
2682    
2683    6. Changed the implementation of Controls.MLRISC back to something closer
2684       to the original.  In particular, this module (and therefore MLRISC)
2685       does not depend on Controls.  There now is some link-time code in
2686       int-sys.sml that registers the MLRISC controls with the Controls
2687       module.
2688    
2689    CM:
2690    ---
2691    
2692      * One can now specify the lambda-split aggressiveness in init.cmi.
2693    
2694    ----------------------------------------------------------------------
2695    Name: Allen Leung
2696    Date: 2002/03/13 17:30:00 EST
2697    Tag: leunga-20020313-x86-fp-unary
2698    Description:
2699    
2700    Bug fix for:
2701    
2702    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
2703    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
2704    > - fun f(x,(y,z)) = Real.~ y;
2705    > [autoloading]
2706    > [autoloading done]
2707    >       fchsl   (%eax), 184(%esp)
2708    > Error: MLRisc bug: X86MCEmitter.emitInstr
2709    >
2710    > uncaught exception Error
2711    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
2712    
2713    The problem was that the code generator did not generate any fp registers
2714    in this case, and the ra didn't know that it needed to run the X86FP phase to
2715    translate the pseudo fp instruction.   This only happened with unary fp
2716    operators in certain situations.
2717    
2718    ----------------------------------------------------------------------
2719    Name: Matthias Blume
2720    Date: 2002/03/13 14:00:00 EST
2721    Tag: blume-20020313-overload-etc
2722    Description:
2723    
2724    1. Added _overload as a synonym for overload for backward compatibility.
2725       (Control.overloadKW must be true for either version to be accepted.)
2726    
2727    2. Fixed bug in install script that caused more things to be installed
2728       than what was requested in config/targets.
2729    
2730    3. Made CM aware of the (_)overload construct so that autoloading
2731       works.
2732    
2733    ----------------------------------------------------------------------
2734    Name: Matthias Blume
2735    Date: 2002/03/12 22:03:00 EST
2736    Tag: blume-20020312-url
2737    Description:
2738    
2739    Forgot to update BOOT and srcarchiveurl.
2740    
2741    ----------------------------------------------------------------------
2742    Name: Matthias Blume
2743    Date: 2002/03/12 17:30:00 EST
2744    Tag: blume-20020312-version110392
2745    Description:
2746    
2747    Yet another version number bump (because of small changes to the
2748    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
2749    
2750    Changes:
2751    
2752      The new pid generation scheme described a few weeks ago was overly
2753      complicated.  I implemented a new mechanism that is simpler and
2754      provides a bit more "stability":  Once CM has seen a compilation
2755      unit, it keeps its identity constant (as long as you do not delete
2756      those crucial CM/GUID/* files).  This means that when you change
2757      an interface, compile, then go back to the old interface, and
2758      compile again, you arrive at the original pid.
2759    
2760      There now also is a mechanism that instructs CM to use the plain
2761      environment hash as a module's pid (effectively making its GUID
2762      the empty string).  For this, "noguid" must be specified as an
2763      option to the .sml file in question within its .cm file.
2764      This is most useful for code that is being generated by tools such
2765      as ml-nlffigen (because during development programmers tend to
2766      erase the tool's entire output directory tree including CM's cached
2767      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
2768      revert to the old, broken behavior of SML/NJ, but in specific cases
2769      where there is no danger of interface confusion, its use is ok
2770      (I think).
2771    
2772      ml-nlffigen by default generates "noguid" annotations.  They can be
2773      turned off by specifying -guid in its command line.
2774    
2775    ----------------------------------------------------------------------
2776    Name: Lal George
2777    Date: 2002/03/12 12 14:42:36 EST
2778    Tag: george-20020312-frequency-computation
2779    Description:
2780    
2781    Integrated jump chaining and static block frequency into the
2782    compiler. More details and numbers later.
2783    
2784    ----------------------------------------------------------------------
2785    Name: Lal George
2786    Date: 2002/03/11 11 22:38:53 EST
2787    Tag: george-20020311-jump-chain-elim
2788    Description:
2789    
2790    Tested the jump chain elimination on all architectures (except the
2791    hppa).  This is on by default right now and is profitable for the
2792    alpha and x86, however, it may not be profitable for the sparc and ppc
2793    when compiling the compiler.
2794    
2795    The gc test will typically jump to a label at the end of the cluster,
2796    where there is another jump to an external cluster containing the actual
2797    code to invoke gc. This is to allow factoring of common gc invocation
2798    sequences. That is to say, we generate:
2799    
2800            f:
2801               testgc
2802               ja   L1      % jump if above to L1
2803    
2804            L1:
2805               jmp L2
2806    
2807    
2808    After jump chain elimination the 'ja L1' instructions is converted to
2809    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
2810    up being implemented in their long form (if L2 is far away) using:
2811    
2812            jbe     L3      % jump if below or equal to L3
2813            jmp     L2
2814         L3:
2815            ...
2816    
2817    
2818    For large compilation units L2  may be far away.
2819    
2820    
2821    ----------------------------------------------------------------------
2822    Name: Matthias Blume
2823    Date: 2002/03/11 13:30:00 EST
2824    Tag: blume-20020311-mltreeeval
2825    Description:
2826    
2827    A functor parameter was missing.
2828    
2829    ----------------------------------------------------------------------
2830    Name: Allen Leung
2831    Date: 2002/03/11 10:30:00 EST
2832    Tag: leunga-20020311-runtime-string0
2833    Description:
2834    
2835       The representation of the empty string now points to a
2836    legal null terminated C string instead of unit.  It is now possible
2837    to convert an ML string into C string with InlineT.CharVector.getData.
