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 716, Sat Nov 4 05:42:35 2000 UTC revision 1384, Thu Sep 18 20:35:34 2003 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: 2003/09/18 16:00:00 CDT
18    Tag: blume-20030918-110_43_1
19    Description:
20    
21    Made a new interim version and bootfiles for developer's bootstrapping
22    convenience.
23    
24       110.43.1 -- NEW BOOTFILES
25    
26    ----------------------------------------------------------------------
27    Name: Matthias Blume (blume (at) tti - c (dot) org)
28    Date: 2003/09/18 15:20:00 CDT
29    Tag: blume-20030918-misc
30    Description:
31    
32    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
33       this seems to be controversial at the moment) and used those in
34       Posix.ProcEnv.times.
35    
36    2. Added Time.{from,to}Nanoseconds to Time.
37    
38    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
39       For example, fromLargeInt never needs to look at more than 3 "big
40       digits" to get its 53 bits of precision.
41    
42    ----------------------------------------------------------------------
43    Name: Matthias Blume (blume (at) tti - c (dot) org)
44    Date: 2003/09/17 16:30:00 CDT
45    Tag: blume-20030917-real32-slices
46    Description:
47    
48    Added an entry to the primitive environment
49    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
50    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
51    
52    Removed some of the "magic" constants in real64.sml and replaced them
53    with code that generates these values from their corresponding
54    integer counterparts.
55    
56    Made all(?) the slice-related changes to the Basis and made everything
57    compile again...
58    
59    ----------------------------------------------------------------------
60    Name: Matthias Blume (blume (at) tti - c (dot) org)
61    Date: 2003/09/15 17:45:00 CDT
62    Tag: blume-20030915-rbase
63    Description:
64    
65    Fixed bug in Real.fromLargeInt.
66    
67    ----------------------------------------------------------------------
68    Name: Matthias Blume (blume (at) tti - c (dot) org)
69    Date: 2003/09/13 18:11:00 CDT
70    Tag: blume-20030913-libinstall
71    Description:
72    
73    Minor bugfix in config/libinstall (set anchor with path to
74    standalone tool after installing it, otherwise libraries that
75    need ml-lex or ml-yacc won't compile the first time the installer
76    runs).
77    
78    ----------------------------------------------------------------------
79    Name: Matthias Blume (blume (at) tti - c (dot) org)
80    Date: 2003/09/12 11:45:00 CDT
81    Tag: blume-20030912-various
82    Description:
83    
84     - fixed bug in Real.toLargeInt
85     - fixed bug in Posix.ProcEnv.times
86     - changed inputLine functions to return an option
87     - minor installer improvements / bugfixes
88     - changed default @SMLalloc parameter for x86/celeron to 64k
89    
90    ----------------------------------------------------------------------
91    Name: Matthias Blume (blume (at) tti - c (dot) org)
92    Date: 2003/09/09 22:00:00 CDT
93    Tag: Release_110_43
94    Description:
95    
96        New working release 110.43.  New bootfiles.
97    
98    ----------------------------------------------------------------------
99    Name: Matthias Blume (blume (at) tti - c (dot) org)
100    Date: 2003/09/09 19:20:00 CDT
101    Tag: blume-20030909-installer
102    Description:
103    
104    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
105    Modified config/install.bat to take advantage of it.  Also modified
106    config/install.sh (and called it config/new-install.sh) to take advantage
107    of it on Unix systems.  (The SML code is (supposed to be) platform-
108    independent.)
109    
110    The installer can now install everything under Win32
111    as well as under *nix as long as it compiles.
112    
113    Other changes:
114    
115      - made CML compile again under Win32
116      - made eXene compile under Win32 (by providing a fake structure UnixSock
117        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
118      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
119        same as type int;  under Win32 it isn't
120      - fixed some slice-related problems in the win32-specific parts of CML
121      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
122        (under Win32, the volume name is case-insensitive, and the
123        OS.Path code compares volume names for equality)
124    
125    ----------------------------------------------------------------------
126    Name: Matthias Blume (blume (at) tti - c (dot) org)
127    Date: 2003/09/08 11:55:00 CDT
128    Tag: blume-20030908-fullpath
129    Description:
130    
131    Made Win32 version of OS.FileSys.fullPath return current directory
132    when given an empty string.  This is what the spec says, and incidentally,
133    CM depends on it.  (CM otherwise goes into an infinite loop in certain
134    cases when presented with the name of a non-existing .cm file.)
135    
136    ----------------------------------------------------------------------
137    Name: Matthias Blume (blume (at) tti - c (dot) org)
138    Date: 2003/09/04 16:30:00 CDT
139    Tag: blume-20030905-slices-etc
140    Description:
141    
142      1. Changed interface to vectors and arrays in Basis to match
143         (draft) Basis spec.
144      2. Added signatures and implementations of slices according to
145         Basis spec.
146      3. Edited source code throughout the system to make it compile again
147         under 1. and 2.  (In some cases code had to be added to have it
148         match the new signatures.)
149      4. MLRISC should be backward-compatible: the copies of the originals
150         of files that needed to change under 3. were retained, the .cm files
151         check the compiler version number and use old versions when
152         appropriate.
153      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
154         dirstream -> string option (in accordance with Basis spec).
155      6. When generating code that counts lines, ml-lex used function
156         CharVector.foldli, taking advantage of its old interface.
157         This has been replaced with the corresponding code from
158         CharVectorSlice.  (html-lex must be re-lexed!)
159      7. BitArray in smlnj-lib/Util has been extended/modified to match the
160         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
161      8. Removed temporary additions (fromInternal, toInternal) from the
162         (now obsolete) IntInf in smlnj-lib/Util.
163      9. Cleaned up structure Byte.
164     10. Added localOffset, scan, and fromString to Date (according to spec).
165         Cleaned/corrected implementation of Date.
166         (Still need to check for correctness; implement better canonicalizeDate.)
167     11. Added "scan" to signature IEEE_REAL.
168     12. Some improvements to IntInf [in particular: efficiency-hack for
169         mod and rem when second operand is 2 (for parity checks).]
170     13. Changed representation of type Time.time, using a single IntInf.int
171         value counting microseconds.  This considerably simplified the
172         implementation of structure Time.  We now support negative time
173         values; scan and fromString handle signs.
174     14. Functor PrimIO now takes two additional arguments (VectorSlice and
175         ArraySlice).
176    
177    ----------------------------------------------------------------------
178    Name: Matthias Blume (blume (at) tti - c (dot) org)
179    Date: 2003/08/28 17:00:00 CDT
180    Tag: blume-20030828-intinf
181    Description:
182    
183    This is a major update which comes with a version number bump
184    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
185    and an implementation of IntInf in the Basis.
186    
187    There are a fairly large number of related changes and updates throughout
188    the system:
189    
190    Basis:
191            - Implemented IntInf.
192            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
193            - Added some missing Real64 operations, most notably Real.toLargeInt.
194            - Added FixedInt as a synonym for Int32.
195    
196    compiler:
197            * Added support for a built-in intinf type.
198              - literals
199              - pattern matching
200              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
201              - overloading on literals and operations
202    
203            This required adding a primitive type intinf, some additional
204            primops, and implementations for several non-trivial intinf
205            operations in Core.  (The intinf type is completely abstract
206            to the compiler; all operations get delegated back to the Core.)
207    
208            * Intinf equality is handled by polyequal.  However, the compiler
209            does not print its usual warning in this case (since polyequal
210            is the right thing to do there).
211    
212            * Improved the organization of structure InlineT.
213    
214            * A word about conversion primops:
215            If conversions involving intinf do not cancel out during
216            CPS contract, then the compiler must insert calls to Core functions.
217            Since all core access must be resolved already during the FLINT
218            translate phase, it would be too late a the time of CPS contract
219            to add new Core calls.  For this reason, conversion primops
220            for intinf carry two arguments:  1. the numeric argument that
221            they are supposed to convert, and 2. the Core function that
222            can help with this conversion if necessary.  If CPS contract
223            eliminates a primop, then the associated Core function becomes
224            dead and goes away.  Intinf conversion primops that do not get
225            eliminated by CPS contract get rewritten into calls of their
226            core functions by a separate, new phase.
227    
228    interactive system:
229            - Control.Print.intinfDepth controls max length of intinf constants
230            being printed. (Analogous to Control.Print.stringDepth.)
231            - Cleanup in printutil and pputil: got rid of unused stuff and
232            duplicates; replaced some of the code with code that makes better
233            use of library functionality.
234    
235    CM:
236            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
237            are no longer silent.
238    
239    CKIT:
240            Fixed mismatched uses of Int32 and LargeInt.  I always decided
241            in favor of LargeInt -- which is now the same as IntInf.
242            CKIT-knowledgable people should check whether this is what's
243            intended and otherwise change things back to using Int32 or
244            FixedInt.
245    
246    Throughout the code:
247            Started using IntInf.int literals and built-in operations
248            (e.g., comparison with 0) where this seems appropriate.
249    
250    
251    ----------------------------------------------------------------------
252    Name: Dave MacQueen (dbm@cs.uchicago.edu)
253    Date: 2003/08/13 11:36:00 CDT
254    Tag: dbm-20030813-mcz-merge1
255    Description:
256    
257    Merging changes from the mcz-branch development branch into trunk.
258    These changes involve replacement of the emulated old prettyprinter
259    interface with direct use of the SML/NJ Lib PP library, and fixing
260    of a couple of bugs (895, 1186) relating to error messages.  A new
261    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
262    has been added.
263    
264    ----------------------------------------------------------------------
265    Name: Matthias Blume (blume (at) tti - c (dot) org)
266    Date: 2003/08/11 15:45:00 CDT
267    Tag: blume-20030811-windows
268    Description:
269    
270    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
271    
272        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
273    
274    This patch restores SML/NJ's ability to run under win32.  There are a
275    number of changes, including fixes for several bugs that had gone
276    unnoticed until now:
277    
278      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
279        trying to run the win32 version from within cygwin.)
280    
281      - There are a number of simple .bat scripts that substitute for their
282        corresponding Unix shell-scripts.  (See below.)
283    
284      - The internals of ml-build have been modified slightly.  The main
285        difference is that instead of calling ".link-sml" (or link-sml.bat)
286        using OS.Process.system, the ML process delegates this task back
287        to the script.  Otherwise problems arise in mixed environments such
288        as Cygwin where scripts look and work like Unix scripts, but
289        where OS.Process.system cannot run them.
290    
291      - In CM, the srcpath pickler used native pathname syntax -- which
292        is incorrect in the case of cross-compilation.  The new pickle format
293        is independent of platform-specific naming conventions.
294    
295      - Path configuration files (such as lib/pathconfig) can now choose
296        between native and standard syntax.  Placing a line of the form
297    
298            standard!
299    
300        into the file causes all subsequent paths to be interpreted using
301        CM standard pathname syntax (= Unix conventions); a line
302    
303            native!
304    
305        switches back to native style.  This was needed so that
306        path config files can be written portably, see src/system/pathconfig.
307    
308      - Runtime system:
309    
310          - win32-filesys.c:  get_file_time and set_file_time now
311            access modification time, not creation time.
312    
313          - I/O code made aware of new array representation.
314    
315          - Bug fixes in X86.prim.masm.
316    
317       - src/system/makeml made aware of win32. (For use under cygwin
318            and other Unix-environments for windows.)
319    
320       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
321         which caused BinIO.inputAll to fail consistently.
322    
323    .bat scripts:
324    
325       Windows .bat scripts assume that SMLNJ_HOME is defined.
326    
327       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
328         applications (sml, ml-yacc, ml-lex).
329       - ml-build.bat: analogous to ml-build.
330       - config\install.bat:  Analogous to config/install.sh.  This requires
331         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
332         (nmake etc. must be on the path, and vcvars32 must have been run.)
333         Moreover, sources for ml-lex and ml-yacc need to exist under src,
334         and the bootfile hierarchy must have been unpacked under
335         sml.boot.x86-win32.
336         The script is very primitive and does a poor job at error checking.
337         It only installs the base system, ml-lex, and ml-yacc.  No other
338         libraries are being installed (i.e., you get only those that
339         are part of the compiler.)
340       - link-sml.bat:  analogous to .link-sml, but not currently used
341    
342    Unrelated bug fixes:
343    
344      - ml-nlffigen now exports structures ST_* corresponding to incomplete
345        types.
346      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
347        otherwise.)
348    
349    ----------------------------------------------------------------------
350    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
351    Date: 2003/06/17
352    Tag: macqueen-20030617-bug895
353    Description:
354    
355    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
356    Tag will be used for new development branch (mcz-branch) for
357    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
358    
359    ----------------------------------------------------------------------
360    Name: Matthias Blume (blume (at) tti - c (dot) org)
361    Date: 2003/05/27 16:55:00 CDT
362    Tag: blume-20030527-polyeq
363    Description:
364    
365    Tried to eliminated most cases of polymorphic equality.
366    
367    ----------------------------------------------------------------------
368    Name: Matthias Blume (blume (at) tti - c (dot) org)
369    Date: 2003/05/21 17:45:00 CDT
370    Tag: blume-20030517-complete
371    Description:
372    
373    Two changes:
374    
375      1. Added a flag for controlling whether non-exhaustive bindings will
376         be treated as errors (default is false).
377      2. Cleaned up the *entire* source tree so that CMB.make goes through
378         without a single non-exhaustive match- or bind warning.
379    
380    ----------------------------------------------------------------------
381    Name: Matthias Blume (blume (at) tti - c (dot) org)
382    Date: 2003/05/17 10:20:00 CDT
383    Tag: blume-20030517-absyn
384    Description:
385    
386    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
387    
388       This mainly affects the quality of error messages.  However, some
389       of the code is now more straightforward than before. (Treatment of
390       the above four constructs in translate.sml is much simpler than
391       the "macro-expansion" that was going on before.  Plus, the mach-
392       compiler no longer gets invoked just to be able to compile an
393       if-expression.)
394    
395    2. The ErrorMsg.Error exception is now caught and absorbed by the
396       interactive loop.
397    
398    ----------------------------------------------------------------------
399    Name: Allen Leung
400    Date: 2003/05/16 13:05:00 CDT
401    Tag: leunga-20030516-cygwin-runtime
402    Description:
403    
404        Ported the runtime system to cygwin, which uses the unix
405    x86-unix bin files.  Missing/buggy features:
406    
407       o getnetbyname, getnetbyaddr: these functions seem to be missing in
408         the Cygwin library.
409       o Ctrl-C handling may be flaky.
410       o Windows system calls and Windows I/O are not supported.
411    
412       A new set of binfiles is located at:
413    
414         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
415    
416       This is only needed for bootstrapping the cygwin version of smlnj.
417       Other x86 versions can use the existing binfiles.
418    
419    ----------------------------------------------------------------------
420    Name: Matthias Blume
421    Date: 2003/04/08 15:42:00 CDT
422    Tag: blume-20030408-listpair
423    Description:
424    
425    1. Added a target 'mlrisc' to installer.
426    
427    2. Added missing elements to structure ListPair.
428    
429    ----------------------------------------------------------------------
430    Name: Allen Leung
431    Date: 2003/01/07 10:40:00 EST
432    Tag: leunga-20030107-int-rem
433    Description:
434    
435    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
436    The arguments to the SUBL instruction were swapped.
437    
438    ----------------------------------------------------------------------
439    Name: Matthias Blume
440    Date: 2002/12/12 16:25:00 EST
441    Tag: blume-20021212-risc-ra
442    Description:
443    
444    Fixed a serious bug in the rewrite code for FP spilling/reloading that
445    sent the RA into an infinite loop when floating point registers get
446    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
447    110.37 and 110.38.)
448    There was another set of potential problems related to the handling of
449    MLRISC annotations (but those did not yet cause real problems, apparently).
450    
451    ----------------------------------------------------------------------
452    Name: Matthias Blume
453    Date: 2002/12/06 22:40:00 EST
454    Tag: blume-20021206-cm-fileid
455    Description:
456    
457    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
458    This fixes the problem of CM getting confused by files that suddenly
459    change their identity (e.g., by getting unlinked and recreated by some
460    text editor such as vi).  There might be a better/cheaper/cleaner way
461    of doing this, but for now this will have to do.
462    
463    ----------------------------------------------------------------------
464    Name: Matthias Blume
465    Date: 2002/10/28 09:50:00 EST
466    Tag: blume-20021028-typecheck
467    Description:
468    
469    Exported structure Typecheck from $smlnj/viscomp/core.cm.
470    
471    ----------------------------------------------------------------------
472    Name: Matthias Blume
473    Date: 2002/10/17 09:10:00 EDT
474    Tag: Release_110_42
475    Description:
476    
477    In good old tradition, there has been a slight hiccup so that we have
478    to patch 110.42 after the fact.   The old release tag has been replaced
479    (see below).
480    
481    The change solves a problem with two competing approaches the
482    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
483    each other's way.
484    
485    This change only affects the runtime system code and the installer script.
486    (No new bootfiles.)
487    
488    ----------------------------------------------------------------------
489    Name: Matthias Blume
490    Date: 2002/10/16 12:00:00 EDT
491    Tag: Release_110_42_removed
492    Description:
493    
494    New working release.  New bootfiles.
495    
496    ----------------------------------------------------------------------
497    Name: Matthias Blume
498    Date: 2002/10/10 13:10:00 EDT
499    Tag: blume-20021010-ppc-divs
500    Description:
501    
502    The mltree operator DIVS must be implemented with an overflow check on
503    the PPC because the hardware indicates divide-by-zero using "overflow" as
504    well.
505    
506    ----------------------------------------------------------------------
507    Name: Matthias Blume
508    Date: 2002/07/23 11:45:00
509    Tag: blume-20020723-smlnj-home
510    Description:
511    
512    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
513    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
514    CM_PATHCONFIG is also set), the path configuration file is assumed
515    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
516    the entire tree to some other place and everything will "just work".
517    
518    (Companion commands such as ml-build and ml-makedepend also sense this
519    variable.)
520    
521    ----------------------------------------------------------------------
522    Name: Matthias Blume
523    Date: 2002/07/12 21:19:00 EDT
524    Tag: blume-20020712-liveness
525    Description:
526    
527    Exported two useful "step" functions from liveness module (MLRISC).
528    
529    ----------------------------------------------------------------------
530    Name: Matthias Blume
531    Date: 2002/07/05 16:00 EDT
532    Tag: Release_110_41
533    Description:
534    
535    New working release.  New bootfiles.
536    
537    ----------------------------------------------------------------------
538    Name: Matthias Blume
539    Date: 2002/07/05 10:25:00 EDT
540    Tag: blume-20020705-btimp
541    Description:
542    
543    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
544    other clients can set up backtracing support.
545    
546    ----------------------------------------------------------------------
547    Name: Matthias Blume
548    Date: 2002/06/25 17:23:00 EDT
549    Tag: blume-20020625-fpmax
550    Description:
551    
552    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
553    case.  (The sense of the isNaN test was reversed -- which made min and
554    max always return their first argument.)
555    
556    ----------------------------------------------------------------------
557    Name: Matthias Blume
558    Date: 2002/06/11
559    Tag: blume-20020611-unixpath
560    Description:
561    
562    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
563    
564    ----------------------------------------------------------------------
565    Name: Matthias Blume
566    Date: 2002/06/10 16:35:00 EDT
567    Tag: blume-20020610-ieeereal
568    Description:
569    
570    I back-ported my implementation of IEEEReal.fromString from the
571    idlbasis-devel branch so that we can test it.
572    
573    Another small change is that ppDec tries to give more information
574    than just "<sig>" in the case of functors.  However, this code is
575    broken in some mysterious way if the functor's body's signature
576    has not been declared by ascription but gets inferred from the
577    implementation.  This needs fixing...
578    
579    ----------------------------------------------------------------------
580    Name: Matthias Blume
581    Date: 2002/05/31
582    Tag: blume-20020531-btrace-mode
583    Description:
584    
585    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
586    the wayside when I switched over to using Controls everywhere.)
587    
588    ----------------------------------------------------------------------
589    Name: Lal George
590    Date: 2002/05/23 12:21:40 EDT
591    Tag: george-20020523-visual-labels
592    Description:
593    
594    Labels are now displayed in the graphical output to make
595    the fall-through and target blocks obvious.
596    
597    ----------------------------------------------------------------------
598    Name: Matthias Blume
599    Date: 2002/05/22 11:03:00 EDT
600    Tag: blume-20020522-shrink
601    Description:
602    
603    John tweaked yesterday's fix for 1131 to handle an out-of-memory
604    situation that comes up when allocating huge arrays.
605    
606    ----------------------------------------------------------------------
607    Name: Matthias Blume
608    Date: 2002/05/21 16:00:00 EDT
609    Tag: Release_110_40
610    Description:
611    
612    New working release (110.40).  New bootfiles.
613    
614    [Also: John Reppy fixed GC bug 1131.]
615    
616    ----------------------------------------------------------------------
617    Name: Matthias Blume
618    Date: 2002/05/21 12:35:00 EDT
619    Tag: blume-20020521-cmdoc
620    Description:
621    
622    CM documentation update.
623    
624    ----------------------------------------------------------------------
625    Name: Matthias Blume
626    Date: 2002/05/21 10:55:00 EDT
627    Tag: blume-20020521-misc
628    Description:
629    
630    - John tweaked runtime to be silent on heap export (except when
631      GC messages are on).
632    - I added a few more things (cross-compiling versions of CMB) to
633      config/preloads (as suggestions).
634    
635    ----------------------------------------------------------------------
636    Name: Matthias Blume
637    Date: 2002/05/20 22:25:00 EDT
638    Tag: blume-20020520-controls
639    Description:
640    
641    - Added ControlUtil structure to control-lib.cm.
642    - Use it throughout.
643    - Used Controls facility to define MLRISC controls (as opposed to
644      registering MLRISC control ref cells with Controls after the fact)
645    - Fixed messed-up controls priorities.
646    
647    * Removed again all the stuff from config/preloads that one wouldn't
648      be able to preload at the time the initial heap image is built.
649      (Many libraries, e.g., CML, do not exist yet at this time.  The
650       only libraries that can be preloaded via config/preloads are those
651       that come bundled with the bootfiles.)
652    
653    ----------------------------------------------------------------------
654    Name: Matthias Blume
655    Date: 2002/05/20 10:59:00 EDT
656    Tag: blume-20020520-preloads
657    Description:
658    
659    Added a lot of commented-out suggestions for things to be included
660    in config/preloads.
661    
662    ----------------------------------------------------------------------
663    Name: Allen Leung
664    Date: 2002/05/18 14:20:00 EDT
665    Tag: leunga-20020518-mdl
666    Description:
667    
668    o Made the mdl tool stuff compile and run again.
669    o I've disabled all the stuff that depends on RTL specifications; they
670      are all badly broken anyway.
671    
672    ----------------------------------------------------------------------
673    Name: Matthias Blume
674    Date: 2002/05/17 16:49:00 EDT
675    Tag: blume-20020517-controls
676    Description:
677    
678    0. John Reppy made several modifications to the SML/NJ library.
679       In particular, there is a shiny new controls-lib.cm.
680    
681    1. Pushed new controls interface through compiler so that everything
682       compiles again.
683    
684    2. Added FormatComb and FORMAT_COMB to the CML version of the
685       SML/NJ library (so that CML compiles again).
686    
687    3. Modified init scripts because XXX_DEFAULT environment variables
688       are no longer with us.  (Boot-time initialization is now done
689       using the same environment variables that are also used for
690       startup-time initialization of controls.)
691    
692    ----------------------------------------------------------------------
693    Name: Lal George
694    Date: 2002/05/15 09:20:10 EDT
695    Tag: george-20020515-pseudo-op-decls
696    Description:
697    
698            All pseudo-ops emitted before the first segment declaration
699    such as TEXT, DATA, and BSS directives are assumed to be global
700    declarations and are emitted first in the assembly file. This is
701    useful in a number of situations where one has pseudo-ops that are not
702    specific to any segment, and also works around the constraint that one
703    cannot have client pseudo-ops in the TEXT segment.
704    
705            Because no segment is associated with these declarations it is
706    an error to allocate any space or objects before the first segment
707    directive and an exception will be raised. However, we cannot make
708    this check for client pseudo-ops.
709    
710            These top level declarations are a field in the CFG graph_info.
711    In theory you can continue to add to this field after the CFG has been
712    built -- provided you know what you are doing;-)
713    
714    ----------------------------------------------------------------------
715    Name: Matthias Blume
716    Date: 2002/05/13 16:40:00 EDT
717    Tag: blume-20020513-pp-etc
718    Description:
719    
720    A few minor bugfixes:
721    
722      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
723        (Bogus printouts for redefined bindings still occur.  Compiler
724        bug should no longer occur now.  We need to redo the prettyprinter
725        from scratch.)
726    
727      - CM pathname printer now also adds escape sequences for ( and )
728    
729      - commend and docu fixes for ml-nlffi
730    
731    ----------------------------------------------------------------------
732    Name: Matthias Blume
733    Date: 2002/05/10 16:40:00 EDT
734    Tag: blume-20020510-erg-textio
735    Description:
736    
737    Applied the following bugfix provided by Emden Gansner:
738    
739        Output is corrupted when outputSubstr is used rather than output.
740    
741        The problem occurs when a substring
742    
743            ss = (s, dataStart, dataLen)
744    
745        where dataStart > 0, fills a stream buffer with avail bytes left.
746        avail bytes of s, starting at index dataStart, are copied into the
747        buffer, the buffer is flushed, and then the remaining dataLen-avail
748        bytes of ss are copied into the beginning of the buffer. Instead of
749        starting this copy at index dataStart+avail in s, the current code
750        starts the copy at index avail.
751    
752        Fix:
753        In text-io-fn.sml, change line 695 from
754             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
755        to
756             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
757    
758    ----------------------------------------------------------------------
759    Name: Matthias Blume
760    Date: 2002/04/12 13:55:00 EDT
761    Tag: blume-20020412-assyntax
762    Description:
763    
764    1. Grabbed newer assyntax.h from the XFree86 project.
765    2. Fiddled with how to compile X86.prim.asm without warnings.
766    3. (Very) Minor cleanup in CM.
767    
768    ----------------------------------------------------------------------
769    Name: Matthias Blume
770    Date: 2002/04/01 (no joke!) 17:07:00 EST
771    Tag: blume-20020401-x86div
772    Description:
773    
774    Added full support for div/mod/rem/quot on the x86, using the machine
775    instruction's two results (without clumsily recomputing the remainder)
776    directly where appropriate.
777    
778    Some more extensive power-of-two support was added to the x86 instruction
779    selector (avoiding expensive divs, mods, and muls where they can be
780    replaced with cheaper shifts and masks).  However, this sort of thing
781    ought to be done earlier, e.g., within the CPS optimizer so that
782    all architectures benefit from it.
783    
784    The compiler compiles to a fixed point, but changes might be somewhat
785    fragile nevertheless.  Please, report any strange things that you might
786    see wrt. div/mod/quot/rem...
787    
788    ----------------------------------------------------------------------
789    Name: Matthias Blume
790    Date: 2002/03/29 17:22:00
791    Tag: blume-20020329-div
792    Description:
793    
794    Fixed my broken div/mod logic.  Unfortunately, this means that the
795    inline code for div/mod now has one more comparison than before.
796    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
797    The problem was with quotient = 0, because that alone does not tell
798    us which way the rounding went.  One then has to look at whether
799    remainder and divisor have the same sign...  :(
800    
801    Anyway, I replaced the bootfiles with fresh ones...
802    
803    ----------------------------------------------------------------------
804    Name: Matthias Blume
805    Date: 2002/03/29 14:10:00 EST
806    Tag: blume-20020329-inlprims
807    Description:
808    
809    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
810    
811    Primops have changed. This means that the bin/boot-file formats have
812    changed as well.
813    
814    To make sure that there is no confusion, I made a new version.
815    
816    
817    CHANGES:
818    
819    * removed REMT from mltree (remainder should never overflow).
820    
821    * added primops to deal with divisions of all flavors to the frontend
822    
823    * handled these primops all the way through so they map to their respective
824      MLRISC support
825    
826    * used these primops in the implementation of Int, Int32, Word, Word32
827    
828    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
829    
830    * parameterized INLMIN, INLMAX, and INLABS by a numkind
831    
832    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
833      floating point
834    
835    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
836      and Real (but Real.abs maps to a separate floating-point-only primop)
837    
838    
839    TODO items:
840    
841    * Hacked Alpha32 instruction selection, disabling the selection of REMx
842      instructions because the machine instruction encoder cannot handle
843      them.  (Hppa, PPC, and Sparc instruction selection did not handle
844      REM in the first place, and REM is supported by the x86 machine coder.)
845    
846    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
847      selection phase.  (The two can be streamlined because the hardware
848      delivers both quotient and remainder at the same time anyway.)
849    
850    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
851      (Currently the behavior is inconsistent both across architectures and
852      wrt. the draft Basis spec.)
853    
854    * Word8 should eventually be handled natively, too.
855    
856    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
857      as if there currently is no execution path that could trigger it in
858      SML/NJ.  (The assumptions underlying functions arith and promotable do not
859      hold for things like multiplication and division.)
860    
861    ----------------------------------------------------------------------
862    Name: Matthias Blume
863    Date: 2002/03/27 16:27:00 EST
864    Tag: blume-20020327-mlrisc-divisions
865    Description:
866    
867    Added support for all four division operations (ML's div, mod, quot,
868    and rem) to MLRISC.  In the course of doing so, I also rationalized
869    the naming (no more annoying switch-around of DIV and QUOT), by
870    parameterizing the operation by div_rounding_mode (which can be either
871    DIV_TO_ZERO or DIV_TO_NEGINF).
872    
873    The generic MLTreeGen functor takes care of compiling all four
874    operations down to only round-to-zero div.
875    
876    Missing pieces:
877    
878      * Doing something smarter than relying on MLTreeGen on architectures
879        like, e.g., the x86 where hardware division delivers both quotient and
880        remainder at the same time.  With this, the implementation of the
881        round-to-neginf operations could be further streamlined.
882    
883      * Remove inlining support for div/mod/rem from the frontend and replace it
884        with primops that get carried through to the backend.  Do this for all
885        int and word types.
886    
887    ----------------------------------------------------------------------
888    Name: Matthias Blume
889    Date: 2002/03/25 17:25:00 EST
890    Tag: blume-20020325-divmod
891    Description:
892    
893    I improved (hopefully without breaking them) the implementation of Int.div,
894    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
895    advantage of the following observations:
896    
897      Let  q = x quot y      r = x rem y
898           d = x div  y      m = x mod y
899    
900    where "quot" is the round-to-zero version of integer division that
901    hardware usually provides.  Then we have:
902    
903         r = x - q * y        where neither the * nor the - will overflow
904         d = if q >= 0 orelse x = q * y then q else q - 1
905                              where neither the * nor the - will overflow
906         m = if q >= 0 orelse r = 0 then r else r + y
907                              where the + will not overflow
908    
909    This results in substantial simplification of the generated code.
910    The following table shows the number of CFG nodes and edges generated
911    for
912            fun f (x, y) = x OPER y
913            (* with OPER \in div, mod, quot, rem *)
914    
915    
916        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
917        --------------------------------------------------------
918         div |         24 |         39 |         12 |         16
919         mod |         41 |         71 |         12 |         16
920        quot |          8 |         10 |          8 |         10
921         rem |         10 |         14 |          8 |         10
922    
923    
924    ----------------------------------------------------------------------
925    Name: Matthias Blume
926    Date: 2002/03/25 22:06:00 EST
927    Tag: blume-20020325-cprotobug
928    Description:
929    
930    Fixed a bug in cproto (c prototype decoder).
931    
932    ----------------------------------------------------------------------
933    Name: Matthias Blume
934    Date: 2002/03/25 16:00:00 EST
935    Tag: blume-20020325-raw-primops
936    Description:
937    
938    I did some cleanup to Allen's new primop code and
939    replaced yesterday's bootfiles with new ones.
940    (But they are stored in the same place.)
941    
942    ----------------------------------------------------------------------
943    Name: Matthias Blume
944    Date: 2002/03/24 22:40:00 EST
945    Tag: blume-20020324-bootfiles
946    Description:
947    
948    Made the bootfiles that Allen asked for.
949    
950    ----------------------------------------------------------------------
951    Name: Allen Leung
952    Date: 2002/03/23 15:50:00 EST
953    Tag: leunga-20020323-flint-cps-rcc-primops
954    Description:
955    
956      1. Changes to FLINT primops:
957    
958        (* make a call to a C-function;
959         * The primop carries C function prototype information and specifies
960         * which of its (ML-) arguments are floating point. C prototype
961         * information is for use by the backend, ML information is for
962         * use by the CPS converter. *)
963      | RAW_CCALL of { c_proto: CTypes.c_proto,
964                       ml_args: ccall_type list,
965                       ml_res_opt: ccall_type option,
966                       reentrant : bool
967                     } option
968       (* Allocate uninitialized storage on the heap.
969        * The record is meant to hold short-lived C objects, i.e., they
970        * are not ML pointers.  With the tag, the representation is
971        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
972        *)
973      | RAW_RECORD of {tag:bool,sz:int}
974      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
975    
976      2.  These CPS primops are now overloaded:
977    
978           rawload of {kind:numkind}
979           rawstore of {kind:numkind}
980    
981          The one argument form is:
982    
983             rawload {kind} address
984    
985          The two argument form is:
986    
987             rawload {kind} [ml object, byte-offset]
988    
989      3. RAW_CCALL/RCC now takes two extra arguments:
990    
991         a. The first is whether the C call is reentrant, i.e., whether
992            ML state should be saved and restored.
993         b. The second argument is a string argument specifying the name of
994            library and the C function.
995    
996         These things are currently not handled in the code generator, yet.
997    
998      4. In CProto,
999    
1000         An encoding type of "bool" means "ml object" and is mapped into
1001         C prototype of PTR.  Note that "bool" is different than "string",
1002         even though "string" is also mapped into PTR, because "bool"
1003         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1004    
1005      5. Pickler/unpicker
1006    
1007         Changed to handle RAW_RECORD and newest RAW_CCALL
1008    
1009      6. MLRiscGen,
1010    
1011         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1012         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1013            in the file CodeGen/cpscompile/cps-c-calls.sml
1014    
1015      7. Added the conditional move operator
1016    
1017             condmove of branch
1018    
1019         to cps.  Generation of this is still buggy so it is currently
1020         disabled.