2838    This compiles into one single machine instruction.
2839    
2840    ----------------------------------------------------------------------
2841    Name: Allen Leung
2842    Date: 2002/03/10 23:55:00 EST
2843    Tag: leunga-20020310-x86-call
2844    Description:
2845    
2846       Added machine generation for CALL instruction (relative displacement mode)
2847    
2848    ----------------------------------------------------------------------
2849    Name: Matthias Blume
2850    Date: 2002/03/08 16:05:00
2851    Tag: blume-20020308-entrypoints
2852    Description:
2853    
2854    Version number bumped to 110.39.1.  NEW BOOTFILES!
2855    
2856    Entrypoints: non-zero offset into a code object where execution should begin.
2857    
2858    - Added the notion of an entrypoint to CodeObj.
2859    - Added reading/writing of entrypoint info to Binfile.
2860    - Made runtime system bootloader aware of entrypoints.
2861    - Use the address of the label of the first function given to mlriscGen
2862      as the entrypoint.  This address is currently always 0, but it will
2863      not be 0 once we turn on block placement.
2864    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
2865      with entry points) from mlriscGen.
2866    
2867    ----------------------------------------------------------------------
2868    Name: Allen Leung
2869    Date: 2002/03/07 20:45:00 EST
2870    Tag: leunga-20020307-x86-cmov
2871    Description:
2872    
2873       Bug fixes for CMOVcc on x86.
2874    
2875       1. Added machine code generation for CMOVcc
2876       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
2877       3. CMOVcc cannot have an immediate operand as argument.
2878    
2879    ----------------------------------------------------------------------
2880    Name: Matthias Blume
2881    Date: 2002/03/07 16:15:00 EST
2882    Tag: blume-20020307-controls
2883    Description:
2884    
2885    This is a very large but mostly boring patch which makes (almost)
2886    every tuneable compiler knob (i.e., pretty much everything under
2887    Control.* plus a few other things) configurable via both the command
2888    line and environment variables in the style CM did its configuration
2889    until now.
2890    
2891    Try starting sml with '-h' (or, if you are brave, '-H')
2892    
2893    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
2894    implements the underlying generic mechanism.
2895    
2896    The interface to some of the existing such facilities has changed somewhat.
2897    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
2898    (The getFoo interface is still there for backward-compatibility, but its
2899    use is deprecated.)
2900    
2901    The ml-build script passes -Cxxx=yyy command-line arguments through so
2902    that one can now twiddle the compiler settings when using this "batch"
2903    compiler.
2904    
2905    TODO items:
2906    
2907    We should go through and throw out all controls that are no longer
2908    connected to anything.  Moreover, we should go through and provide
2909    meaningful (and correct!) documentation strings for those controls
2910    that still are connected.
2911    
2912    Currently, multiple calls to Controls.new are accepted (only the first
2913    has any effect).  Eventually we should make sure that every control
2914    is being made (via Controls.new) exactly once.  Future access can then
2915    be done using Controls.acc.
2916    
2917    Finally, it would probably be a good idea to use the getter-setter
2918    interface to controls rather than ref cells.  For the time being, both
2919    styles are provided by the Controls module, but getter-setter pairs are
2920    better if thread-safety is of any concern because they can be wrapped.
2921    
2922    *****************************************
2923    
2924    One bug fix: The function blockPlacement in three of the MLRISC
2925    backpatch files used to be hard-wired to one of two possibilities at
2926    link time (according to the value of the placementFlag).  But (I
2927    think) it should rather sense the flag every time.
2928    
2929    *****************************************
2930    
2931    Other assorted changes (by other people who did not supply a HISTORY entry):
2932    
2933    1. the cross-module inliner now works much better (Monnier)
2934    2. representation of weights, frequencies, and probabilities in MLRISC
2935       changed in preparation of using those for weighted block placement
2936       (Reppy, George)
2937    
2938    ----------------------------------------------------------------------
2939    Name: Lal George
2940    Date: 2002/03/07 14:44:24 EST 2002
2941    Tag: george-20020307-weighted-block-placement
2942    
2943    Tested the weighted block placement optimization on all architectures
2944    (except the hppa) using AMPL to generate the block and edge frequencies.
2945    Changes were required in the machine properties to correctly
2946    categorize trap instructions. There is an MLRISC flag
2947    "weighted-block-placement" that can be used to enable weighted block
2948    placement, but this will be ineffective without block/edge
2949    frequencies (coming soon).
2950    
2951    
2952    ----------------------------------------------------------------------
2953    Name: Lal George
2954    Date: 2002/03/05 17:24:48 EST
2955    Tag: george-20020305-linkage-cluster
2956    
2957    In order to support the block placement optimization, a new cluster
2958    is generated as the very first cluster (called the linkage cluster).
2959    It contains a single jump to the 'real' entry point for the compilation
2960    unit. Block placement has no effect on the linkage cluster itself, but
2961    all the other clusters  have full freedom in the manner in which they
2962    reorder blocks or functions.