1021    
1022    ----------------------------------------------------------------------
1023    Name: Lal George
1024    Date: 2002/03/22 14:18:25 EST
1025    Tag: george-20020322-cps-branch-prob
1026    Description:
1027    
1028    Implemented the Ball-Larus branch prediction-heuristics, and
1029    incorporated graphical viewers for control flow graphs.
1030    
1031    Ball-Larus Heuristics:
1032    ---------------------
1033    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1034    
1035    By design it uses the Dempster-Shafer theory for combining
1036    probabilities.  For example, in the function:
1037    
1038        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1039    
1040    the ball-larus heuristics predicts that the n=0 is unlikely
1041    (OH-heuristic), and the 'then' branch is unlikely because of the
1042    RH-heuristic -- giving the 'then' branch an even lower combined
1043    probability using the Dempster-Shafer theory.
1044    
1045    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1046    probability of the 'then' branch because of the loop in the else
1047    branch.
1048    
1049    
1050    Graphical Viewing:
1051    ------------------
1052    I merely plugged in Allen's graphical viewers into the compiler. The
1053    additional code is not much. At the top level, saying:
1054    
1055            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1056    
1057    will display the graphical view of the control flow graph just before
1058    back-patching.  daVinci must be in your path for this to work. If
1059    daVinci is not available, then the default viewer can be changed
1060    using:
1061    
1062            Control.MLRISC.getString "viewer"
1063    
1064    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1065    course, these viewers must be in your path.
1066    
1067    The above will display the compilation unit at the level of clusters,
1068    many of which are small, boring, and un-interesting. Also setting:
1069    
1070            Control.MLRISC.getInt "cfg-graphical-view_size"
1071    
1072    will display clusters that are larger than the value set by the above.
1073    
1074    
1075    ----------------------------------------------------------------------
1076    Name: Matthias Blume
1077    Date: 2002/03/21 22:20:00 EST
1078    Tag: blume-20020321-kmp-bugfix
1079    Description:
1080    
1081    Changed the interface to the KMP routine in PreString and fixed
1082    a minor bug in one place where it was used.
1083    
1084    ----------------------------------------------------------------------
1085    Name: Allen Leung
1086    Date: 2002/03/21 20:30:00 EST
1087    Tag: leunga-20020321-cfg
1088    Description:
1089    
1090      Fixed a potential problem in cfg edge splitting.
1091    
1092    ----------------------------------------------------------------------
1093    Name: Allen Leung
1094    Date: 2002/03/21 17:15:00 EST
1095    Tag: leunga-20020321-x86-fp-cfg
1096    Description:
1097    
1098      1. Recoded the buggy parts of x86-fp.
1099    
1100         a. All the block reordering code has been removed.
1101            We now depend on the block placement phases to do this work.
1102    
1103         b. Critical edge splitting code has been simplified and moved into the
1104            CFG modules, as where they belong.
1105    
1106         Both of these were quite buggy and complex.  The code is now much, much
1107         simpler.
1108    
1109      2. X86 backend.
1110    
1111         a. Added instructions for 64-bit support.  Instruction selection for
1112            64-bit has not been committed, however, since that
1113            requires changes to MLTREE which haven't been approved by
1114            Lal and John.
1115    
1116         b. Added support for FUCOMI and FUCOMIP when generating code for
1117            PentiumPro and above.  We only generate these instructions in
1118            the fast-fp mode.
1119    
1120         c. Added cases for JP and JNP in X86FreqProps.
1121    
1122      3. CFG
1123    
1124         CFG now has a bunch of methods for edge splitting and merging.
1125    
1126      4. Machine description.
1127    
1128         John's simplification of MLTREE_BASIS.fcond broke a few machine
1129         description things:
1130    
1131         rtl-build.{sig,sml} and hppa.mdl fixed.
1132    
1133         NOTE: the machine description stuff in the repository is still broken.
1134               Again, I can't put my fixes in because that involves
1135               changes to MLTREE.
1136    
1137    ----------------------------------------------------------------------
1138    Name: Matthias Blume
1139    Date: 2002/03/20 15:55:00 EST
1140    Tag: blume-20020320-kmp
1141    Description:
1142    
1143    Implemented Knuth-Morris-Pratt string matching in PreString and used
1144    it for String.isSubstring, Substring.isSubstring, and
1145    Substring.position.
1146    
1147    (Might need some stress-testing.  Simple examples worked fine.)
1148    
1149    ----------------------------------------------------------------------
1150    Name: Matthias Blume
1151    Date: 2002/03/19 16:37:00 EST
1152    Tag: blume-20020319-witnesses
1153    Description:
1154    
1155    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1156    
1157    This implements a generic mechanism for changing constness qualifiers
1158    anywhere within big C types without resorting to outright "casts".
1159    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1160    modify the constness at the outermost level.)
1161    The implementation of "convert" is based on the idea of "witness"
1162    values -- values that are not used by the operation but whose types
1163    "testify" to their applicability.  On the implementation side, "convert"
1164    is simply a projection (returning its second curried argument).  With
1165    cross-module inlining, it should not result in any machine code being
1166    generated.
1167    
1168    ----------------------------------------------------------------------
1169    Name: Matthias Blume
1170    Date: 2002/03/15 16:40:00 EST
1171    Tag: blume-20020315-basis
1172    Description:
1173    
1174    Provided (preliminary?) implementations for
1175    
1176      {String,Substring}.{concatWith,isSuffix,isSubstring}
1177    
1178    and
1179    
1180      Substring.full
1181    
1182    Those are in the Basis spec but they were missing in SML/NJ.
1183    
1184    ----------------------------------------------------------------------
1185    Name: Matthias Blume
1186    Date: 2002/03/14 21:30:00 EST
1187    Tag: blume-20020314-controls
1188    Description:
1189    
1190    Controls:
1191    ---------
1192    
1193    1. Factored out the recently-added Controls : CONTROLS stuff and put
1194       it into its own library $/controls-lib.cm.  The source tree for
1195       this is under src/smlnj-lib/Controls.
1196    
1197    2. Changed the names of types and functions in this interface, so they
1198       make a bit more "sense":
1199    
1200          module -> registry
1201          'a registry -> 'a group
1202    
1203    3. The interface now deals in ref cells only.  The getter/setter interface
1204       is (mostly) gone.
1205    
1206    4. Added a function that lets one register an already-existing ref cell.
1207    
1208    5. Made the corresponding modifications to the rest of the code so that
1209       everything compiles again.
1210    
1211    6. Changed the implementation of Controls.MLRISC back to something closer
1212       to the original.  In particular, this module (and therefore MLRISC)
1213       does not depend on Controls.  There now is some link-time code in
1214       int-sys.sml that registers the MLRISC controls with the Controls
1215       module.
1216    
1217    CM:
1218    ---
1219    
1220      * One can now specify the lambda-split aggressiveness in init.cmi.
1221    
1222    ----------------------------------------------------------------------
1223    Name: Allen Leung
1224    Date: 2002/03/13 17:30:00 EST
1225    Tag: leunga-20020313-x86-fp-unary
1226    Description:
1227    
1228    Bug fix for:
1229    
1230    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1231    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1232    > - fun f(x,(y,z)) = Real.~ y;
1233    > [autoloading]
1234    > [autoloading done]
1235    >       fchsl   (%eax), 184(%esp)
1236    > Error: MLRisc bug: X86MCEmitter.emitInstr
1237    >
1238    > uncaught exception Error
1239    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1240    
1241    The problem was that the code generator did not generate any fp registers
1242    in this case, and the ra didn't know that it needed to run the X86FP phase to
1243    translate the pseudo fp instruction.   This only happened with unary fp
1244    operators in certain situations.
1245    
1246    ----------------------------------------------------------------------
1247    Name: Matthias Blume
1248    Date: 2002/03/13 14:00:00 EST
1249    Tag: blume-20020313-overload-etc
1250    Description:
1251    
1252    1. Added _overload as a synonym for overload for backward compatibility.
1253       (Control.overloadKW must be true for either version to be accepted.)
1254    
1255    2. Fixed bug in install script that caused more things to be installed
1256       than what was requested in config/targets.
1257    
1258    3. Made CM aware of the (_)overload construct so that autoloading
1259       works.
1260    
1261    ----------------------------------------------------------------------
1262    Name: Matthias Blume
1263    Date: 2002/03/12 22:03:00 EST
1264    Tag: blume-20020312-url
1265    Description:
1266    
1267    Forgot to update BOOT and srcarchiveurl.
1268    
1269    ----------------------------------------------------------------------
1270    Name: Matthias Blume
1271    Date: 2002/03/12 17:30:00 EST
1272    Tag: blume-20020312-version110392
1273    Description:
1274    
1275    Yet another version number bump (because of small changes to the
1276    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1277    
1278    Changes:
1279    
1280      The new pid generation scheme described a few weeks ago was overly
1281      complicated.  I implemented a new mechanism that is simpler and
1282      provides a bit more "stability":  Once CM has seen a compilation
1283      unit, it keeps its identity constant (as long as you do not delete
1284      those crucial CM/GUID/* files).  This means that when you change
1285      an interface, compile, then go back to the old interface, and
1286      compile again, you arrive at the original pid.
1287    
1288      There now also is a mechanism that instructs CM to use the plain
1289      environment hash as a module's pid (effectively making its GUID
1290      the empty string).  For this, "noguid" must be specified as an
1291      option to the .sml file in question within its .cm file.
1292      This is most useful for code that is being generated by tools such
1293      as ml-nlffigen (because during development programmers tend to
1294      erase the tool's entire output directory tree including CM's cached
1295      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1296      revert to the old, broken behavior of SML/NJ, but in specific cases
1297      where there is no danger of interface confusion, its use is ok
1298      (I think).
1299    
1300      ml-nlffigen by default generates "noguid" annotations.  They can be
1301      turned off by specifying -guid in its command line.
1302    
1303    ----------------------------------------------------------------------
1304    Name: Lal George
1305    Date: 2002/03/12 12 14:42:36 EST
1306    Tag: george-20020312-frequency-computation
1307    Description:
1308    
1309    Integrated jump chaining and static block frequency into the
1310    compiler. More details and numbers later.
1311    
1312    ----------------------------------------------------------------------
1313    Name: Lal George
1314    Date: 2002/03/11 11 22:38:53 EST
1315    Tag: george-20020311-jump-chain-elim
1316    Description:
1317    
1318    Tested the jump chain elimination on all architectures (except the
1319    hppa).  This is on by default right now and is profitable for the
1320    alpha and x86, however, it may not be profitable for the sparc and ppc
1321    when compiling the compiler.
1322    
1323    The gc test will typically jump to a label at the end of the cluster,
1324    where there is another jump to an external cluster containing the actual
1325    code to invoke gc. This is to allow factoring of common gc invocation
1326    sequences. That is to say, we generate:
1327    
1328            f:
1329               testgc
1330               ja   L1      % jump if above to L1
1331    
1332            L1:
1333               jmp L2
1334    
1335    
1336    After jump chain elimination the 'ja L1' instructions is converted to
1337    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1338    up being implemented in their long form (if L2 is far away) using:
1339    
1340            jbe     L3      % jump if below or equal to L3
1341            jmp     L2
1342         L3:
1343            ...
1344    
1345    
1346    For large compilation units L2  may be far away.
1347    
1348    
1349    ----------------------------------------------------------------------
1350    Name: Matthias Blume
1351    Date: 2002/03/11 13:30:00 EST
1352    Tag: blume-20020311-mltreeeval
1353    Description:
1354    
1355    A functor parameter was missing.
1356    
1357    ----------------------------------------------------------------------
1358    Name: Allen Leung
1359    Date: 2002/03/11 10:30:00 EST
1360    Tag: leunga-20020311-runtime-string0
1361    Description:
1362    
1363       The representation of the empty string now points to a
1364    legal null terminated C string instead of unit.  It is now possible
1365    to convert an ML string into C string with InlineT.CharVector.getData.
1366    This compiles into one single machine instruction.
1367    
1368    ----------------------------------------------------------------------
1369    Name: Allen Leung
1370    Date: 2002/03/10 23:55:00 EST
1371    Tag: leunga-20020310-x86-call
1372    Description:
1373    
1374       Added machine generation for CALL instruction (relative displacement mode)
1375    
1376    ----------------------------------------------------------------------
1377    Name: Matthias Blume
1378    Date: 2002/03/08 16:05:00
1379    Tag: blume-20020308-entrypoints
1380    Description:
1381    
1382    Version number bumped to 110.39.1.  NEW BOOTFILES!
1383    
1384    Entrypoints: non-zero offset into a code object where execution should begin.
1385    
1386    - Added the notion of an entrypoint to CodeObj.
1387    - Added reading/writing of entrypoint info to Binfile.
1388    - Made runtime system bootloader aware of entrypoints.
1389    - Use the address of the label of the first function given to mlriscGen
1390      as the entrypoint.  This address is currently always 0, but it will
1391      not be 0 once we turn on block placement.
1392    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
1393      with entry points) from mlriscGen.
1394    
1395    ----------------------------------------------------------------------
1396    Name: Allen Leung
1397    Date: 2002/03/07 20:45:00 EST
1398    Tag: leunga-20020307-x86-cmov
1399    Description:
1400    
1401       Bug fixes for CMOVcc on x86.
1402    
1403       1. Added machine code generation for CMOVcc
1404       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1405       3. CMOVcc cannot have an immediate operand as argument.
1406    
1407    ----------------------------------------------------------------------
1408    Name: Matthias Blume
1409    Date: 2002/03/07 16:15:00 EST
1410    Tag: blume-20020307-controls
1411    Description:
1412    
1413    This is a very large but mostly boring patch which makes (almost)
1414    every tuneable compiler knob (i.e., pretty much everything under
1415    Control.* plus a few other things) configurable via both the command
1416    line and environment variables in the style CM did its configuration
1417    until now.
1418    
1419    Try starting sml with '-h' (or, if you are brave, '-H')
1420    
1421    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1422    implements the underlying generic mechanism.
1423    
1424    The interface to some of the existing such facilities has changed somewhat.
1425    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1426    (The getFoo interface is still there for backward-compatibility, but its
1427    use is deprecated.)
1428    
1429    The ml-build script passes -Cxxx=yyy command-line arguments through so
1430    that one can now twiddle the compiler settings when using this "batch"
1431    compiler.
1432    
1433    TODO items:
1434    
1435    We should go through and throw out all controls that are no longer
1436    connected to anything.  Moreover, we should go through and provide
1437    meaningful (and correct!) documentation strings for those controls
1438    that still are connected.
1439    
1440    Currently, multiple calls to Controls.new are accepted (only the first
1441    has any effect).  Eventually we should make sure that every control
1442    is being made (via Controls.new) exactly once.  Future access can then
1443    be done using Controls.acc.
1444    
1445    Finally, it would probably be a good idea to use the getter-setter
1446    interface to controls rather than ref cells.  For the time being, both
1447    styles are provided by the Controls module, but getter-setter pairs are
1448    better if thread-safety is of any concern because they can be wrapped.
1449    
1450    *****************************************
1451    
1452    One bug fix: The function blockPlacement in three of the MLRISC
1453    backpatch files used to be hard-wired to one of two possibilities at
1454    link time (according to the value of the placementFlag).  But (I
1455    think) it should rather sense the flag every time.
1456    
1457    *****************************************
1458    
1459    Other assorted changes (by other people who did not supply a HISTORY entry):
1460    
1461    1. the cross-module inliner now works much better (Monnier)
1462    2. representation of weights, frequencies, and probabilities in MLRISC
1463       changed in preparation of using those for weighted block placement
1464       (Reppy, George)
1465    
1466    ----------------------------------------------------------------------
1467    Name: Lal George
1468    Date: 2002/03/07 14:44:24 EST 2002
1469    Tag: george-20020307-weighted-block-placement
1470    
1471    Tested the weighted block placement optimization on all architectures
1472    (except the hppa) using AMPL to generate the block and edge frequencies.
1473    Changes were required in the machine properties to correctly
1474    categorize trap instructions. There is an MLRISC flag
1475    "weighted-block-placement" that can be used to enable weighted block
1476    placement, but this will be ineffective without block/edge
1477    frequencies (coming soon).
1478    
1479    
1480    ----------------------------------------------------------------------
1481    Name: Lal George
1482    Date: 2002/03/05 17:24:48 EST
1483    Tag: george-20020305-linkage-cluster
1484    
1485    In order to support the block placement optimization, a new cluster
1486    is generated as the very first cluster (called the linkage cluster).
1487    It contains a single jump to the 'real' entry point for the compilation
1488    unit. Block placement has no effect on the linkage cluster itself, but
1489    all the other clusters  have full freedom in the manner in which they
1490    reorder blocks or functions.