2963    
2964    On the x86 the typical linkage code that is generated is:
2965       ----------------------
2966            .align 2
2967       L0:
2968            addl    $L1-L0, 72(%esp)
2969            jmp     L1
2970    
2971    
2972            .align  2
2973       L1:
2974       ----------------------
2975    
2976    72(%esp) is the memory location for the stdlink register. This
2977    must contain the address of the CPS function being called. In the
2978    above example, it contains the address of  L0; before
2979    calling L1 (the real entry point for the compilation unit), it
2980    must contain the address for L1, and hence
2981    
2982            addl $L1-L0, 72(%esp)
2983    
2984    I have tested this on all architectures except the hppa.The increase
2985    in code size is of course negligible
2986    
2987    ----------------------------------------------------------------------
2988    Name: Allen Leung
2989    Date: 2002/03/03 13:20:00 EST
2990    Tag: leunga-20020303-mlrisc-tools
2991    
2992      Added #[ ... ] expressions to mlrisc tools
2993    
2994    ----------------------------------------------------------------------
2995    Name: Matthias Blume
2996    Date: 2002/02/27 12:29:00 EST
2997    Tag: blume-20020227-cdebug
2998    Description:
2999    
3000    - made types in structure C and C_Debug to be equal
3001    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
3002    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
3003    
3004    ----------------------------------------------------------------------
3005    Name: Matthias Blume
3006    Date: 2002/02/26 12:00:00 EST
3007    Tag: blume-20020226-ffi
3008    Description:
3009    
3010    1. Fixed a minor bug in CM's "noweb" tool:
3011       If numbering is turned off, then truly don't number (i.e., do not
3012       supply the -L option to noweb).  The previous behavior was to supply
3013       -L'' -- which caused noweb to use the "default" line numbering scheme.
3014       Thanks to Chris Richards for pointing this out (and supplying the fix).
3015    
3016    2. Once again, I reworked some aspects of the FFI:
3017    
3018       A. The incomplete/complete type business:
3019    
3020       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
3021         gone!
3022       - ML types representing an incomplete type are now *equal* to
3023         ML types representing their corresponding complete types (just like
3024         in C).  This is still safe because ml-nlffigen will not generate
3025         RTTI for incomplete types, nor will it generate functions that
3026         require access to such RTTI.   But when ML code generated from both
3027         incomplete and complete versions of the C type meet, the ML types
3028         are trivially interoperable.
3029    
3030         NOTE:  These changes restore the full generality of the translation
3031         (which was previously lost when I eliminated functorization)!
3032    
3033       B. Enum types:
3034    
3035       - Structure C now has a type constructor "enum" that is similar to
3036         how the "su" constructor works.  However, "enum" is not a phantom
3037         type because each "T enum" has values (and is isomorphic to
3038         MLRep.Signed.int).
3039       - There are generic access operations for enum objects (using
3040         MLRep.Signed.int).
3041       - ml-nlffigen will generate a structure E_foo for each "enum foo".
3042         * The structure contains the definition of type "mlrep" (the ML-side
3043         representation type of the enum).  Normally, mlrep is the same
3044         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
3045         then mlrep will be defined as a datatype -- thus facilitating
3046         pattern matching on mlrep values.
3047         ("-ec" will be suppressed if there are duplicate values in an
3048          enumeration.)
3049         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
3050         will be generated for each C enum constant xxx.
3051         * Conversion functions m2i and i2m convert between mlrep and
3052         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
3053         * Coversion functions c and ml convert between mlrep and "tag enum".
3054         * Access functions (get/set) fetch and store mlrep values.
3055       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
3056         enumerations are merged into one single enumeration represented by
3057         structure E_'.
3058    
3059    ----------------------------------------------------------------------
3060    Name: Allen Leung
3061    Date: 2002/02/25 04:45:00 EST
3062    Tag: leunga-20020225-cps-spill
3063    
3064    This is a new implementation of the CPS spill phase.
3065    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
3066    In case of problems, replace it with the old file spill.sml
3067    
3068    The current compiler runs into some serious performance problems when
3069    constructing a large record.  This can happen when we try to compile a
3070    structure with many items.  Even a very simple structure like the following
3071    makes the compiler slow down.
3072    
3073        structure Foo = struct
3074           val x_1 = 0w1 : Word32.int
3075           val x_2 = 0w2 : Word32.int
3076           val x_3 = 0w3 : Word32.int
3077           ...
3078           val x_N = 0wN : Word32.int
3079        end
3080    
3081    The following table shows the compile time, from N=1000 to N=4000,
3082    with the old compiler:
3083    
3084    N
3085    1000   CPS 100 spill                           0.04u  0.00s  0.00g
3086           MLRISC ra                               0.06u  0.00s  0.05g
3087              (spills = 0 reloads = 0)
3088           TOTAL                                   0.63u  0.07s  0.21g
3089    
3090    1100   CPS 100 spill                           8.25u  0.32s  0.64g
3091           MLRISC ra                               5.68u  0.59s  3.93g
3092              (spills = 0 reloads = 0)
3093           TOTAL                                   14.71u  0.99s  4.81g
3094    
3095    1500   CPS 100 spill                           58.55u  2.34s  1.74g
3096           MLRISC ra                               5.54u  0.65s  3.91g
3097              (spills = 543 reloads = 1082)
3098           TOTAL                                   65.40u  3.13s  6.00g
3099    
3100    2000   CPS 100 spill                           126.69u  4.84s  3.08g
3101           MLRISC ra                               0.80u  0.10s  0.55g
3102              (spills = 42 reloads = 84)
3103           TOTAL                                   129.42u  5.10s  4.13g
3104    
3105    3000   CPS 100 spill                           675.59u  19.03s  11.64g
3106           MLRISC ra                               2.69u  0.27s  1.38g
3107              (spills = 62 reloads = 124)
3108           TOTAL                                   682.48u  19.61s  13.99g
3109    
3110    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
3111           MLRISC ra                               4.96u  0.27s  2.72g
3112              (spills = 85 reloads = 170)
3113           TOTAL                                   2375.26u  57.21s  48.00g
3114    
3115    As you can see the old cps spill module suffers from some serious
3116    performance problem.  But since I cannot decipher the old code fully,
3117    instead of patching the problems up, I'm reimplementing it
3118    with a different algorithm.  The new code is more modular,
3119    smaller when compiled, and substantially faster
3120    (O(n log n) time and O(n) space).  Timing of the new spill module:
3121    
3122    4000  CPS 100 spill                           0.02u  0.00s  0.00g
3123          MLRISC ra                               0.25u  0.02s  0.15g
3124             (spills=1 reloads=3)
3125          TOTAL                                   7.74u  0.34s  1.62g
3126    
3127    Implementation details:
3128    
3129    As far as I can tell, the purpose of the CPS spill module is to make sure the
3130    number of live variables at any program point (the bandwidth)
3131    does not exceed a certain limit, which is determined by the
3132    size of the spill area.