1491    
1492    On the x86 the typical linkage code that is generated is:
1493       ----------------------
1494            .align 2
1495       L0:
1496            addl    $L1-L0, 72(%esp)
1497            jmp     L1
1498    
1499    
1500            .align  2
1501       L1:
1502       ----------------------
1503    
1504    72(%esp) is the memory location for the stdlink register. This
1505    must contain the address of the CPS function being called. In the
1506    above example, it contains the address of  L0; before
1507    calling L1 (the real entry point for the compilation unit), it
1508    must contain the address for L1, and hence
1509    
1510            addl $L1-L0, 72(%esp)
1511    
1512    I have tested this on all architectures except the hppa.The increase
1513    in code size is of course negligible
1514    
1515    ----------------------------------------------------------------------
1516    Name: Allen Leung
1517    Date: 2002/03/03 13:20:00 EST
1518    Tag: leunga-20020303-mlrisc-tools
1519    
1520      Added #[ ... ] expressions to mlrisc tools
1521    
1522    ----------------------------------------------------------------------
1523    Name: Matthias Blume
1524    Date: 2002/02/27 12:29:00 EST
1525    Tag: blume-20020227-cdebug
1526    Description:
1527    
1528    - made types in structure C and C_Debug to be equal
1529    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1530    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1531    
1532    ----------------------------------------------------------------------
1533    Name: Matthias Blume
1534    Date: 2002/02/26 12:00:00 EST
1535    Tag: blume-20020226-ffi
1536    Description:
1537    
1538    1. Fixed a minor bug in CM's "noweb" tool:
1539       If numbering is turned off, then truly don't number (i.e., do not
1540       supply the -L option to noweb).  The previous behavior was to supply
1541       -L'' -- which caused noweb to use the "default" line numbering scheme.
1542       Thanks to Chris Richards for pointing this out (and supplying the fix).
1543    
1544    2. Once again, I reworked some aspects of the FFI:
1545    
1546       A. The incomplete/complete type business:
1547    
1548       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
1549         gone!
1550       - ML types representing an incomplete type are now *equal* to
1551         ML types representing their corresponding complete types (just like
1552         in C).  This is still safe because ml-nlffigen will not generate
1553         RTTI for incomplete types, nor will it generate functions that
1554         require access to such RTTI.   But when ML code generated from both
1555         incomplete and complete versions of the C type meet, the ML types
1556         are trivially interoperable.
1557    
1558         NOTE:  These changes restore the full generality of the translation
1559         (which was previously lost when I eliminated functorization)!
1560    
1561       B. Enum types:
1562    
1563       - Structure C now has a type constructor "enum" that is similar to
1564         how the "su" constructor works.  However, "enum" is not a phantom
1565         type because each "T enum" has values (and is isomorphic to
1566         MLRep.Signed.int).
1567       - There are generic access operations for enum objects (using
1568         MLRep.Signed.int).
1569       - ml-nlffigen will generate a structure E_foo for each "enum foo".
1570         * The structure contains the definition of type "mlrep" (the ML-side
1571         representation type of the enum).  Normally, mlrep is the same
1572         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
1573         then mlrep will be defined as a datatype -- thus facilitating
1574         pattern matching on mlrep values.
1575         ("-ec" will be suppressed if there are duplicate values in an
1576          enumeration.)
1577         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
1578         will be generated for each C enum constant xxx.
1579         * Conversion functions m2i and i2m convert between mlrep and
1580         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
1581         * Coversion functions c and ml convert between mlrep and "tag enum".
1582         * Access functions (get/set) fetch and store mlrep values.
1583       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
1584         enumerations are merged into one single enumeration represented by
1585         structure E_'.
1586    
1587    ----------------------------------------------------------------------
1588    Name: Allen Leung
1589    Date: 2002/02/25 04:45:00 EST
1590    Tag: leunga-20020225-cps-spill
1591    
1592    This is a new implementation of the CPS spill phase.
1593    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
1594    In case of problems, replace it with the old file spill.sml
1595    
1596    The current compiler runs into some serious performance problems when
1597    constructing a large record.  This can happen when we try to compile a
1598    structure with many items.  Even a very simple structure like the following
1599    makes the compiler slow down.
1600    
1601        structure Foo = struct
1602           val x_1 = 0w1 : Word32.int
1603           val x_2 = 0w2 : Word32.int
1604           val x_3 = 0w3 : Word32.int
1605           ...
1606           val x_N = 0wN : Word32.int
1607        end
1608    
1609    The following table shows the compile time, from N=1000 to N=4000,
1610    with the old compiler:
1611    
1612    N
1613    1000   CPS 100 spill                           0.04u  0.00s  0.00g
1614           MLRISC ra                               0.06u  0.00s  0.05g
1615              (spills = 0 reloads = 0)
1616           TOTAL                                   0.63u  0.07s  0.21g
1617    
1618    1100   CPS 100 spill                           8.25u  0.32s  0.64g
1619           MLRISC ra                               5.68u  0.59s  3.93g
1620              (spills = 0 reloads = 0)
1621           TOTAL                                   14.71u  0.99s  4.81g
1622    
1623    1500   CPS 100 spill                           58.55u  2.34s  1.74g
1624           MLRISC ra                               5.54u  0.65s  3.91g
1625              (spills = 543 reloads = 1082)
1626           TOTAL                                   65.40u  3.13s  6.00g
1627    
1628    2000   CPS 100 spill                           126.69u  4.84s  3.08g
1629           MLRISC ra                               0.80u  0.10s  0.55g
1630              (spills = 42 reloads = 84)
1631           TOTAL                                   129.42u  5.10s  4.13g
1632    
1633    3000   CPS 100 spill                           675.59u  19.03s  11.64g
1634           MLRISC ra                               2.69u  0.27s  1.38g
1635              (spills = 62 reloads = 124)
1636           TOTAL                                   682.48u  19.61s  13.99g
1637    
1638    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
1639           MLRISC ra                               4.96u  0.27s  2.72g
1640              (spills = 85 reloads = 170)
1641           TOTAL                                   2375.26u  57.21s  48.00g
1642    
1643    As you can see the old cps spill module suffers from some serious
1644    performance problem.  But since I cannot decipher the old code fully,
1645    instead of patching the problems up, I'm reimplementing it
1646    with a different algorithm.  The new code is more modular,
1647    smaller when compiled, and substantially faster
1648    (O(n log n) time and O(n) space).  Timing of the new spill module:
1649    
1650    4000  CPS 100 spill                           0.02u  0.00s  0.00g
1651          MLRISC ra                               0.25u  0.02s  0.15g
1652             (spills=1 reloads=3)
1653          TOTAL                                   7.74u  0.34s  1.62g
1654    
1655    Implementation details:
1656    
1657    As far as I can tell, the purpose of the CPS spill module is to make sure the
1658    number of live variables at any program point (the bandwidth)
1659    does not exceed a certain limit, which is determined by the
1660    size of the spill area.
1661    
1662    When the bandwidth is too large, we decrease the register pressure by
1663    packing live variables into spill records.  How we achieve this is
1664    completely different than what we did in the old code.
1665    
1666    First, there is something about the MLRiscGen code generator
1667    that we should be aware of:
1668    
1669    o MLRiscGen performs code motion!
1670    
1671       In particular, it will move floating point computations and
1672       address computations involving only the heap pointer to
1673       their use sites (if there is only a single use).
1674       What this means is that if we have a CPS record construction
1675       statement
1676    
1677           RECORD(k,vl,w,e)
1678    
1679       we should never count the new record address w as live if w
1680       has only one use (which is often the case).
1681    
1682       We should do something similar to floating point, but the transformation
1683       there is much more complex, so I won't deal with that.
1684    
1685    Secondly, there are now two new cps primops at our disposal:
1686    
1687     1. rawrecord of record_kind option
1688        This pure operator allocates some uninitialized storage from the heap.
1689        There are two forms:
1690    
1691         rawrecord NONE [INT n]  allocates a tagless record of length n
1692         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
1693                                     and initializes the tag.
1694    
1695     2. rawupdate of cty
1696          rawupdate cty (v,i,x)
1697          Assigns to x to the ith component of record v.
1698          The storelist is not updated.
1699    
1700    We use these new primops for both spilling and increment record construction.
1701    
1702     1. Spilling.
1703    
1704        This is implemented with a linear scan algorithm (but generalized
1705        to trees).  The algorithm will create a single spill record at the
1706        beginning of the cps function and use rawupdate to spill to it,
1707        and SELECT or SELp to reload from it.  So both spills and reloads
1708        are fine-grain operations.  In contrast, in the old algorithm
1709        "spills" have to be bundled together in records.
1710    
1711        Ideally, we should sink the spill record construction to where
1712        it is needed.  We can even split the spill record into multiple ones
1713        at the places where they are needed.  But CPS is not a good
1714        representation for global code motion, so I'll keep it simple and
1715        am not attempting this.
1716    
1717     2. Incremental record construction (aka record splitting).
1718    
1719        Long records with many component values which are simulatenously live
1720        (recall that single use record addresses are not considered to
1721         be live) are constructed with rawrecord and rawupdate.
1722        We allocate space on the heap with rawrecord first, then gradually
1723        fill it in with rawupdate.  This is the technique suggested to me
1724        by Matthias.
1725    
1726        Some restrictions on when this is applicable:
1727        1. It is not a VECTOR record.  The code generator currently does not handle
1728           this case. VECTOR record uses double indirection like arrays.
1729        2. All the record component values are defined in the same "basic block"
1730           as the record constructor.  This is to prevent speculative
1731           record construction.
1732    
1733    ----------------------------------------------------------------------
1734    Name: Allen Leung
1735    Date: 2002/02/22 01:02:00 EST
1736    Tag: leunga-20020222-mlrisc-tools
1737    
1738    Minor bug fixes in the parser and rewriter
1739    
1740    ----------------------------------------------------------------------
1741    Name: Allen Leung
1742    Date: 2002/02/21 20:20:00 EST
1743    Tag: leunga-20020221-peephole
1744    
1745    Regenerated the peephole files.  Some contained typos in the specification
1746    and some didn't compile because of pretty printing bugs in the old version
1747    of 'nowhere'.
1748    
1749    ----------------------------------------------------------------------
1750    Name: Allen Leung
1751    Date: 2002/02/19 20:20:00 EST
1752    Tag: leunga-20020219-mlrisc-tools
1753    Description:
1754    
1755       Minor bug fixes to the mlrisc-tools library:
1756    
1757       1.  Fixed up parsing colon suffixed keywords
1758       2.  Added the ability to shut the error messages up
1759       3.  Reimplemented the pretty printer and fixed up/improved
1760           the pretty printing of handle and -> types.
1761       4.  Fixed up generation of literal symbols in the nowhere tool.
1762       5.  Added some SML keywords to to sml.sty
1763    
1764    ----------------------------------------------------------------------
1765    Name: Matthias Blume
1766    Date: 2002/02/19 16:20:00 EST
1767    Tag: blume-20020219-cmffi
1768    Description:
1769    
1770    A wild mix of changes, some minor, some major:
1771    
1772    * All C FFI-related libraries are now anchored under $c:
1773        $/c.cm      --> $c/c.cm
1774        $/c-int.cm  --> $c/internals/c-int.cm
1775        $/memory.cm --> $c/memory/memory.cm
1776    
1777    * "make" tool (in CM) now treats its argument pathname slightly
1778      differently:
1779        1. If the native expansion is an absolute name, then before invoking
1780           the "make" command on it, CM will apply OS.Path.mkRelative
1781           (with relativeTo = OS.FileSys.getDir()) to it.
1782        2. The argument will be passed through to subsequent phases of CM
1783           processing without "going native".  In particular, if the argument
1784           was an anchored path, then "make" will not lose track of that anchor.
1785    
1786    * Compiler backends now "know" their respective C calling conventions
1787      instead of having to be told about it by ml-nlffigen.  This relieves
1788      ml-nlffigen from one of its burdens.
1789    
1790    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
1791    
1792    * Export C_DEBUG and C_Debug from $c/c.cm.
1793    
1794    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
1795      subtyping relationship between incomplete pointers and their complete
1796      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
1797      with the convention of instantiating 'o with ('t, 'c) obj whenever
1798      the pointer target type is complete.  In the incomplete case, 'o
1799      will be instantiated with some "'c iobj" -- a type obtained by
1800      using one of the functors PointerToIncompleteType or PointerToCompleteType.
1801    
1802      Operations that work on both incomplete and complete pointer types are
1803      typed as taking an 'o ptr while operations that require the target to
1804      be known are typed as taking some ('t, 'c) obj ptr.
1805    
1806      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
1807      where void is an eqtype without any values.  This makes it possible
1808      to work on voidptr values using functions meant to operate on light
1809      incomplete pointers.
1810    
1811    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
1812      been vastly simplified.
1813    
1814    ----------------------------------------------------------------------
1815    Name: Matthias Blume
1816    Date: 2002/02/19 10:48:00 EST
1817    Tag: blume-20020219-pqfix
1818    Description:
1819    
1820    Applied Chris Okasaki's bug fix for priority queues.
1821    
1822    ----------------------------------------------------------------------
1823    Name: Matthias Blume
1824    Date: 2002/02/15 17:05:00
1825    Tag: Release_110_39
1826    Description:
1827    
1828    Last-minute retagging is becoming a tradition... :-(
1829    
1830    This is the working release 110.39.
1831    
1832    ----------------------------------------------------------------------
1833    Name: Matthias Blume
1834    Date: 2002/02/15 16:00:00 EST
1835    Tag: Release_110_39-orig
1836    Description:
1837    
1838    Working release 110.39.  New bootfiles.
1839    
1840    (Update: There was a small bug in the installer so it wouldn't work
1841    with all shells.  So I retagged. -Matthias)
1842    
1843    ----------------------------------------------------------------------
1844    Name: Matthias Blume
1845    Date: 2002/02/15 14:17:00 EST
1846    Tag: blume-20020215-showbindings
1847    Description:
1848    
1849    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
1850    the latter can be useful for exploring what bindings are available at
1851    the interactive prompt.  (The first function returns only the list
1852    of symbols that are really bound, the second prints those but also the
1853    ones that CM's autoloading mechanism knows about.)
1854    
1855    ----------------------------------------------------------------------
1856    Name: Matthias Blume
1857    Date: 2002/02/15 12:08:00 EST
1858    Tag: blume-20020215-iptrs
1859    Description:
1860    
1861    Two improvements to ml-nlffigen:
1862    
1863      1. Write files only if they do not exist or if their current contents
1864         do not coincide with what's being written.  (That is, avoid messing
1865         with the time stamps unless absolutely necessary.)
1866    
1867      2. Implement a "repository" mechanism for generated files related
1868         to "incomplete pointer types".   See the README file for details.
1869    
1870    ----------------------------------------------------------------------
1871    Name: Matthias Blume
1872    Date: 2002/02/14 11:50:00 EST
1873    Tag: blume-20020214-quote
1874    Description:
1875    
1876    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
1877    because of the new and improved tag generation scheme.  (Thanks to Allen
1878    Leung for pointing it out.)
1879    
1880    ----------------------------------------------------------------------
1881    Name: Lal George
1882    Date: 2002/02/14 09:55:27 EST 2002
1883    Tag: george-20020214-isabelle-bug
1884    Description:
1885    
1886    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
1887    of Isabelle on the x86.
1888    
1889    From Allen:
1890    -----------
1891     I've found the problem:
1892    
1893         in ra-core.sml, I use the counter "blocked" to keep track of the
1894         true number of elements in the freeze queue.  When the counter goes
1895         to zero, I skip examining the queue.  But I've messed up the
1896         bookkeeping in combine():
1897    
1898             else ();
1899             case !ucol of
1900               PSEUDO => (if !cntv > 0 then
1901                     (if !cntu > 0 then blocked := !blocked - 1 else ();
1902                                        ^^^^^^^^^^^^^^^^^^^^^^^
1903                      moveu := mergeMoveList(!movev, !moveu)
1904                     )
1905                  else ();
1906    
1907         combine() is called to coalesce two nodes u and v.
1908         I think I was thinking that if the move counts of u and v are both
1909         greater than zero then after they are coalesced then one node is
1910         removed from the freeze queue.  Apparently I was thinking that
1911         both u and v are of low degree, but that's clearly not necessarily true.
1912    
1913    
1914    02/12/2002:
1915        Here's the patch.  HOL now compiles.
1916    
1917        I don't know how this impact on performance (compile
1918        time or runtime).  This bug caused the RA (especially on the x86)
1919        to go thru the potential spill phase when there are still nodes on the
1920        freeze queue.
1921    
1922    
1923    
1924    
1925    ----------------------------------------------------------------------
1926    Name: Matthias Blume
1927    Date: 2002/02/13 22:40:00 EST
1928    Tag: blume-20020213-fptr-rtti
1929    Description:
1930    
1931    Fixed a bug in ml-nlffigen that was introduced with one of the previous
1932    updates.
1933    
1934    ----------------------------------------------------------------------
1935    Name: Matthias Blume
1936    Date: 2002/02/13 16:41:00 EST
1937    Tag: blume-20020213-cmlpq
1938    Description:
1939    
1940    Added new priority queue export symbols (which have just been added to
1941    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
1942    not compile and the installer would choke.)
1943    
1944    ----------------------------------------------------------------------
1945    Name: Matthias Blume
1946    Date: 2002/02/13 16:15:00 EST
1947    Tag: blume-20020213-various
1948    Description:
1949    
1950    1. More tweaks to ml-nlffigen:
1951    
1952       - better internal datastructures (resulting in slight speedup)
1953       - "-match" option requires exact match
1954       - "localized" gensym counters (untagged structs/unions nested within
1955         other structs/unions or within typedefs get a fresh counter; their
1956         tag will be prefixed by a concatenation of their parents' tags)
1957       - bug fixes (related to calculation of transitive closure of types
1958         to be included in the output)
1959    
1960    2. Minor Basis updates:
1961    
1962       - added implementations for List.collate and Option.app
1963    
1964    ----------------------------------------------------------------------
1965    Name: Matthias Blume
1966    Date: 2002/02/11 15:55:00 EST
1967    Tag: blume-20020211-gensym
1968    Description:
1969    
1970    Added a "-gensym" option to command line of ml-nlffigen.  This can be
1971    used to specify a "stem" -- a string that is inserted in all "gensym'd"
1972    names (ML structure names that correspond to unnamed C structs, unions,
1973    and enums), so that separate runs of ml-nlffigen do not clash.
1974    
1975    ----------------------------------------------------------------------
1976    Name: Matthias Blume
1977    Date: 2002/02/11 12:05:00 EST
1978    Tag: blume-20020211-gensml
1979    Description:
1980    
1981    A quick fix for a problem with GenSML (in the pgraph-util library):
1982    Make generation of toplevel "local" optional.  (Strictly speaking,
1983    signature definitions within "local" are not legal SML.)
1984    
1985    Other than that: updates to INSTALL and cm/TODO.
1986    
1987    ----------------------------------------------------------------------
1988    Name: Matthias Blume
1989    Date: 2002/02/08 15:00:00 EST
1990    Tag: blume-20020208-uniquepid
1991    Description:
1992    
1993    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
1994    
1995    1. The installer (config/install.sh) has gotten smarter:
1996    
1997         - Configuration options are a bit easier to specify now
1998           (in config/targets).
1999         - Bug in recognizing .tar.bz2 files fixed.
2000         - Installer automatically resolves dependencies between
2001           configuration options (e.g., if you ask for eXene, you will
2002           also get cml -- regardless whether you asked for it or not).
2003         - Installer can run in "quieter mode" by setting the environment
2004           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2005           "completely silent", though.
2006         - Build HashCons library as part of smlnj-lib.
2007    
2008    2. A new scheme for assigning persistent identifiers to compilation
2009       units (and, by extension, to types etc.) has been put into place.
2010       This fixes a long-standing bug where types and even dynamic values
2011       can get internally confused, thereby compromising type safety
2012       (abstraction) and dynamic correctness.  See
2013    
2014         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2015    
2016       for an example of how things could go wrong until now.
2017    
2018       The downside of the new scheme is that pids are not quite as
2019       persistent as they used to be: CM will generate a fresh pid
2020       for every compilation unit that it thinks it sees for the first
2021       time.  That means that if you compile starting from a clean, fresh
2022       source tree at two different times, you end up with different
2023       binaries.
2024    
2025       Cutoff recompilation, however, has not been compromised because
2026       CM keeps pid information in special caches between runs.
2027    
2028    ----------------------------------------------------------------------
2029    Name: Lal George
2030    Date: 2002/02/07 15:34:13 EST 2002
2031    Tag: <none>
2032    Description:
2033    
2034    Compilers that generate assembly code may produce  global labels
2035    whose value is resolved at link time. The various peephole optimization
2036    modules did not take this in account.
2037    
2038    TODO. The Labels.addrOf function should really return an option
2039    type so that clients are forced to deal with this issue, rather
2040    than an exception being raised.
2041    
2042    ----------------------------------------------------------------------
2043    Name: Lal George
2044    Date: 2002/02/06 13:55:02 EST
2045    Tag: george-20020206-ra-breakup
2046    Description:
2047    
2048    1. A bug fix from Allen.
2049    
2050        A typo causes extra fstp %st(0)'s to be generated at compensation
2051        edges, which might cause stack underflow traps at runtime.  This
2052        occurs in fft where there are extraneous fstps right before the 'into'
2053        trap instruction (in this case they are harmless since none of the
2054        integers overflow.)
2055    
2056    2. Pulled out various utility modules that were embedded in the modules
2057       of the register allocator. I need these modules for other purposes, but
2058       they are not complete enough to put into a library (just yet).
2059    ----------------------------------------------------------------------
2060    Name: Matthias Blume
2061    Date: 2002/01/31 16:05:00 EST
2062    Tag: blume-20020131-sparc-ccalls
2063    Description:
2064    
2065    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2066       of extra stack space by mistake.  Fixed.
2067    
2068    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2069    
2070    ----------------------------------------------------------------------
2071    Name: Allen Leung
2072    Date: 2002/01/30
2073    Tag: leunga-20020130-nowhere-bug-fix
2074    Description:
2075    
2076       MLRISC bug fixes:
2077       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2078       2. MachineInt.fromString was negating its value.
2079    
2080    ----------------------------------------------------------------------
2081    Name: Matthias Blume
2082    Date: 2002/01/29
2083    Tag: blume-20020129-INSTALL
2084    Description:
2085    
2086    - Added somewhat detailed installation instructions (file INSTALL).
2087    - Fixed curl-detection bug in config/install.sh.
2088    - It is now possible to select the URL getter using the URLGETTER
2089      environment variable:
2090    
2091          not set / "unknown"      --> automatic detection (script tries wget,
2092                                       curl, and lynx)
2093          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2094                                       how to properly invoke them)
2095          other                    --> use $URLGETTER directly, it must take
2096                                       precisely two command-line arguments
2097                                       (source URL and destination file name)
2098    
2099    ----------------------------------------------------------------------
2100    Name: Matthias Blume
2101    Date: 2002/01/28
2102    Tag: blume-20020128-sparc-ccalls
2103    Description:
2104    
2105    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2106    - Make use of the allocParam argument in sparc-c-calls.
2107    
2108    ----------------------------------------------------------------------
2109    Name: Matthias Blume
2110    Date: 2002/01/28
2111    Tag: blume-20020128-allocParam
2112    Description:
2113    
2114    John Reppy:  Changes c-calls API to accept client-callback for
2115    allocating extra stack space.
2116    me: Corresponding changes to mlriscGen (using a dummy argument that
2117        does not change the current behavior).
2118    
2119    ----------------------------------------------------------------------
2120    Name: Matthias Blume
2121    Date: 2002/01/28 12:00:00
2122    Tag: Release_110_38
2123    Description:
2124    
2125    This time for real!!!
2126    
2127    ----------------------------------------------------------------------
2128    Name: Matthias Blume
2129    Date: 2002/01/28 10:56:00 EST
2130    Tag: blume-20020128-retraction
2131    Description:
2132    
2133    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
2134       with blume-Release_110_38-retracted.)
2135    
2136    1. Fixed a problem with incorrect rounding modes in real64.sml.
2137       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2138    
2139    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2140       and enums fixed.  The naming of corresponding ML identifiers should
2141       now be consistent again.
2142    
2143    ----------------------------------------------------------------------
2144    Name: Allen Leung
2145    Date: 2002/01/27
2146    Tag: leunga-20020127-nowhere
2147    Description:
2148    
2149       Added a target called nowhere in the configuration scripts.
2150       Enabling this will build the MLRISC 'nowhere' tool (for translating
2151       programs with where-clauses into legal SML code) during installation.
2152    
2153    ----------------------------------------------------------------------
2154    Name: Matthias Blume
2155    Date: 2002/01/25 21:27:00 EST
2156    Tag: blume-Release_110_38-retracted
2157    Description:
2158    
2159    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
2160    
2161    README will be added later.
2162    
2163    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
2164    (Release_110_38) removed.  Reason: Last-minute bug fixes.
2165    
2166    ----------------------------------------------------------------------
2167    Name: Matthias Blume
2168    Date: 2002/01/25
2169    Tag: blume-20020125-ffi
2170    Description:
2171    
2172    A large number of tweaks and improvements to ml-nlffi-lib and
2173    ml-nlffigen:
2174    
2175       - ML represenation types have been streamlined
2176       - getter and setter functions work with concrete values, not abstract
2177         ones where possible
2178       - ml-nlffigen command line more flexible (see README file there)
2179       - some bugs have been fixed (hopefully)
2180    
2181    ----------------------------------------------------------------------
2182    Name: Lal George
2183    Date: 2002/01/24
2184    Tag: george-20020124-risc-ra-interface
2185    Description:
2186    
2187       There is a dramatic simplification in the interface to the
2188       register allocator for RISC architectures as a result of making
2189       parallel copy instructions explicit.
2190    
2191    ----------------------------------------------------------------------
2192    Name: Matthias Blume
2193    Date: 2002/01/22
2194    Tag: blume-20020122-x86-ccalls
2195    Description:
2196    
2197    Bug fix for c-calls on x86 (having to do with how char- and
2198    short-arguments are being handled).
2199    
2200    ----------------------------------------------------------------------
2201    Name: Matthias Blume
2202    Date: 2002/01/21
2203    Tag: blume-20020121-ff
2204    Description:
2205    
2206    Another day of fiddling with the FFI...
2207    
2208    1. Bug fix/workaround:  CKIT does not complain about negative array
2209       dimensions, so ml-nlffigen has to guard itself against this possibility.
2210       (Otherwise a negative dimension would send it into an infinite loop.)
2211    
2212    2. Some of the abstract types (light objects, light pointers, most "base"
2213       types) in structure C are now eqtypes.
2214    
2215    3. Added constructors and test functions for NULL function pointers.
2216    
2217    ----------------------------------------------------------------------
2218    Name: Matthias Blume
2219    Date: 2002/01/18
2220    Tag: blume-20020118-ready-for-new-release
2221    Description:
2222    
2223    Made config/srcarchiveurl point to a new place.  (Will provide boot
2224    files shortly.)
2225    
2226    Maybe we christen this to be 110.38?
2227    
2228    ----------------------------------------------------------------------
2229    Name: Matthias Blume
2230    Date: 2002/01/18
2231    Tag: blume-20020118-more-ffifiddle
2232    Description:
2233    
2234    Today's FFI fiddling:
2235    
2236      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
2237        C.Set.* that use concrete (MLRep.*) arguments and results instead
2238        of abstract ones.
2239    
2240      - Provided word-style bit operations etc. for "int" representation
2241        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
2242        and Long.
2243    
2244    ----------------------------------------------------------------------
2245    Name: Matthias Blume
2246    Date: 2002/01/18
2247    Tag: blume-20020118-use-x86-fp
2248    Description:
2249    
2250    Now that x86-fast-fp seems to be working, I turned it back on again
2251    by default.  (Seems to work fine now, even with the FFI.)
2252    
2253    Other than that, I added some documentation about the FFI to
2254    src/ml-nlffigen/README and updated the FFI test examples in
2255    src/ml-nlffi-lib/Tests/*.
2256    
2257    ----------------------------------------------------------------------
2258    Name: Allen Leung
2259    Date: 2002/01/17
2260    Tag: leunga-20020117-x86-fast-fp-call
2261    Description:
2262    
2263       1. Fixed a problem with handling return fp values when x86's fast fp
2264          mode is turned on.
2265    
2266       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
2267          of %f32.
2268    
2269       3. Added a constructor INT32lit to the ast of MLRISC tools.
2270    
2271    ----------------------------------------------------------------------
2272    Name: Matthias Blume
2273    Date: 2002/01/16
2274    Tag: blume-20020116-ffifiddle
2275    Description:
2276    
2277    More fiddling with the FFI interface:
2278    
2279     - Make constness 'c instead of rw wherever possible.  This eliminates
2280       the need for certain explicit coercions.  (However, due to ML's
2281       value polymorphism, there will still be many cases where explicit
2282       coercions are necessary.  Phantom types are not the whole answer
2283       to modeling a subtyping relationship in ML.)
2284    
2285     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
2286    
2287     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
2288       clumsy workaround for varargs problem.)
2289    
2290    ----------------------------------------------------------------------
2291    Name: Lal George
2292    Date: 2002/01/15
2293    Tag: <none>
2294    Description:
2295    
2296    1. Since COPY instructions are no longer native to the architecture,
2297       a generic functor can be used to implement the expandCopies function.
2298    
2299    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
2300       TEXT segment.
2301    
2302    ----------------------------------------------------------------------
2303    Name: Matthias Blume
2304    Date: 2002/01/15
2305    Tag: blume-20020115-ffiupdates
2306    Description:
2307    
2308    1. Fix for bug resulting in single-precision float values being returned
2309       incorrectly from FFI calls.
2310    
2311    2. Small modifications to C FFI API:
2312    
2313        - memory-allocation routines return straight objects (no options)
2314          and raise an exception in out-of-memory situations
2315        - unsafe extensions to cast between function pointers and pointers
2316          from/to ints
2317        - added structure C_Debug as an alternative to structure C where
2318          pointer-dereferencing (|*| and |*!) always check for null-pointers
2319        - added open_lib' to DynLinkage;  open_lib' works like open_lib
2320          but also takes a (possibly empty) list of existing library handles
2321          that the current library depends on
2322    
2323    ----------------------------------------------------------------------
2324    Name: Matthias Blume
2325    Date: 2002/01/10
2326    Tag: blume-20020110-newffigen
2327    Description:
2328    
2329    1. Updates to portable graph code.
2330    
2331    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
2332       more scalable now so that even huge interfaces such as the one
2333       for GTK compile in finite time and space. :-)
2334       See src/ml-nlffigen/README for details on what's new.
2335    
2336    ----------------------------------------------------------------------
2337    Name: Lal George
2338    Date: 2001/01/09 14:31:35 EST 2002
2339    Tag: george-20011206-rm-native-copy
2340    Description:
2341    
2342            Removed the native COPY and FCOPY instructions
2343            from all the architectures and replaced it with the
2344            explicit COPY instruction from the previous commit.
2345    
2346            It is now possible to simplify many of the optimizations
2347            modules that manipulate copies. This has not been
2348            done in this change.
2349    
2350    ----------------------------------------------------------------------
2351    Name: Lal George
2352    Date: 2001/12/06 16:50:13 EST 2001
2353    Tag: george-20011206-mlrisc-instruction
2354    Description:
2355    
2356    Changed the representation of instructions from being fully abstract
2357    to being partially concrete. That is to say:
2358    
2359      from
2360            type instruction
2361    
2362      to
2363            type instr                              (* machine instruction *)
2364    
2365            datatype instruction =
2366                LIVE of {regs: C.cellset, spilled: C.cellset}
2367              | KILL of {regs: C.cellset, spilled: C.cellset}
2368              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
2369              | ANNOTATION of {i: instruction, a: Annotations.annotation}
2370              | INSTR of instr
2371    
2372    This makes the handling of certain special instructions that appear on
2373    all architectures easier and uniform.
2374    
2375    LIVE and KILL say that a list of registers are live or killed at the
2376    program point where they appear. No spill code is generated when an
2377    element of the 'regs' field is spilled, but the register is moved to
2378    the 'spilled' (which is present, more for debugging than anything else).
2379    
2380    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
2381    We used to generate:
2382    
2383            DEFFREG f1
2384            f1 := f2 + f3
2385            trapb
2386    
2387    but now generate:
2388    
2389            f1 := f2 + f3
2390            trapb
2391            LIVE {regs=[f1,f2,f3], spilled=[]}
2392    
2393    Furthermore, the DEFFREG (hack) required that all floating point instruction
2394    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
2395    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
2396    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
2397    this architecture).
2398    
2399    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
2400    all the architectures. This will result in further simplification of the
2401    register allocator that must be aware of them for coalescing purposes, and
2402    will also simplify certain aspects of the machine description that provides
2403    callbacks related to parallel copies.
2404    
2405    ANNOTATION should be obvious, and now INSTR represents the honest to God
2406    machine instruction set!
2407    
2408    The <arch>/instructions/<arch>Instr.sml files define certain utility
2409    functions for making porting easier -- essentially converting upper case
2410    to lower case. All machine instructions (of type instr) are in upper case,
2411    and the lower case form generates an MLRISC instruction. For example on
2412    the alpha we have:
2413    
2414      datatype instr =
2415         LDA of {r:cell, b:cell, d:operand}
2416       | ...
2417    
2418      val lda : {r:cell, b:cell, d:operand} -> instruction
2419        ...
2420    
2421    where lda is just (INSTR o LDA), etc.
2422    
2423    ----------------------------------------------------------------------
2424    Name: Matthias Blume
2425    Date: 2001/11/22 21:40:00 EST
2426    Tag: Release_110_37
2427    Description:
2428    
2429    Release 110.37.  This time for real.
2430    
2431    ----------------------------------------------------------------------
2432    Name: Matthias Blume
2433    Date: 2001/11/21 16:35:00 EST
2434    Tag: blume-20011121-foot-in-mouth
2435    Description:
2436    
2437    Removed the "Release_110_37" tag because of a serious bug.
2438    This will be re-tagged once the bug is fixed.
2439    
2440    ----------------------------------------------------------------------
2441    Name: Matthias Blume
2442    Date: 2001/11/21 16:14:00 EST
2443    Tag: blume-20011121-forgottenfile
2444    Description:
2445    
2446    Forgot to add a file.  (Just a .tex-file -- part of
2447    the CM manual source.)
2448    
2449    ----------------------------------------------------------------------
2450    Name: Matthias Blume
2451    Date: 2001/11/21 16:10:00 EST
2452    Tag: blume-20011121-invalid_110_37
2453    Description:
2454    
2455    Note: I removed the original tag "Release_110_37" from this commit
2456          because we found a serious bug in all non-x86 backends.
2457          - Matthias
2458    
2459    1. Modifications to the SML/NJ code generator and to the runtime system
2460       so that code object name strings are directly inserted into code
2461       objects at code generation time.  The only business the runtime system
2462       has with this is now to read the name strings on occasions.
2463       (The encoding of the name string has also changed somewhat.)
2464    
2465    2. CM now implements a simple "set calculus" for specifying export lists.
2466       In particular, it is now possible to refer to the export lists of
2467       other libraries/groups/sources and form unions as well as differences.