3133    
3134    When the bandwidth is too large, we decrease the register pressure by
3135    packing live variables into spill records.  How we achieve this is
3136    completely different than what we did in the old code.
3137    
3138    First, there is something about the MLRiscGen code generator
3139    that we should be aware of:
3140    
3141    o MLRiscGen performs code motion!
3142    
3143       In particular, it will move floating point computations and
3144       address computations involving only the heap pointer to
3145       their use sites (if there is only a single use).
3146       What this means is that if we have a CPS record construction
3147       statement
3148    
3149           RECORD(k,vl,w,e)
3150    
3151       we should never count the new record address w as live if w
3152       has only one use (which is often the case).
3153    
3154       We should do something similar to floating point, but the transformation
3155       there is much more complex, so I won't deal with that.
3156    
3157    Secondly, there are now two new cps primops at our disposal:
3158    
3159     1. rawrecord of record_kind option
3160        This pure operator allocates some uninitialized storage from the heap.
3161        There are two forms:
3162    
3163         rawrecord NONE [INT n]  allocates a tagless record of length n
3164         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
3165                                     and initializes the tag.
3166    
3167     2. rawupdate of cty
3168          rawupdate cty (v,i,x)
3169          Assigns to x to the ith component of record v.
3170          The storelist is not updated.
3171    
3172    We use these new primops for both spilling and increment record construction.
3173    
3174     1. Spilling.
3175    
3176        This is implemented with a linear scan algorithm (but generalized
3177        to trees).  The algorithm will create a single spill record at the
3178        beginning of the cps function and use rawupdate to spill to it,
3179        and SELECT or SELp to reload from it.  So both spills and reloads
3180        are fine-grain operations.  In contrast, in the old algorithm
3181        "spills" have to be bundled together in records.
3182    
3183        Ideally, we should sink the spill record construction to where
3184        it is needed.  We can even split the spill record into multiple ones
3185        at the places where they are needed.  But CPS is not a good
3186        representation for global code motion, so I'll keep it simple and
3187        am not attempting this.
3188    
3189     2. Incremental record construction (aka record splitting).
3190    
3191        Long records with many component values which are simulatenously live
3192        (recall that single use record addresses are not considered to
3193         be live) are constructed with rawrecord and rawupdate.
3194        We allocate space on the heap with rawrecord first, then gradually
3195        fill it in with rawupdate.  This is the technique suggested to me
3196        by Matthias.
3197    
3198        Some restrictions on when this is applicable:
3199        1. It is not a VECTOR record.  The code generator currently does not handle
3200           this case. VECTOR record uses double indirection like arrays.
3201        2. All the record component values are defined in the same "basic block"
3202           as the record constructor.  This is to prevent speculative
3203           record construction.
3204    
3205    ----------------------------------------------------------------------
3206    Name: Allen Leung
3207    Date: 2002/02/22 01:02:00 EST
3208    Tag: leunga-20020222-mlrisc-tools
3209    
3210    Minor bug fixes in the parser and rewriter
3211    
3212    ----------------------------------------------------------------------
3213    Name: Allen Leung
3214    Date: 2002/02/21 20:20:00 EST
3215    Tag: leunga-20020221-peephole
3216    
3217    Regenerated the peephole files.  Some contained typos in the specification
3218    and some didn't compile because of pretty printing bugs in the old version
3219    of 'nowhere'.
3220    
3221    ----------------------------------------------------------------------
3222    Name: Allen Leung
3223    Date: 2002/02/19 20:20:00 EST
3224    Tag: leunga-20020219-mlrisc-tools
3225    Description:
3226    
3227       Minor bug fixes to the mlrisc-tools library:
3228    
3229       1.  Fixed up parsing colon suffixed keywords
3230       2.  Added the ability to shut the error messages up
3231       3.  Reimplemented the pretty printer and fixed up/improved
3232           the pretty printing of handle and -> types.
3233       4.  Fixed up generation of literal symbols in the nowhere tool.
3234       5.  Added some SML keywords to to sml.sty
3235    
3236    ----------------------------------------------------------------------
3237    Name: Matthias Blume
3238    Date: 2002/02/19 16:20:00 EST
3239    Tag: blume-20020219-cmffi
3240    Description:
3241    
3242    A wild mix of changes, some minor, some major:
3243    
3244    * All C FFI-related libraries are now anchored under $c:
3245        $/c.cm      --> $c/c.cm
3246        $/c-int.cm  --> $c/internals/c-int.cm
3247        $/memory.cm --> $c/memory/memory.cm
3248    
3249    * "make" tool (in CM) now treats its argument pathname slightly
3250      differently:
3251        1. If the native expansion is an absolute name, then before invoking
3252           the "make" command on it, CM will apply OS.Path.mkRelative
3253           (with relativeTo = OS.FileSys.getDir()) to it.