2468       See the latest CM manual for details.
2469    
2470    3. An separate notion of "proxy" libraries has again be eliminated from
2471       CM's model.  (Proxy libraries are now simply a special case of using
2472       the export list calculus.)
2473    
2474    4. Some of the existing libraries now take advantage of the new set
2475       calculus.
2476       (Notice that not all libraries have been converted because some
2477       of the existing .cm-files are supposed to be backward compatible
2478       with 110.0.x.)
2479    
2480    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
2481       "General.exnMessage"!  The former relies on a certain hook to be
2482       initialized, and that often does not happen in the stand-alone case.)
2483    
2484    ----------------------------------------------------------------------
2485    Name: Lal George
2486    Date: 2001/11/21  13:56:18 EST
2487    Tag: george-2001121-pseudo-ops
2488    Description:
2489    
2490      Implemented a complete redesign of MLRISC pseudo-ops. Now there
2491      ought to never be any question of incompatabilities with
2492      pseudo-op syntax expected by host assemblers.
2493    
2494      For now, only modules supporting GAS syntax are implemented
2495      but more should follow, such as MASM, and vendor assembler
2496      syntax, e.g. IBM as, Sun as, etc.
2497    
2498    ----------------------------------------------------------------------
2499    Name: Matthias Blume
2500    Date: 2001/11/14 11:52:00 EST
2501    Tag: blume-20011114-srcname
2502    Description:
2503    
2504    1. Routed the name of the current source file to mlriscgen where it
2505       should be directly emitted into the code object.  (This last part
2506       is yet to be done.)
2507    
2508    2. Some cleanup of the pgraph code to make it match the proposal that
2509       I put out the other day.  (The proposal notwithstanding, things are
2510       still in flux here.)
2511    
2512    ----------------------------------------------------------------------
2513    Name: Lal George
2514    Date: 2001/11/14 09:44:04 EST
2515    Tag:
2516    Description:
2517    
2518      Fix for a backpatching bug reported by Allen.
2519    
2520      Because the boundary between short and long span-dependent
2521      instructions is +/- 128, there are an astounding number of
2522      span-dependent instructions whose size is over estimated.
2523    
2524      Allen came up with the idea of letting the size of span
2525      dependent instructions be non-monotonic, for a maxIter
2526      number of times, after which the size must be monotonically
2527      increasing.
2528    
2529      This table shows the number of span-dependent instructions
2530      whose size was over-estimated as a function of maxIter, for the
2531      file Parse/parse/ml.grm.sml:
2532    
2533         maxIter            # of instructions:
2534            10                      687
2535            20                      438
2536            30                      198
2537            40                        0
2538    
2539      In compiling the compiler, there is no significant difference in
2540      compilation speed between maxIter=10 and maxIter=40. Actually,
2541      my measurements showed that maxIter=40 was a tad faster than
2542      maxIter=10! Also 96% of the  files in the compiler reach a fix
2543      point within 13 iterations, so fixing maxIter at 40, while high,
2544      is okay.
2545    
2546    ----------------------------------------------------------------------
2547    Name: Matthias Blume
2548    Date: 2001/10/31 15:25:00 EST
2549    Tag: blume-20011031-pgraph
2550    Description:
2551    
2552    CKIT:
2553    * Changed the "Function" constructor of type Ast.ctype to carry optional
2554      argument identifiers.
2555    * Changed the return type of TypeUtil.getFunction accordingly.
2556    * Type equality ignores the argument names.
2557    * TypeUtil.composite tries to preserve argument names but gives up quickly
2558      if there is a mismatch.
2559    
2560    installation script:
2561    * attempts to use "curl" if available (unless "wget" is available as well)
2562    
2563    CM:
2564    * has an experimental implementation of "portable graphs" which I will
2565      soon propose as an implementation-independent library format
2566    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
2567    
2568    NLFFI-LIB:
2569    * some cleanup (all cosmetic)
2570    
2571    NLFFIGEN:
2572    * temporarily disabled the mechanism that suppresses ML output for
2573      C definitions whose identifiers start with an underscore character
2574    * generate val bindings for enum constants
2575    * user can request that only one style (light or heavy) is being used;
2576      default is to use both (command-line arguments: -heavy and -light)
2577    * fixed bug in handling of function types involving incomplete pointers
2578    * generate ML entry points that take record arguments (i.e., using
2579      named arguments) for C functions that have a prototype with named
2580      arguments
2581      (see changes to CKIT)
2582    
2583    ----------------------------------------------------------------------
2584    Name: Allen Leung
2585    Date: 2001/10/27 20:34:00 EDT
2586    Tag: leunga-20011027-x86-fast-fp-call
2587    Description:
2588    
2589       Fixed the bug described in blume-20010920-slowfp.
2590    
2591       The fix involves
2592          1. generating FCOPYs in FSTP in ia32-svid
2593          2. marking a CALL with the appropriate annotation
2594    
2595    ----------------------------------------------------------------------
2596    Name: Matthias Blume
2597    Date: 2001/10/16 11:32:00 EDT
2598    Tag: blume-20011016-netbsd
2599    Description:
2600    
2601    Underscore patch from Chris Richards (fixing problem with compiling
2602    runtime system under recent NetBSD).
2603    
2604    ----------------------------------------------------------------------
2605    Name: Allen Leung
2606    Date: 2001/10/12 17:18:32 EDT 2001
2607    Tag: leung-20011012-x86-printflowgraph
2608    Description:
2609    
2610    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
2611    
2612    ----------------------------------------------------------------------
2613    Name: Lal George
2614    Date: 2001/10/11 23:51:34 EDT
2615    Tag: george-20011011-too-many-instrs
2616    Description:
2617    
2618    The representation of a program point never expected to see more
2619    than 65536 instructions in a basic block!
2620    
2621    ----------------------------------------------------------------------
2622    Name: Lal George
2623    Date: 2001/10/09 09:41:37 EDT
2624    Tag: george-20011008-mlrisc-labels
2625    Description:
2626    
2627    Changed the machine description files to support printing of
2628    local and global labels in assembly code, based on host assembler
2629    conventions.
2630    
2631    ----------------------------------------------------------------------
2632    Name: Matthias Blume
2633    Date: 2001/09/25 15:25:00 EDT
2634    Tag: blume-20010925-exninfo
2635    Description:
2636    
2637    I provided a non-hook implementation of exnName (at the toplevel) and
2638    made the "dummy" implementation of exnMessage (at the toplevel) more
2639    useful: if nothing gets "hooked in", then at least you are going to
2640    see the exception name and a message indicating why you don't see more.
2641    
2642    [For the time being, programs that need exnMessage and want to use
2643    ml-build should either use General.exnMessage (strongly recommended) or
2644    refer to structure General at some other point so that CM sees a
2645    static dependency.]
2646    
2647    [Similar remarks go for "print" and "use":  If you want to use their
2648    functionality in stand-alone programs generated by ml-build, then use
2649    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
2650    
2651    ----------------------------------------------------------------------
2652    Name: Matthias Blume
2653    Date: 2001/09/20 17:28:00 EDT
2654    Tag: blume-20010920-slowfp
2655    Description:
2656    
2657    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
2658    
2659    ----------------------------------------------------------------------
2660    Name: Matthias Blume
2661    Date: 2001/09/20 17:20:00 EDT
2662    Tag: blume-20010920-canonicalpaths
2663    Description:
2664    
2665    0. Updated the BOOT file (something that I forgot to do earlier).
2666    
2667    1. Small internal change to CM so that it avoids "/../" in filenames
2668       as much as possible (but only where it is safe).
2669    
2670    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
2671       that arguments that contain delimiters are passed through correctly.
2672       This change also means that all "special" arguments of the form
2673       @SMLxxx... must come first.
2674    
2675    3. Changed install script to put relative anchor names for tool commands
2676       into pathconfig.
2677    
2678    ----------------------------------------------------------------------
2679    Name: Matthias Blume
2680    Date: 2001/09/18 15:35:00 EDT
2681    Tag: blume-20010918-readme11036
2682    Description:
2683    
2684    Added README files.
2685    
2686    ----------------------------------------------------------------------
2687    Name: Matthias Blume
2688    Date: 2001/09/18 11:45:00 EDT
2689    Tag: Release_110_36 (retag)
2690    Description:
2691    
2692    Fixed mistake in config/preloads. Retagged as 110.36.
2693    
2694    ----------------------------------------------------------------------
2695    Name: Matthias Blume
2696    Date: 2001/09/18 09:40:00 EDT
2697    Tag: Release_110_36_orig (tag changed)
2698    Description:
2699    
2700    New version (110.36).  New bootfiles.
2701    
2702    ----------------------------------------------------------------------
2703    Name: Matthias Blume
2704    Date: 2001/09/14 16:15:00 EDT
2705    Tag: blume-20010914-x86fastfp
2706    Description:
2707    
2708    John committed some changes that Allen made, in particular a (hopefully)
2709    correctly working version of the x86-fp module.
2710    
2711    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
2712    flag to "true".  Everything seems to compile to a fixpoint ok, and
2713    "mandelbrot" speeds up by about 15%.
2714    
2715    ----------------------------------------------------------------------
2716    Name: Matthias Blume
2717    Date: 2001/09/13 11:20:00 EDT
2718    Tag: blume-20010913-minimal
2719    Description:
2720    
2721    1. Stefan Monnier's patch to fix a miscompilation problem that
2722       was brought to light by John Reppy's work on Moby.
2723    
2724    2. Implemented a minimal "structure Compiler" that contains just
2725       "version" and "architecture".  The minimal version will be
2726       available when the full version is not.  This is for backward-
2727       compatibility with code that wants to test Compiler.version.
2728    
2729    ----------------------------------------------------------------------
2730    Name: Matthias Blume
2731    Date: 2001/08/28 14:03:00 EDT
2732    Tag: blume-20010828-ml-lex
2733    Description:
2734    
2735    Fix for bug 1581, received from Neophytos Michael.
2736    
2737    ----------------------------------------------------------------------
2738    Name: Matthias Blume
2739    Date: 2001/08/27 11:20:00 EDT
2740    Tag: blume-20010827-readme11035
2741    Description:
2742    
2743    Fleshed out the README file for 110.35.
2744    
2745    ----------------------------------------------------------------------
2746    Name: Matthias Blume
2747    Date: 2001/08/24 17:10:00 EDT
2748    Tag: Release_110_35
2749    Description:
2750    
2751    New version number (110.35).  New bootfiles.
2752    
2753    ----------------------------------------------------------------------
2754    Name: Lal George
2755    Date: 2001/08/24 13:47:18 EDT 2001
2756    Tag: george-20010824-MLRISC-graphs
2757    Description:
2758    
2759     removed clusters from MLRISC completely and replaced with graphs.
2760    
2761    ----------------------------------------------------------------------
2762    Name: Matthias Blume
2763    Date: 2001/08/23 17:50:00 EDT
2764    Tag: blume-20010823-toplevel
2765    Description:
2766    
2767    - some reorganization of the code that implements various kinds of
2768      environments in the compiler (static, dynamic, symbolic, combined)
2769    - re-implemented the EnvRef module so that evalStream works properly
2770      (if the stream contains references to "use", "CM.make", etc.)
2771    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
2772    
2773    ----------------------------------------------------------------------
2774    Name: Matthias Blume
2775    Date: 2001/08/20 15:50 EDT
2776    Tag: blume20010820-slipup
2777    Description:
2778    
2779    I forgot to commit a few files.  Here they are...
2780    
2781    ----------------------------------------------------------------------
2782    Name: Matthias Blume
2783    Date: 2001/08/20 15:35:00 EDT
2784    Tag: blume-20010820-debugprof
2785    Description:
2786    
2787    !!!! NEW BOOTFILES !!!!
2788    
2789    This is another round of reorganizing the compiler sources.  This
2790    time the main goal was to factor out all the "instrumentation"
2791    passes (for profiling and backtracing) into their own library.
2792    The difficulty was to do it in such a way that it does not depend
2793    on elaborate.cm but only on elabdata.cm.
2794    
2795    Therefore there have been further changes to both elaborate.cm and
2796    elabdata.cm -- more "generic" things have been moved from the former
2797    to the latter.  As a result, I was forced to split the assignment
2798    of numbers indicating "primtyc"s into two portions: SML-generic and
2799    SML/NJ-specific.  Since it would have been awkward to maintain,
2800    I bit the bullet and actually _changed_ the mapping between these
2801    numbers and primtycs.  The bottom line of this is that you need
2802    a new set of bin- and bootfiles.
2803    
2804    I have built new bootfiles for all architectures, so doing a fresh
2805    checkout and config/install.sh should be all you need.
2806    
2807    The newly created library's name is
2808    
2809        $smlnj/viscomp/debugprof.cm
2810    
2811    and its sources live under
2812    
2813        src/compiler/DebugProf
2814    
2815    ----------------------------------------------------------------------
2816    Name: Matthias Blume
2817    Date: 2001/08/15 17:15:00 EDT
2818    Tag: blume-20010815-compreorg
2819    Description:
2820    
2821    This is a first cut at reorganizing the CM libraries that make up the
2822    core of the compiler.  The idea is to separate out pieces that could
2823    be used independently by tools, e.g., the parser, the typechecker, etc.
2824    
2825    The current status is a step in this direction, but it is not quite
2826    satisfactory yet.  Expect more changes in the future.
2827    
2828    Here is the current (new) organization...
2829    
2830        What used to be $smlnj/viscomp/core.cm is now divided into
2831        six CM libraries:
2832    
2833             $smlnj/viscomp/basics.cm
2834                           /parser.cm
2835                           /elabdata.cm
2836                           /elaborate.cm
2837                           /execute.cm
2838                           /core.cm
2839    
2840        The CM files for these libraries live under src/system/smlnj/viscomp.
2841        All these libraries are proxy libraries that contain precisely
2842        one CM library component.  Here are the locations of the components
2843        (all within the src/compiler tree):
2844    
2845             Basics/basics.cm
2846             Parse/parser.cm
2847             ElabData/elabdata.cm
2848             Elaborator/elaborate.cm
2849             Execution/execute.cm
2850             core.cm
2851    
2852         [This organization is the same that has been used already
2853         for a while for the architecture-specific parts of the visible
2854         compiler and for the old version of core.cm.]
2855    
2856         As you will notice, many source files have been moved from their
2857         respective original locations to a new home in one of the above
2858         subtrees.
2859    
2860         The division of labor between the new libraries is the following:
2861    
2862             basics.cm:
2863                - Simple, basic definitions that pertain to many (or all) of
2864                  the other libraries.
2865             parser.cm:
2866                - The SML parser, producing output of type Ast.dec.
2867                - The type family for Ast is also defined and exported here.
2868             elabdata.cm:
2869                - The datatypes that describe input and output of the elaborator.
2870                  This includes types, absyn, and static environments.
2871             elaborator.cm:
2872                - The SML/NJ type checker and elaborator.
2873                  This maps an Ast.dec (with a given static environment) to
2874                  an Absyn.dec (with a new static environment).
2875                - This libraries implements certain modules that used to be
2876                  structures as functors (to remove dependencies on FLINT).
2877             execute.cm:
2878                - Everything having to do with executing binary code objects.
2879                - Dynamic environments.
2880             core.cm:
2881                - SML/NJ-specific instantiations of the elaborator and MLRISC.
2882                - Top-level modules.
2883                - FLINT (this should eventually become its own library)
2884    
2885    Notes:
2886    
2887    I am not 100% happy with the way I separated the elaborator (and its
2888    data structures) from FLINT.  Two instances of the same problem:
2889    
2890        1. Data structures contain certain fields that carry FLINT-specific
2891           information.  I hacked around this using exn and the property list
2892           module from smlnj-lib.  But the fact that there are middle-end
2893           specific fields around at all is a bit annoying.
2894    
2895        2. The elaborator calculates certain FLINT-related information.  I tried
2896           to make this as abstract as I could using functorization, but, again,
2897           the fact that the elaborator has to perform calculations on behalf
2898           of the middle-end at all is not nice.
2899    
2900        3. Having to used exn and property lists is unfortunate because it
2901           weakens type checking.  The other alternative (parameterizing
2902           nearly *everything*) is not appealing, though.
2903    
2904    I removed the "rebinding =" warning hack because due to the new organization
2905    it was awkward to maintain it.  As a result, the compiler now issues some of
2906    these warnings when compiling init.cmi during bootstrap compilation. On
2907    the plus side, you also get a warning when you do, for example:
2908       val op = = Int32.+
2909    which was not the case up to now.
2910    
2911    I placed "assign" and "deref" into the _Core structure so that the
2912    code that deals with the "lazy" keyword can find them there.  This
2913    removes the need for having access to the primitive environment
2914    during elaboration.
2915    
2916    ----------------------------------------------------------------------
2917    Name: Matthias Blume
2918    Date: 2001/08/13
2919    Tag: blume-20010813-closures
2920    Description:
2921    
2922    This fix was sent to us by Zhong Shao.  It is supposed to improve the
2923    performance of certain loops by avoiding needless closure allocation.
2924    
2925    ----------------------------------------------------------------------
2926    Name: Lal George
2927    Date: 2001/07/31 10:03:23 EDT 2001
2928    Tag: george-20010731-x86-fmalloc
2929    Description: Fixed bug in x86 calls
2930    
2931        There was a bug where call instructions would mysteriously
2932        vanish. The call instruction had to be one that returned
2933        a floating point value.