3254        2. The argument will be passed through to subsequent phases of CM
3255           processing without "going native".  In particular, if the argument
3256           was an anchored path, then "make" will not lose track of that anchor.
3257    
3258    * Compiler backends now "know" their respective C calling conventions
3259      instead of having to be told about it by ml-nlffigen.  This relieves
3260      ml-nlffigen from one of its burdens.
3261    
3262    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
3263    
3264    * Export C_DEBUG and C_Debug from $c/c.cm.
3265    
3266    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
3267      subtyping relationship between incomplete pointers and their complete
3268      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
3269      with the convention of instantiating 'o with ('t, 'c) obj whenever
3270      the pointer target type is complete.  In the incomplete case, 'o
3271      will be instantiated with some "'c iobj" -- a type obtained by
3272      using one of the functors PointerToIncompleteType or PointerToCompleteType.
3273    
3274      Operations that work on both incomplete and complete pointer types are
3275      typed as taking an 'o ptr while operations that require the target to
3276      be known are typed as taking some ('t, 'c) obj ptr.
3277    
3278      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
3279      where void is an eqtype without any values.  This makes it possible
3280      to work on voidptr values using functions meant to operate on light
3281      incomplete pointers.
3282    
3283    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
3284      been vastly simplified.
3285    
3286    ----------------------------------------------------------------------
3287    Name: Matthias Blume
3288    Date: 2002/02/19 10:48:00 EST
3289    Tag: blume-20020219-pqfix
3290    Description:
3291    
3292    Applied Chris Okasaki's bug fix for priority queues.
3293    
3294    ----------------------------------------------------------------------
3295    Name: Matthias Blume
3296    Date: 2002/02/15 17:05:00
3297    Tag: Release_110_39
3298    Description:
3299    
3300    Last-minute retagging is becoming a tradition... :-(
3301    
3302    This is the working release 110.39.
3303    
3304    ----------------------------------------------------------------------
3305    Name: Matthias Blume
3306    Date: 2002/02/15 16:00:00 EST
3307    Tag: Release_110_39-orig
3308    Description:
3309    
3310    Working release 110.39.  New bootfiles.
3311    
3312    (Update: There was a small bug in the installer so it wouldn't work
3313    with all shells.  So I retagged. -Matthias)
3314    
3315    ----------------------------------------------------------------------
3316    Name: Matthias Blume
3317    Date: 2002/02/15 14:17:00 EST
3318    Tag: blume-20020215-showbindings
3319    Description:
3320    
3321    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
3322    the latter can be useful for exploring what bindings are available at
3323    the interactive prompt.  (The first function returns only the list
3324    of symbols that are really bound, the second prints those but also the
3325    ones that CM's autoloading mechanism knows about.)
3326    
3327    ----------------------------------------------------------------------
3328    Name: Matthias Blume
3329    Date: 2002/02/15 12:08:00 EST
3330    Tag: blume-20020215-iptrs
3331    Description:
3332    
3333    Two improvements to ml-nlffigen:
3334    
3335      1. Write files only if they do not exist or if their current contents
3336         do not coincide with what's being written.  (That is, avoid messing
3337         with the time stamps unless absolutely necessary.)
3338    
3339      2. Implement a "repository" mechanism for generated files related
3340         to "incomplete pointer types".   See the README file for details.
3341    
3342    ----------------------------------------------------------------------
3343    Name: Matthias Blume
3344    Date: 2002/02/14 11:50:00 EST
3345    Tag: blume-20020214-quote
3346    Description:
3347    
3348    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
3349    because of the new and improved tag generation scheme.  (Thanks to Allen
3350    Leung for pointing it out.)
3351    
3352    ----------------------------------------------------------------------
3353    Name: Lal George
3354    Date: 2002/02/14 09:55:27 EST 2002
3355    Tag: george-20020214-isabelle-bug
3356    Description:
3357    
3358    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
3359    of Isabelle on the x86.
3360    
3361    From Allen:
3362    -----------
3363     I've found the problem:
3364    
3365         in ra-core.sml, I use the counter "blocked" to keep track of the
3366         true number of elements in the freeze queue.  When the counter goes
3367         to zero, I skip examining the queue.  But I've messed up the
3368         bookkeeping in combine():
3369    
3370             else ();
3371             case !ucol of
3372               PSEUDO => (if !cntv > 0 then
3373                     (if !cntu > 0 then blocked := !blocked - 1 else ();
3374                                        ^^^^^^^^^^^^^^^^^^^^^^^
3375                      moveu := mergeMoveList(!movev, !moveu)
3376                     )
3377                  else ();
3378    
3379         combine() is called to coalesce two nodes u and v.
3380         I think I was thinking that if the move counts of u and v are both
3381         greater than zero then after they are coalesced then one node is
3382         removed from the freeze queue.  Apparently I was thinking that
3383         both u and v are of low degree, but that's clearly not necessarily true.
3384    
3385    
3386    02/12/2002:
3387        Here's the patch.  HOL now compiles.
3388    
3389        I don't know how this impact on performance (compile
3390        time or runtime).  This bug caused the RA (especially on the x86)
3391        to go thru the potential spill phase when there are still nodes on the
3392        freeze queue.
3393    
3394    
3395    
3396    
3397    ----------------------------------------------------------------------
3398    Name: Matthias Blume
3399    Date: 2002/02/13 22:40:00 EST
3400    Tag: blume-20020213-fptr-rtti
3401    Description:
3402    
3403    Fixed a bug in ml-nlffigen that was introduced with one of the previous
3404    updates.
3405    
3406    ----------------------------------------------------------------------
3407    Name: Matthias Blume
3408    Date: 2002/02/13 16:41:00 EST
3409    Tag: blume-20020213-cmlpq
3410    Description:
3411    
3412    Added new priority queue export symbols (which have just been added to
3413    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
3414    not compile and the installer would choke.)
3415    
3416    ----------------------------------------------------------------------
3417    Name: Matthias Blume
3418    Date: 2002/02/13 16:15:00 EST
3419    Tag: blume-20020213-various
3420    Description:
3421    
3422    1. More tweaks to ml-nlffigen:
3423    
3424       - better internal datastructures (resulting in slight speedup)
3425       - "-match" option requires exact match
3426       - "localized" gensym counters (untagged structs/unions nested within
3427         other structs/unions or within typedefs get a fresh counter; their
3428         tag will be prefixed by a concatenation of their parents' tags)
3429       - bug fixes (related to calculation of transitive closure of types
3430         to be included in the output)
3431    
3432    2. Minor Basis updates:
3433    
3434       - added implementations for List.collate and Option.app
3435    
3436    ----------------------------------------------------------------------
3437    Name: Matthias Blume
3438    Date: 2002/02/11 15:55:00 EST
3439    Tag: blume-20020211-gensym
3440    Description:
3441    
3442    Added a "-gensym" option to command line of ml-nlffigen.  This can be
3443    used to specify a "stem" -- a string that is inserted in all "gensym'd"
3444    names (ML structure names that correspond to unnamed C structs, unions,
3445    and enums), so that separate runs of ml-nlffigen do not clash.
3446    
3447    ----------------------------------------------------------------------
3448    Name: Matthias Blume
3449    Date: 2002/02/11 12:05:00 EST
3450    Tag: blume-20020211-gensml
3451    Description:
3452    
3453    A quick fix for a problem with GenSML (in the pgraph-util library):
3454    Make generation of toplevel "local" optional.  (Strictly speaking,
3455    signature definitions within "local" are not legal SML.)
3456    
3457    Other than that: updates to INSTALL and cm/TODO.
3458    
3459    ----------------------------------------------------------------------
3460    Name: Matthias Blume
3461    Date: 2002/02/08 15:00:00 EST
3462    Tag: blume-20020208-uniquepid
3463    Description:
3464    
3465    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
3466    
3467    1. The installer (config/install.sh) has gotten smarter:
3468    
3469         - Configuration options are a bit easier to specify now
3470           (in config/targets).
3471         - Bug in recognizing .tar.bz2 files fixed.
3472         - Installer automatically resolves dependencies between
3473           configuration options (e.g., if you ask for eXene, you will
3474           also get cml -- regardless whether you asked for it or not).
3475         - Installer can run in "quieter mode" by setting the environment
3476           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
3477           "completely silent", though.
3478         - Build HashCons library as part of smlnj-lib.
3479    
3480    2. A new scheme for assigning persistent identifiers to compilation
3481       units (and, by extension, to types etc.) has been put into place.
3482       This fixes a long-standing bug where types and even dynamic values
3483       can get internally confused, thereby compromising type safety
3484       (abstraction) and dynamic correctness.  See
3485    
3486         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
3487    
3488       for an example of how things could go wrong until now.
3489    
3490       The downside of the new scheme is that pids are not quite as
3491       persistent as they used to be: CM will generate a fresh pid
3492       for every compilation unit that it thinks it sees for the first
3493       time.  That means that if you compile starting from a clean, fresh
3494       source tree at two different times, you end up with different
3495       binaries.
3496    
3497       Cutoff recompilation, however, has not been compromised because
3498       CM keeps pid information in special caches between runs.
3499    
3500    ----------------------------------------------------------------------
3501    Name: Lal George
3502    Date: 2002/02/07 15:34:13 EST 2002
3503    Tag: <none>
3504    Description:
3505    
3506    Compilers that generate assembly code may produce  global labels
3507    whose value is resolved at link time. The various peephole optimization
3508    modules did not take this in account.
3509    
3510    TODO. The Labels.addrOf function should really return an option
3511    type so that clients are forced to deal with this issue, rather
3512    than an exception being raised.
3513    
3514    ----------------------------------------------------------------------
3515    Name: Lal George
3516    Date: 2002/02/06 13:55:02 EST
3517    Tag: george-20020206-ra-breakup
3518    Description:
3519    
3520    1. A bug fix from Allen.
3521    
3522        A typo causes extra fstp %st(0)'s to be generated at compensation
3523        edges, which might cause stack underflow traps at runtime.  This
3524        occurs in fft where there are extraneous fstps right before the 'into'
3525        trap instruction (in this case they are harmless since none of the
3526        integers overflow.)
3527    
3528    2. Pulled out various utility modules that were embedded in the modules
3529       of the register allocator. I need these modules for other purposes, but
3530       they are not complete enough to put into a library (just yet).