2934    
2935    ----------------------------------------------------------------------
2936    Name: Lal George
2937    Date: 2001/07/19 16:36:29 EDT 2001
2938    Tag: george-20010719-simple-cells
2939    Description:
2940    
2941    I have dramatically simplified the interface for CELLS in MLRISC.
2942    
2943    In summary, the cells interface is broken up into three parts:
2944    
2945      1. CellsBasis : CELLS_BASIS
2946    
2947            CellsBasis is a top level structure and common for all
2948            architectures.  it contains the definitions of basic datatypes
2949            and utility  functions over these types.
2950    
2951      2. functor Cells() : CELLS
2952    
2953            Cells generates an interface for CELLS that incorporates the
2954            specific resources on the target architecture, such as the
2955            presence of special register classes, their number and size,
2956            and various useful substructures.
2957    
2958      3. <ARCH>CELLS
2959    
2960            e.g. SparcCells: SPARCCELLS
2961    
2962            <ARCH>CELLS usually contains additional bindings for special
2963            registers  on the architecture, such as:
2964    
2965                    val r0 : cell           (* register zero *)
2966                    val y : cell            (* Y register *)
2967                    val psr : cell          (* processor status register *)
2968                    ...
2969    
2970            The structure returned by applying the Cells functor is opened
2971            in this interface.
2972    
2973    The main implication of all this is that the datatypes for cells is
2974    split between CellsBasis and CELLS -- a fairly simple change for user
2975    code.
2976    
2977    In the old scheme the CELLS interface had a definitional binding of
2978    the form:
2979    
2980            signature CELLS = sig
2981    
2982               structure CellsBasis = CellsBasis
2983    
2984               ...
2985    
2986            end
2987    
2988    With all the sharing constraints that goes on in MLRISC, this old
2989    design  quickly leads to errors such as:
2990    
2991            "structure definition spec inside of sharing ... "
2992    
2993    
2994    and appears to require an unacceptable amount of sharing and where
2995    constraint hackery.
2996    
2997    I think this error message (the interaction of definitional specs and
2998    sharing) requires more explanation on our web page.
2999    
3000    ----------------------------------------------------------------------
3001    Name: Matthias Blume
3002    Date: 2001/07/19 15:00:00 EDT
3003    Tag: blume-20010719-libreorg
3004    Description:
3005    
3006    This update puts together a fairly extensive but straightforward change
3007    to the way the libraries that implement the interactive system are
3008    organized:
3009    
3010       The biggest change is the elimination of structure Compiler.  As a
3011       replacement for this structure, there is now a CM library
3012       (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
3013       that exports all the substructures of the original structure Compiler
3014       directly.  So instead of saying Compiler.Foo.bar one now simply
3015       says Foo.bar.  (The CM libraries actually export a collection of
3016       structures that is richer than the collection of substructures of
3017       structure Compiler.)
3018    
3019       To make the transition smooth, there is a separate library called
3020       $smlnj/compiler/compiler.cm which puts together and exports the
3021       original structure Compiler (or at least something very close to it).
3022    
3023       There are five members of the original structure Compiler
3024       that are not exported directly but which instead became members
3025       of a new structure Backend (described by signature BACKEND).  These are:
3026       structure Profile (: PROFILE), structure Compile (: COMPILE), structure
3027       Interact (: INTERACT), structure Machine (: MACHINE), and val
3028       architecture (: string).
3029    
3030       Structure Compiler.Version has become structure CompilerVersion.
3031    
3032       Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
3033       by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
3034       or x86, respectively.
3035       Each of these exports the same frontend structures that
3036       $smlnj/compiler.cm exports.  But they do not have a structure Backend
3037       and instead export some structure <Arch>Backend where <Arch> is Alpha32,
3038       Hppa, PPC, Sparc, or X86, respectively.
3039    
3040       Library $smlnj/compiler/all.cm exports the union of the exports of
3041       $smlnj/compiler/<arch>.cm
3042    
3043       There are no structures <Arch>Compiler anymore, use
3044       $smlnj/compiler/<arch>.cm instead.
3045    
3046       Library host-compiler-0.cm is gone.  Instead, the internal library
3047       that instantiates CM is now called cm0.cm.  Selection of the host
3048       compiler (backend) is no longer done here but. (Responsibility for it
3049       now lies with $smlnj/compiler/current.cm.  This seems to be more
3050       logical.)
3051    
3052       Many individual files have been moved or renamed.  Some files have
3053       been split into multiple files, and some "dead" files have been deleted.
3054    
3055    Aside from these changes to library organization, there are also changes
3056    to the way the code itself is organized:
3057    
3058       Structure Binfile has been re-implemented in such a way that it no
3059       longer needs any knowledge of the compiler.  It exclusively deals
3060       with the details of binfile layout.  It no longer invokes the
3061       compiler (for the purpose of creating new prospective binfile
3062       content), and it no longer has any knowledge of how to interpret
3063       pickles.
3064    
3065       Structure Compile (: COMPILE) has been stripped down to the bare
3066       essentials of compilation.  It no longer deals with linking/execution.
3067       The interface has been cleaned up considerably.
3068    
3069       Utility routines for dealing with linking and execution have been
3070       moved into their own substructures.
3071    
3072       (The ultimate goal of these changes is to provide a light-weight
3073       binfile loader/linker (at least for, e.g., stable libraries) that
3074       does not require CM or the compiler to be present.)
3075    
3076    CM documentation has been updated to reflect the changes to library
3077    organization.
3078    
3079    ----------------------------------------------------------------------
3080    Name: Matthias Blume
3081    Date: 2001/07/10 17:30:00 EDT
3082    Tag: Release_110_34
3083    Description:
3084    
3085    Minor tweak to 110.34 (re-tagged):
3086    
3087      - README.html file added to CVS repository
3088      - runtime compiles properly under FreeBSD 3.X and 4.X
3089    
3090    ----------------------------------------------------------------------
3091    Name: Matthias Blume
3092    Date: 2001/07/10 17:30:00 EDT
3093    Tag: Release_110_34
3094    Description:
3095    
3096    New version number (110.34). New bootfiles.
3097    
3098    ----------------------------------------------------------------------
3099    Name: Matthias Blume
3100    Date: 2001/07/09 16:00:00 EDT
3101    Tag: blume-20010709-more-varargs
3102    Description:
3103    
3104    I changed the handling of varargs in ml-nlffigen again:
3105    The ellipsis ... will now simply be ignored (with an accompanying warning).
3106    
3107    The immediate effect is that you can actually call a varargs function
3108    from ML -- but you can't actually supply any arguments beyond the ones
3109    specified explicitly.  (For example, you can call printf with its format
3110    string, but you cannot pass additional arguments.)
3111    
3112    This behavior is only marginally more useful than the one before, but
3113    it has the advantage that a function or, more importantly, a function
3114    type never gets dropped on the floor, thus avoiding follow-up problems with
3115    other types that refer to the offending one.
3116    
3117    ----------------------------------------------------------------------
3118    Name: Matthias Blume
3119    Date: 2001/07/09 11:25:00 EDT
3120    Tag: blume-20010709-varargs
3121    Description:
3122    
3123    1. ckit-lib.cm now exports structure Error
3124    2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
3125       with a warning accompanied by a source location.  Moreover, it
3126       merely skips the offending function or type and proceeds with the
3127       rest of its work.u  As a result, one can safely feed C code containing
3128       "..." to ml-nlffigen.
3129    3. There are some internal improvements to CM, providing slightly
3130       more general string substitutions in the tools subsystem.
3131    
3132    ----------------------------------------------------------------------
3133    Name: Matthias Blume
3134    Date: 2001/06/27 15:10:00 EDT
3135    Tag: blume-20010627-concur
3136    Description:
3137    
3138    Fixed a small bug in CM's handling of parallel compilation.
3139    (You could observe the bug by Control-C-interrupting an ordinary
3140    CMB.make or CM.stabilize and then attaching some compile servers.
3141    The result was that all of a sudden the previously interrupted
3142    compilation would continue on its own.  This was because of
3143    an over-optimization: CM did not bother to clean out certain queues
3144    when no servers were attached "anyway", resulting in the contents
3145    of these queues to grab control when new servers did get attached.)
3146    
3147    There is also another minor update to the CM manual.
3148    
3149    ----------------------------------------------------------------------
3150    Name: Matthias Blume
3151    Date: 2001/06/26 16:15:00 EDT
3152    Tag: blume-20010626-cmdoc
3153    Description:
3154    
3155    Minor typo fixed in CM manual (syntax diagram for libraries).
3156    
3157    ----------------------------------------------------------------------
3158    Name: Matthias Blume
3159    Date: 2001/06/25 22:55:00 EDT
3160    Tag: blume-20010625-x86pc
3161    Description:
3162    
3163    Fixed a nasty bug in the X86 assembly code that caused signal
3164    handlers to fail (crash) randomly.
3165    
3166    ----------------------------------------------------------------------
3167    Name: Matthias Blume
3168    Date: 2001/06/25 12:05:00 EDT
3169    Tag: blume-20010625-nlffigen
3170    Description:
3171    
3172    This update fixes a number of minor bugs in ml-nlffigen as reported by
3173    Nick Carter <nbc@andrew.cmu.edu>.
3174    
3175      1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
3176      2. Default names for generated files are now derived from the name of
3177         the C file *without its directory*.  In particular, this causes generated
3178         files to be placed locally even if the C file is in some system directory.
3179      3. Default names for generated signatures and structures are also derived
3180         from the C file name without its directory.  This avoids silly things
3181         like "structure GL/GL".
3182         (Other silly names are still possible because ml-nlffigen does not do
3183          a thorough check of whether generated names are legal ML identifiers.
3184          When in doubt, use command line arguments to force particular names.)
3185    
3186    ----------------------------------------------------------------------
3187    Name: Matthias Blume
3188    Date: 2001/06/21 12:25:00 EDT
3189    Tag: blume-20010621-eXene
3190    Description:
3191    
3192    eXene now compiles and (sort of) works again.
3193    
3194    The library name (for version > 110.33) is $/eXene.cm.
3195    
3196    I also added an new example in src/eXene/examples/nbody.  See the
3197    README file there for details.
3198    
3199    ----------------------------------------------------------------------
3200    Name: Matthias Blume
3201    Date: 2001/06/20 16:40:00 EDT
3202    Tag: blume-20010620-cml
3203    Description:
3204    
3205    CML now compiles and works again.
3206    
3207    Libraries (for version > 110.33):
3208    
3209      $cml/cml.cm            Main CML library.
3210      $cml/basis.cm          CML's version of $/basis.cm.
3211      $cml/cml-internal.cm   Internal helper library.
3212      $cml/core-cml.cm       Internal helper library.
3213      $cml-lib/trace-cml.cm  Tracing facility.
3214      $cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
3215    
3216    The installer (config/install.sh) has been taught how to properly
3217    install this stuff.
3218    
3219    ----------------------------------------------------------------------
3220    Name: Matthias Blume
3221    Date: 2001/06/19 17:55:00 EDT
3222    Tag: blume-20010619-instantiate
3223    Description:
3224    
3225    This un-breaks the fix for bug 1432.
3226    (The bug was originally fixed in 110.9 but I broke it again some
3227    time after that.)
3228    
3229    ----------------------------------------------------------------------
3230    Name: Matthias Blume
3231    Date: 2001/06/19 17:25:00 EDT
3232    Tag: blume-20010619-signals
3233    Description:
3234    
3235    This should (hopefully) fix the long-standing signal handling bug.
3236    (The runtime system was constructing a continuation record with an
3237    incorrect descriptor which would cause the GC to drop data on the floor...)
3238    
3239    ----------------------------------------------------------------------
3240    Name: Matthias Blume
3241    Date: 2001/06/15 15:05:00 EDT
3242    Tag: blume-20010615-moresparc
3243    Description:
3244    
3245    Here is a short late-hour update related to Sparc c-calls:
3246    
3247     -- made handling of double-word arguments a bit smarter
3248    
3249     -- instruction selection phase tries to collapse certain clumsily
3250        constructed ML-Trees; typical example:
3251    
3252            ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
3253    
3254        This currently has no further impact on SML/NJ since mlriscGen does
3255        not seem to generate such patterns in the first place, and c-calls
3256        (which did generate them in the beginning) has meanwhile been fixed
3257        so as to avoid them as well.
3258    
3259    ----------------------------------------------------------------------
3260    Name: Matthias Blume
3261    Date: 2001/06/15 15:05:00 EDT
3262    Tag: blume-20010615-sparc
3263    Description:
3264    
3265    The purpose of this update is to provide an implementation of NLFFI
3266    on Sparc machines.
3267    
3268    Here are the changes in detail:
3269    
3270       * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
3271       the Sparc implementation of the c-calls API.
3272       * The Sparc backend of SML/NJ has been modified to uniformely use %fp
3273       for accessing the ML frame.  Thus, we have a real frame pointer and
3274       can freely modify %sp without need for an omit-frame-ptr phase.
3275       The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
3276       this case.
3277       * ml-nlffigen has been taught to produce code for different architectures
3278       and calling conventions.
3279       * In a way similar to what was done in the x86 case, the Sparc
3280       backend uses its own specific extension to mltree.  (For example,
3281       it needs to be able to generate UNIMP instructions which are part
3282       of the calling convention.)
3283       * ml-nlffi-lib was reorganized to make it more modular (in particular,
3284       to make it easier to plug in new machine- and os-dependent parts).
3285    
3286    There are some other fairly unrelated bug fixes and cleanups as well:
3287    
3288       * I further hacked the .cm files for MLRISC tools (like MDLGen) so
3289       that they properly share their libraries with existing SML/NJ libraries.
3290       * I fixed a minor cosmetic bug in CM, supressing certain spurious
3291       follow-up error messages.
3292       * Updates to CM/CMB documentation.
3293    
3294    TODO items:
3295    
3296       * MLRISC should use a different register as its asmTemp on the Sparc.
3297         (The current %o2 is a really bad choice because it is part of the
3298         calling conventions, so things might interfere in unexpected ways.)
3299    
3300    ----------------------------------------------------------------------
3301    Name: Matthias Blume
3302    Date: 2001/06/07
3303    Tag: blume-20010607-calls
3304    Description:
3305    
3306    A number of internal changes related to C calls and calling conventions:
3307    
3308    1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
3309       number of bytes popped implicitly (by the callee).  In most cases
3310       this field is 0 but on x86/win32 it is some non-zero value.  This
3311       is information provided for the benefit of the "omit-frameptr" pass.
3312    2. The CALL instruction on the x86 carries a similar "pops" field.
3313       The instruction selection phase copies its value from the ML-Tree
3314       CALL statement.
3315    3. On all other architectures, the instruction selection phase checks
3316       whether "pops=0" and complains if not.
3317    4. The c-calls implementation for x86 now accepts two calling conventions:
3318       "ccall" and "stdcall".  When "ccall" is selected, the caller cleans
3319       up after the call and pops is set to 0.  For "stdcall", the caller
3320       does nothing, leaving the cleanup to the callee; pops is set to
3321       the number of bytes that were pushed onto the stack.
3322    5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
3323       distinguish between "ccall" and "stdcall".
3324    6. The UNIMP instruction has been added to the supported Sparc instruction
3325       set. (This is needed for implementing the official C calling convention
3326       on this architecture.)
3327    7. I fixed some of the .cm files under src/MLRISC/Tools to make them
3328       work with the latest CM.
3329    
3330    ----------------------------------------------------------------------
3331    Name: Matthias Blume
3332    Date: 2001/06/05 15:10:00 EDT
3333    Tag: blume-20010605-cm-index
3334    Description:
3335    
3336    0. The "lambdasplit" parameter for class "sml" in CM has been documented.
3337    
3338    1. CM can now generate "index files".  These are human-readable files
3339       that list on a per-.cm-file basis each toplevel symbol defined or
3340       imported.  The location of the index file for
3341       <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
3342       To enable index-file generation, set CM.Control.generate_index to true
3343       or export an environment-symbol: export CM_GENERATE_INDEX=true.
3344    
3345       The CM manual has been updated accordingly.
3346    
3347    2. I made some slight modifications to the c-calls API in MLRISC.
3348    
3349         a) There is now a callback to support saving/restoring of
3350            dedicated but caller-save registers around the actual call
3351            instruction.
3352         b) One can optionally specify a comment-annotation for the
3353            call instruction.
3354    
3355    3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
3356       (For example, the comment annotation shows the C prototype of
3357        the function being called.)
3358    
3359    ----------------------------------------------------------------------
3360    Name: Matthias Blume
3361    Date: 2001/06/01 13:30:00 EDT
3362    Tag: blume-20010601-nlffi-cleanup
3363    Description:
3364    
3365    This is mostly a cleanup of MLFFI stuff:
3366    
3367       - some signature files have been put into a more exposed place
3368       - the ugly 'f type parameter is gone (simplifies types tremendously!)