3531    ----------------------------------------------------------------------
3532    Name: Matthias Blume
3533    Date: 2002/01/31 16:05:00 EST
3534    Tag: blume-20020131-sparc-ccalls
3535    Description:
3536    
3537    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
3538       of extra stack space by mistake.  Fixed.
3539    
3540    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
3541    
3542    ----------------------------------------------------------------------
3543    Name: Allen Leung
3544    Date: 2002/01/30
3545    Tag: leunga-20020130-nowhere-bug-fix
3546    Description:
3547    
3548       MLRISC bug fixes:
3549       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
3550       2. MachineInt.fromString was negating its value.
3551    
3552    ----------------------------------------------------------------------
3553    Name: Matthias Blume
3554    Date: 2002/01/29
3555    Tag: blume-20020129-INSTALL
3556    Description:
3557    
3558    - Added somewhat detailed installation instructions (file INSTALL).
3559    - Fixed curl-detection bug in config/install.sh.
3560    - It is now possible to select the URL getter using the URLGETTER
3561      environment variable:
3562    
3563          not set / "unknown"      --> automatic detection (script tries wget,
3564                                       curl, and lynx)
3565          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
3566                                       how to properly invoke them)
3567          other                    --> use $URLGETTER directly, it must take
3568                                       precisely two command-line arguments
3569                                       (source URL and destination file name)
3570    
3571    ----------------------------------------------------------------------
3572    Name: Matthias Blume
3573    Date: 2002/01/28
3574    Tag: blume-20020128-sparc-ccalls
3575    Description:
3576    
3577    - Fixed problem with calculation of "used" registers in sparc-c-calls.
3578    - Make use of the allocParam argument in sparc-c-calls.
3579    
3580    ----------------------------------------------------------------------
3581    Name: Matthias Blume
3582    Date: 2002/01/28
3583    Tag: blume-20020128-allocParam
3584    Description:
3585    
3586    John Reppy:  Changes c-calls API to accept client-callback for
3587    allocating extra stack space.
3588    me: Corresponding changes to mlriscGen (using a dummy argument that
3589        does not change the current behavior).
3590    
3591    ----------------------------------------------------------------------
3592    Name: Matthias Blume
3593    Date: 2002/01/28 12:00:00
3594    Tag: Release_110_38
3595    Description:
3596    
3597    This time for real!!!
3598    
3599    ----------------------------------------------------------------------
3600    Name: Matthias Blume
3601    Date: 2002/01/28 10:56:00 EST
3602    Tag: blume-20020128-retraction
3603    Description:
3604    
3605    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
3606       with blume-Release_110_38-retracted.)
3607    
3608    1. Fixed a problem with incorrect rounding modes in real64.sml.
3609       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
3610    
3611    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
3612       and enums fixed.  The naming of corresponding ML identifiers should
3613       now be consistent again.
3614    
3615    ----------------------------------------------------------------------
3616    Name: Allen Leung
3617    Date: 2002/01/27
3618    Tag: leunga-20020127-nowhere
3619    Description:
3620    
3621       Added a target called nowhere in the configuration scripts.
3622       Enabling this will build the MLRISC 'nowhere' tool (for translating
3623       programs with where-clauses into legal SML code) during installation.
3624    
3625    ----------------------------------------------------------------------
3626    Name: Matthias Blume
3627    Date: 2002/01/25 21:27:00 EST
3628    Tag: blume-Release_110_38-retracted
3629    Description:
3630    
3631    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
3632    
3633    README will be added later.
3634    
3635    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
3636    (Release_110_38) removed.  Reason: Last-minute bug fixes.
3637    
3638    ----------------------------------------------------------------------
3639    Name: Matthias Blume
3640    Date: 2002/01/25
3641    Tag: blume-20020125-ffi
3642    Description:
3643    
3644    A large number of tweaks and improvements to ml-nlffi-lib and
3645    ml-nlffigen:
3646    
3647       - ML represenation types have been streamlined
3648       - getter and setter functions work with concrete values, not abstract
3649         ones where possible
3650       - ml-nlffigen command line more flexible (see README file there)
3651       - some bugs have been fixed (hopefully)
3652    
3653    ----------------------------------------------------------------------
3654    Name: Lal George
3655    Date: 2002/01/24
3656    Tag: george-20020124-risc-ra-interface
3657    Description:
3658    
3659       There is a dramatic simplification in the interface to the
3660       register allocator for RISC architectures as a result of making
3661       parallel copy instructions explicit.
3662    
3663    ----------------------------------------------------------------------
3664    Name: Matthias Blume
3665    Date: 2002/01/22
3666    Tag: blume-20020122-x86-ccalls
3667    Description:
3668    
3669    Bug fix for c-calls on x86 (having to do with how char- and
3670    short-arguments are being handled).
3671    
3672    ----------------------------------------------------------------------
3673    Name: Matthias Blume
3674    Date: 2002/01/21
3675    Tag: blume-20020121-ff
3676    Description:
3677    
3678    Another day of fiddling with the FFI...
3679    
3680    1. Bug fix/workaround:  CKIT does not complain about negative array
3681       dimensions, so ml-nlffigen has to guard itself against this possibility.
3682       (Otherwise a negative dimension would send it into an infinite loop.)
3683    
3684    2. Some of the abstract types (light objects, light pointers, most "base"
3685       types) in structure C are now eqtypes.
3686    
3687    3. Added constructors and test functions for NULL function pointers.
3688    
3689    ----------------------------------------------------------------------
3690    Name: Matthias Blume
3691    Date: 2002/01/18
3692    Tag: blume-20020118-ready-for-new-release
3693    Description:
3694    
3695    Made config/srcarchiveurl point to a new place.  (Will provide boot
3696    files shortly.)
3697    
3698    Maybe we christen this to be 110.38?
3699    
3700    ----------------------------------------------------------------------
3701    Name: Matthias Blume
3702    Date: 2002/01/18
3703    Tag: blume-20020118-more-ffifiddle
3704    Description:
3705    
3706    Today's FFI fiddling:
3707    
3708      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
3709        C.Set.* that use concrete (MLRep.*) arguments and results instead
3710        of abstract ones.
3711    
3712      - Provided word-style bit operations etc. for "int" representation
3713        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
3714        and Long.
3715    
3716    ----------------------------------------------------------------------
3717    Name: Matthias Blume
3718    Date: 2002/01/18
3719    Tag: blume-20020118-use-x86-fp
3720    Description:
3721    
3722    Now that x86-fast-fp seems to be working, I turned it back on again
3723    by default.  (Seems to work fine now, even with the FFI.)
3724    
3725    Other than that, I added some documentation about the FFI to
3726    src/ml-nlffigen/README and updated the FFI test examples in
3727    src/ml-nlffi-lib/Tests/*.
3728    
3729    ----------------------------------------------------------------------
3730    Name: Allen Leung
3731    Date: 2002/01/17
3732    Tag: leunga-20020117-x86-fast-fp-call
3733    Description:
3734    
3735       1. Fixed a problem with handling return fp values when x86's fast fp
3736          mode is turned on.
3737    
3738       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
3739          of %f32.
3740    
3741       3. Added a constructor INT32lit to the ast of MLRISC tools.
3742    
3743    ----------------------------------------------------------------------
3744    Name: Matthias Blume
3745    Date: 2002/01/16
3746    Tag: blume-20020116-ffifiddle
3747    Description:
3748    
3749    More fiddling with the FFI interface:
3750    
3751     - Make constness 'c instead of rw wherever possible.  This eliminates
3752       the need for certain explicit coercions.  (However, due to ML's
3753       value polymorphism, there will still be many cases where explicit
3754       coercions are necessary.  Phantom types are not the whole answer
3755       to modeling a subtyping relationship in ML.)
3756    
3757     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
3758    
3759     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
3760       clumsy workaround for varargs problem.)
3761    
3762    ----------------------------------------------------------------------
3763    Name: Lal George
3764    Date: 2002/01/15
3765    Tag: <none>
3766    Description:
3767    
3768    1. Since COPY instructions are no longer native to the architecture,
3769       a generic functor can be used to implement the expandCopies function.
3770    
3771    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
3772       TEXT segment.
3773    
3774    ----------------------------------------------------------------------
3775    Name: Matthias Blume
3776    Date: 2002/01/15
3777    Tag: blume-20020115-ffiupdates
3778    Description:
3779    
3780    1. Fix for bug resulting in single-precision float values being returned
3781       incorrectly from FFI calls.
3782    
3783    2. Small modifications to C FFI API:
3784    
3785        - memory-allocation routines return straight objects (no options)
3786          and raise an exception in out-of-memory situations
3787        - unsafe extensions to cast between function pointers and pointers
3788          from/to ints
3789        - added structure C_Debug as an alternative to structure C where
3790          pointer-dereferencing (|*| and |*!) always check for null-pointers
3791        - added open_lib' to DynLinkage;  open_lib' works like open_lib
3792          but also takes a (possibly empty) list of existing library handles
3793          that the current library depends on
3794    
3795    ----------------------------------------------------------------------
3796    Name: Matthias Blume
3797    Date: 2002/01/10
3798    Tag: blume-20020110-newffigen
3799    Description:
3800    
3801    1. Updates to portable graph code.
3802    
3803    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
3804       more scalable now so that even huge interfaces such as the one
3805       for GTK compile in finite time and space. :-)
3806       See src/ml-nlffigen/README for details on what's new.
3807    
3808    ----------------------------------------------------------------------
3809    Name: Lal George
3810    Date: 2001/01/09 14:31:35 EST 2002
3811    Tag: george-20011206-rm-native-copy
3812    Description:
3813    
3814            Removed the native COPY and FCOPY instructions
3815            from all the architectures and replaced it with the
3816            explicit COPY instruction from the previous commit.
3817    
3818            It is now possible to simplify many of the optimizations
3819            modules that manipulate copies. This has not been
3820            done in this change.
3821    
3822    ----------------------------------------------------------------------
3823    Name: Lal George
3824    Date: 2001/12/06 16:50:13 EST 2001
3825    Tag: george-20011206-mlrisc-instruction
3826    Description:
3827    
3828    Changed the representation of instructions from being fully abstract
3829    to being partially concrete. That is to say:
3830    
3831      from
3832            type instruction
3833    
3834      to
3835            type instr                              (* machine instruction *)
3836    
3837            datatype instruction =
3838                LIVE of {regs: C.cellset, spilled: C.cellset}
3839              | KILL of {regs: C.cellset, spilled: C.cellset}
3840              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
3841              | ANNOTATION of {i: instruction, a: Annotations.annotation}
3842              | INSTR of instr
3843    
3844    This makes the handling of certain special instructions that appear on
3845    all architectures easier and uniform.
3846    
3847    LIVE and KILL say that a list of registers are live or killed at the
3848    program point where they appear. No spill code is generated when an
3849    element of the 'regs' field is spilled, but the register is moved to
3850    the 'spilled' (which is present, more for debugging than anything else).
3851    
3852