3369       - ml-nlffigen changed accordingly
3370       - tutorial updated
3371    
3372    Other changes:
3373    
3374       - author's affiliation in CM manual(s) updated
3375       - some more recognized keywords added to Allen's sml.sty
3376    
3377    ----------------------------------------------------------------------
3378    Name: Matthias Blume
3379    Date: 2001/05/25 15:30:00 EDT
3380    Tag: blume-20010525-iptr
3381    Description:
3382    
3383      - put the official 110.33-README (as it appears on the ftp server) under
3384        CVS
3385      - fixed a small bug related to incomplete pointer types in
3386        ml-nlffigen
3387      - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
3388        (it does not need the 'f type parameter)
3389    
3390    ----------------------------------------------------------------------
3391    Name: Matthias Blume
3392    Date: 2001/05/23 14:30:00 EDT
3393    Tag: Release_110_33
3394    Description:
3395    
3396    New version number (110.33).  New bootfiles.
3397    
3398    ----------------------------------------------------------------------
3399    Name: Matthias Blume
3400    Date: 2001/05/22 18:06:00 EDT
3401    Tag: blume-20010522-targets
3402    Description:
3403    
3404    Made install.sh use file config/targets.customized if it exists, falling
3405    back to config/targets if it doesn't.  This way one can have a customized
3406    version of the targets file without touching the "real thing", thus
3407    eliminating the constant fear of accidentally checking something bogus
3408    back into the CVS repository...  (File config/targets.customized must
3409    not be added to the repository!)
3410    
3411    ----------------------------------------------------------------------
3412    Name: Matthias Blume
3413    Date: 2001/05/22 16:30:00 EDT
3414    Tag: blume-20010522-minitut
3415    Description:
3416    
3417    1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
3418       struct returns.
3419    2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
3420       incomplete, preliminary documentation for NLFFI.
3421    
3422    ----------------------------------------------------------------------
3423    Name: Matthias Blume
3424    Date: 2001/05/14 11:30:00 EDT
3425    Tag: blume-20010514-script
3426    Description:
3427    
3428    Some bugs in install script fixed.
3429    
3430    In addition to that I also made a slight change to the NLFFI API:
3431    Functors generated by ml-nlffigen now take the dynamic library as a
3432    straight functor argument, not as a suspended one.  (The original
3433    functor code used to force the suspension right away anyway, so there
3434    was nothing gained by this complication of the interface.)
3435    
3436    ----------------------------------------------------------------------
3437    Name: Matthias Blume
3438    Date: 2001/05/11 14:35:00 EDT
3439    Tag: blume-20010511-ml-nlffi
3440    Description:
3441    
3442    I finally took the plunge and added my new FFI code to the main
3443    repository.  For x86-linux it is now ready for prime-time.
3444    
3445    There are two new subdirectories of "src":
3446    
3447      - ml-nlffi-lib:
3448           The utility library for programs using the FFI interface.
3449           Here is the implementation of $/c.cm and its associated low-level
3450           partners $/c-int.cm and $/memory.cm.
3451      - ml-nlffigen:
3452           A stand-alone program for generating ML glue code from C source
3453           code.
3454    
3455    Building ml-nlffigen requires $/ckit-lib.cm.
3456    
3457    The config/install.sh script has been updates to do the Right Thing
3458    (hopefully).
3459    
3460    Notice that the source tree for the C-Kit will not be put under "src"
3461    but directly under the installation root directory.  (This is the
3462    structure that currently exists on the CVS server when you check out
3463    module "sml".)  Fortunately, config/install.sh knows about this oddity.
3464    
3465    Bugs: No documentation yet.
3466    
3467    ----------------------------------------------------------------------
3468    Name: Matthias Blume
3469    Date: 2001/05/09 16:35:00 EDT
3470    Tag: blume-20010509-cpscontract
3471    Description:
3472    
3473    Fixed a bug in the accounting code in cpsopt/contract.sml.  (The
3474    wrapper/unwrapper elimination did not decrement usage counts and some
3475    dead variables got overlooked by the dead-up logic.)
3476    
3477    ----------------------------------------------------------------------
3478    Name: Lal George
3479    Date: 2001/05/08  17:26:09 EDT
3480    Tag: george-20010508-omit-frameptr
3481    Description:
3482    
3483    Changes to implement the omit-frame-pointer optimization to support
3484    raw C calls. For now, there is only support on the Intel x86, but
3485    other architectures will follow as more experience is gained with this.
3486    
3487    
3488    ----------------------------------------------------------------------
3489    Name: Matthias Blume
3490    Date: 2001/05/07 14:40:00 EDT
3491    Tag: blume-20010507-proxies
3492    Description:
3493    
3494    I made into "proxy libraries" all libraries that qualify for such a
3495    change.  (A qualifying library is a library that has another library or
3496    groups as its sole member and repeats that member's export list
3497    verbatim.  A proxy library avoids this repetition by omitting its export
3498    list, effectively inheriting the list that its (only) member exports.
3499    See the CM manual for more explanation.)
3500    The main effect is that explicit export lists for these libraries
3501    do not have to be kepts in sync, making maintenance a bit easier.
3502    
3503    I also added copyright notices to many .cm-files.
3504    
3505    Last but not least, I made a new set of bootfiles.
3506    
3507    ----------------------------------------------------------------------
3508    Name: Matthias Blume
3509    Date: 2001/05/04 17:00:00 EDT
3510    Tag: blume-20010504-cm-lsplit
3511    Description:
3512    
3513    0. John merged pending changes to $/smlnj-lib.cm
3514    
3515    1. Allen's previous change accidentally backed out of one of Lal's
3516       earlier changes.  I undid this mistake (re-introducing Lal's change).
3517    
3518    2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
3519       within the compiler where applicable.  There is some code simplification
3520       because of that.
3521    
3522    3. The "split" phase (in FLINT) is now part of the default list of phases.
3523       Compiler.Control.LambdaSplitting.* can be used to globally control the
3524       lambda-splitting (cross-module-inlining) engine.  In addition to that,
3525       it can now also be controlled on a per-source basis: CM has been taught
3526       a new tool parameter applicable to ML source files.
3527    
3528       - To turn lambda-splitting off completely:
3529            local open Compiler.Control.LambdaSplitting in
3530                val _ = set Off
3531            end
3532       - To make "no lambda-splitting" the global default (but allow per-source
3533         overriding); this is the initial setting:
3534            local open Compiler.Control.LambdaSplitting in
3535                val _ = set (Default NONE)
3536            end
3537       - To make "lambda-splitting with aggressiveness a" the global default
3538         (and allow per-source overriding):
3539            local open Compiler.Control.LambdaSplitting in
3540                val _ = set (Default (SOME a))
3541            end
3542    
3543       - To turn lambda-splitting off for a given ML souce file (say: a.sml)
3544         write (in the respective .cm-file):
3545            a.sml (lambdasplitting:off)
3546       - To turn lambda-splitting for a.sml on with minimal aggressiveness:
3547            a.sml (lambdasplitting:on)
3548       - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
3549         <a> is a decimal non-negative integer):
3550            a.sml (lambdasplitting:<a>)
3551       - To turn lambda-splitting for a.sml on with maximal aggressiveness:
3552            a.sml (lambdasplitting:infinity)
3553       - To use the global default for a.sml:
3554            a.sml (lambdasplitting:default)
3555         or simply
3556            a.sml
3557    
3558    ----------------------------------------------------------------------
3559    Name: Allen Leung
3560    Date: 2001/05/04 01:57:00 EDT
3561    Tag: leunga-20010504-sync
3562    Description:
3563    
3564      MLRISC features.
3565    
3566      1. Fix to CMPXCHG instructions.
3567      2. Changed RA interface to allow annotations in callbacks.
3568      3. Added a new method to the stream interface to allow annotations updates.
3569    
3570    ----------------------------------------------------------------------
3571    Name: Matthias Blume
3572    Date: 2001/05/01 11:45:00 EDT
3573    Tag: blume-20010501-pcedittmp
3574    Description:
3575    
3576    Changed install.sh to use the current working directory instead of
3577    /usr/tmp for a temporary file (pcedittmp).  The previous choice
3578    of /usr/tmp caused trouble with MacOS X because of file premission
3579    problems.
3580    
3581    ----------------------------------------------------------------------
3582    Name: Matthias Blume
3583    Date: 2001/04/20 11:10:00 EDT
3584    Tag: blume-20010420-inMLflag
3585    Description:
3586    
3587     - added vp_limitPtrMask to vproc-state.h
3588       (for use by the raw-C-calls mechanism to implement proper interrupt
3589        handling)
3590     - made the ML compiler aware of various data-structure offsets so it
3591       can generate code for accessing the vp_inML flag and vp_limitPtrMask
3592     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
3593       raw C-calls
3594    
3595    ----------------------------------------------------------------------
3596    Name: Lal George
3597    Date: 2001/04/20 09:15:28 EDT
3598    Tag: george-20010420-macosX
3599    Description:
3600    
3601     - Changes to port to Mac OS X; Darwin.
3602    
3603     - In the process I found that sqrt was broken on the PPC, because the
3604       fsqrt instruction is not implemented.
3605    
3606    ----------------------------------------------------------------------
3607    Name: Matthias Blume
3608    Date: 2001/04/18 12:45:00 EDT
3609    Tag: blume-20010418-ccalls
3610    Description:
3611    
3612     - fixed two off-by-4 errors in the x86-specific c-calls implementation
3613       (this bug prevented structure arguments containing pointers from being
3614        passed correctly)
3615     - changed the raw-C-call code in mlriscGen.sml in such a way that
3616       structure arguments are represented as a pointer to the beginning
3617       of the structure (instead of having a series of synthesized arguments,
3618       one for each structure member)
3619    
3620     - made makeml script's verbosity level configurable via environment
3621       variable (MAKEML_VERBOSITY)
3622    
3623     - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
3624       in rawmem-x86.sml; we are now using the real thing
3625    
3626    ----------------------------------------------------------------------
3627    Name: Matthias Blume
3628    Date: 2001/03/22 16:25:00 EST
3629    Tag: blume-20010322-bootfiles
3630    Description:
3631    
3632    Created a new set of bootfiles (for your automatic installation convenience).
3633    
3634    ----------------------------------------------------------------------
3635    Name: Matthias Blume
3636    Date: 2001/03/22 15:10:00 EST
3637    Tag: blume-20010322-rawmem-parcm
3638    Description:
3639    
3640    1. All "raw memory access" primitives for the new FFI are implemented now
3641       (at least on the x86).
3642    2. Some further cleanup of CM's parallel make mechanism.
3643    
3644    ----------------------------------------------------------------------
3645    Name: Matthias Blume
3646    Date: 2001/03/19 17:53:00 EST
3647    Tag: blume-20010319-parallel
3648    Description:
3649    
3650    Parallel make (using compile servers) now works again.
3651    
3652    To this end, CM.stabilize and CMB.make have been modified to work in
3653    two passes when compile servers are attached:
3654       1. Compile everything, do not perform stabilization; this pass
3655          uses compile servers
3656       2. Stabilize everything; this pass does not use compile servers
3657    If there are no compile servers, the two passes are combined into one
3658    (as before).  Splitting the passes increases the inherent parallelism
3659    in the dependency graph because the entire graph including all
3660    libraries is available at the same time.  This, in turn, improves
3661    server utilization.  The downside is that the master process will
3662    have to do some extra work after compilation is done (because for
3663    technical reasons it must re-read all the binfiles during stabilization).
3664    
3665    ----------------------------------------------------------------------
3666    Name: Matthias Blume
3667    Date: 2001/03/16 12:22:00 EST
3668    Tag: blume-20010316-bootfiles
3669    Description:
3670    
3671    Created a new set of bootfiles (for your automatic installation convenience).
3672    
3673    ----------------------------------------------------------------------
3674    Name: Matthias Blume
3675    Date: 2001/03/16 11:00:00 EST
3676    Tag: blume-20010316-MLTREE-fixup
3677    Description:
3678    
3679    This is a minor fixup for an (untagged) earlier commit by Allen.
3680    (A file was missing).
3681    
3682    ----------------------------------------------------------------------
3683    Name: Allen Leung
3684    Date: Mon Mar  5 18:54:57 EST 2001
3685    Tag: leunga-20010305-cut-support
3686    
3687    1. New support for alternative control-flow in MLTREE.
3688       Currently we support
3689    
3690          FLOW_TO(CALL ...., [k1,...,kn])
3691    
3692       This is needed for 'cuts to' in C-- and try/handle-like constructs
3693       in Moby
3694    
3695       New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
3696    
3697    2. Register Allocator
3698    
3699       Changes in interface [from Fermin, John]
3700    
3701    3. Alpha 8-bit SLL support [Fermin]
3702    
3703    4. All architectures
3704    
3705       A new module (ClusterExpandCopies) for expanding parallel copies.
3706    
3707    ----------------------------------------------------------------------
3708    Name: Allen Leung
3709    Date: 2001/02/27 23:07:00 EST
3710    Tag: leunga-20010227-minor-stuff
3711    
3712    1. Alpha bug fix for CMOVNE
3713    2. Handle mltree COND(..,FCMP ...,...)
3714    3. Bug fix in simplifier
3715    
3716    ----------------------------------------------------------------------
3717    Name: Matthias Blume
3718    Date: 2001/01/30 17:50:00 EST
3719    Tag: blume-20010130-sync
3720    Description:
3721    
3722    This is just a minor update to sync my devel branch with the main brach.
3723    The only visible change is the addition of some README files.
3724    
3725    ----------------------------------------------------------------------
3726    Name: Matthias Blume
3727    Date: 2001/01/12 23:30:00 JST
3728    Tag: blume-20010112-bootfiles
3729    Description:
3730    
3731    Made a new set of bootfiles that goes with the current state of the
3732    repository.
3733    
3734    ----------------------------------------------------------------------
3735    Name: Matthias Blume
3736    Date: 2001/01/12 21:20:00 JST
3737    Tag: blume-20010112-sync
3738    Description:
3739    
3740    I am just flushing out some minor changes that had accumulated in
3741    my private branch in order to sync with the main tree.  (This is
3742    mainly because I had CVS trouble when trying to merge _into_ my
3743    private branch.)
3744    
3745    Most people should be completely unaffected by this.
3746    
3747    ----------------------------------------------------------------------
3748    Name: Allen Leung
3749    Date: Thu Jan 11 21:03:00 EST 2001
3750    Tag: leunga-20010111-labexp=mltree
3751    Description:
3752    
3753    1.  Removed the type LabelExp and replace it by MLTree.
3754    2.  Rewritten mltree-simplify with the pattern matcher tool.
3755    3.  There were some bugs in alpha code generator which would break
3756        64-bit code generation.
3757    4.  Redo the tools to generate code with the
3758    5.  The CM files in MLRISC (and in src/system/smlnj/MLRISC)
3759        are now generated by perl scripts.
3760    
3761    ----------------------------------------------------------------------
3762    Name: Matthias Blume
3763    Date: 2001/01/10 21:55:00 JST
3764    Tag: blume-20010110-rcc
3765    Description:
3766    
3767    The RCC stuff now seems to work (but only on the x86).
3768    This required hacking of the c-calls interface (and -implementation) in
3769    MLRISC.
3770    
3771    Normal compiler users should be unaffected.
3772    
3773    ----------------------------------------------------------------------
3774    Name: Matthias Blume
3775    Date: 2001/01/09 01:20:00 JST
3776    Tag: blume-20010109-rcc
3777    Description:
3778    
3779    This is a fairly big patch, flushing out a large number of pending
3780    changes that I made to my development copy over the last couple of days.
3781    
3782    Of practical relevance at this moment is a workaround for a pickling
3783    bug that Allen ran into the other day.  The cause of the bug itself is
3784    still unknown and it might be hard to fix it properly, but the
3785    workaround has some merits of its own (namely somewhat reducing pickling
3786    overhead for certain libraries).  Therefore, I think this solution should
3787    be satisfactory at this time.
3788    
3789    The rest of the changes (i.e., the vast majority) has to do with my
3790    ongoing efforts of providing direct support for C function calls from
3791    ML.  At the moment there is a new primop "RAW_CCALL", typing magic
3792    in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
3793    case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
3794    translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
3795    handle RCC.
3796    
3797    The last part (the changes to mlriscGen.sml) are still known to be
3798    wrong on the x86 and not implemented on all other architectures.  But
3799    the infrastructure is in place. I had to change a few functor
3800    signatures in the backend to be able to route the CCalls interface
3801    from MLRISC there, and I had to specialize the mltree type (on the
3802    x86) to include the necessary extensions. (The extensions themselves
3803    were already there and redy to go in MLRISC/x86).
3804    
3805    Everything should be very happy as soon as someone helps me with
3806    mlriscGen.sml...
3807    
3808    In any case, nothing of this should matter to anyone as long as the
3809    new primop is not being used (which is going to be the case unless you
3810    find it where I hid it :). The rest of the compiler is completely
3811    unaffected.
3812    
3813    ----------------------------------------------------------------------
3814    Name: Matthias Blume
3815    Date: 2001/01/05 00:30:00 JST
3816    Tag: blume-20010105-primops
3817    Description:
3818    
3819    Added some experimental support for work that I am doing right now.
3820    These changes mostly concern added primops, but there is also a new
3821    experimental C library in the runtime system (but currently not enabled
3822    anywhere except on Linux/X86).
3823    
3824    In the course of adding primops (and playing with them), I discovered that
3825    Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
3826    broken.  (Zhong was very right he labeled this stuff as "major gross hack".)
3827    To recover, I made type information in INL_PRIM mandatory and changed
3828    prim.sml as well as built-in.sml accordingly.  The InLine structure now
3829    has complete, correct type information (i.e., no bottom types).
3830    
3831    Since all these changes mean that we need new binfiles, I also bumped the
3832    version number to 110.32.1.
3833    
3834    ----------------------------------------------------------------------
3835    Name: Matthias Blume
3836    Date: 2000/12/30 22:10:00 JST
3837    Tag: blume-20001230-various
3838    Description: