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 686, Tue Jul 18 08:58:50 2000 UTC revision 1385, Fri Sep 19 20:55:21 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: 2003/09/19 15:55:00 CDT
18    Tag: blume-20030919-cmdir
19    Description:
20    
21    1. new-install.sh -> install.sh
22    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
23    3. tweaked installer so that another name instead of .cm can be chosen
24       at install time (by setting the CM_DIR_ARC environment variable
25       during installation); once installation is complete, the name is
26       fixed
27    
28    ----------------------------------------------------------------------
29    Name: Matthias Blume (blume (at) tti - c (dot) org)
30    Date: 2003/09/18 16:00:00 CDT
31    Tag: blume-20030918-110_43_1
32    Description:
33    
34    Made a new interim version and bootfiles for developer's bootstrapping
35    convenience.
36    
37       110.43.1 -- NEW BOOTFILES
38    
39    ----------------------------------------------------------------------
40    Name: Matthias Blume (blume (at) tti - c (dot) org)
41    Date: 2003/09/18 15:20:00 CDT
42    Tag: blume-20030918-misc
43    Description:
44    
45    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
46       this seems to be controversial at the moment) and used those in
47       Posix.ProcEnv.times.
48    
49    2. Added Time.{from,to}Nanoseconds to Time.
50    
51    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
52       For example, fromLargeInt never needs to look at more than 3 "big
53       digits" to get its 53 bits of precision.
54    
55    ----------------------------------------------------------------------
56    Name: Matthias Blume (blume (at) tti - c (dot) org)
57    Date: 2003/09/17 16:30:00 CDT
58    Tag: blume-20030917-real32-slices
59    Description:
60    
61    Added an entry to the primitive environment
62    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
63    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
64    
65    Removed some of the "magic" constants in real64.sml and replaced them
66    with code that generates these values from their corresponding
67    integer counterparts.
68    
69    Made all(?) the slice-related changes to the Basis and made everything
70    compile again...
71    
72    ----------------------------------------------------------------------
73    Name: Matthias Blume (blume (at) tti - c (dot) org)
74    Date: 2003/09/15 17:45:00 CDT
75    Tag: blume-20030915-rbase
76    Description:
77    
78    Fixed bug in Real.fromLargeInt.
79    
80    ----------------------------------------------------------------------
81    Name: Matthias Blume (blume (at) tti - c (dot) org)
82    Date: 2003/09/13 18:11:00 CDT
83    Tag: blume-20030913-libinstall
84    Description:
85    
86    Minor bugfix in config/libinstall (set anchor with path to
87    standalone tool after installing it, otherwise libraries that
88    need ml-lex or ml-yacc won't compile the first time the installer
89    runs).
90    
91    ----------------------------------------------------------------------
92    Name: Matthias Blume (blume (at) tti - c (dot) org)
93    Date: 2003/09/12 11:45:00 CDT
94    Tag: blume-20030912-various
95    Description:
96    
97     - fixed bug in Real.toLargeInt
98     - fixed bug in Posix.ProcEnv.times
99     - changed inputLine functions to return an option
100     - minor installer improvements / bugfixes
101     - changed default @SMLalloc parameter for x86/celeron to 64k
102    
103    ----------------------------------------------------------------------
104    Name: Matthias Blume (blume (at) tti - c (dot) org)
105    Date: 2003/09/09 22:00:00 CDT
106    Tag: Release_110_43
107    Description:
108    
109        New working release 110.43.  New bootfiles.
110    
111    ----------------------------------------------------------------------
112    Name: Matthias Blume (blume (at) tti - c (dot) org)
113    Date: 2003/09/09 19:20:00 CDT
114    Tag: blume-20030909-installer
115    Description:
116    
117    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
118    Modified config/install.bat to take advantage of it.  Also modified
119    config/install.sh (and called it config/new-install.sh) to take advantage
120    of it on Unix systems.  (The SML code is (supposed to be) platform-
121    independent.)
122    
123    The installer can now install everything under Win32
124    as well as under *nix as long as it compiles.
125    
126    Other changes:
127    
128      - made CML compile again under Win32
129      - made eXene compile under Win32 (by providing a fake structure UnixSock
130        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
131      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
132        same as type int;  under Win32 it isn't
133      - fixed some slice-related problems in the win32-specific parts of CML
134      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
135        (under Win32, the volume name is case-insensitive, and the
136        OS.Path code compares volume names for equality)
137    
138    ----------------------------------------------------------------------
139    Name: Matthias Blume (blume (at) tti - c (dot) org)
140    Date: 2003/09/08 11:55:00 CDT
141    Tag: blume-20030908-fullpath
142    Description:
143    
144    Made Win32 version of OS.FileSys.fullPath return current directory
145    when given an empty string.  This is what the spec says, and incidentally,
146    CM depends on it.  (CM otherwise goes into an infinite loop in certain
147    cases when presented with the name of a non-existing .cm file.)
148    
149    ----------------------------------------------------------------------
150    Name: Matthias Blume (blume (at) tti - c (dot) org)
151    Date: 2003/09/04 16:30:00 CDT
152    Tag: blume-20030905-slices-etc
153    Description:
154    
155      1. Changed interface to vectors and arrays in Basis to match
156         (draft) Basis spec.
157      2. Added signatures and implementations of slices according to
158         Basis spec.
159      3. Edited source code throughout the system to make it compile again
160         under 1. and 2.  (In some cases code had to be added to have it
161         match the new signatures.)
162      4. MLRISC should be backward-compatible: the copies of the originals
163         of files that needed to change under 3. were retained, the .cm files
164         check the compiler version number and use old versions when
165         appropriate.
166      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
167         dirstream -> string option (in accordance with Basis spec).
168      6. When generating code that counts lines, ml-lex used function
169         CharVector.foldli, taking advantage of its old interface.
170         This has been replaced with the corresponding code from
171         CharVectorSlice.  (html-lex must be re-lexed!)
172      7. BitArray in smlnj-lib/Util has been extended/modified to match the
173         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
174      8. Removed temporary additions (fromInternal, toInternal) from the
175         (now obsolete) IntInf in smlnj-lib/Util.
176      9. Cleaned up structure Byte.
177     10. Added localOffset, scan, and fromString to Date (according to spec).
178         Cleaned/corrected implementation of Date.
179         (Still need to check for correctness; implement better canonicalizeDate.)
180     11. Added "scan" to signature IEEE_REAL.
181     12. Some improvements to IntInf [in particular: efficiency-hack for
182         mod and rem when second operand is 2 (for parity checks).]
183     13. Changed representation of type Time.time, using a single IntInf.int
184         value counting microseconds.  This considerably simplified the
185         implementation of structure Time.  We now support negative time
186         values; scan and fromString handle signs.
187     14. Functor PrimIO now takes two additional arguments (VectorSlice and
188         ArraySlice).
189    
190    ----------------------------------------------------------------------
191    Name: Matthias Blume (blume (at) tti - c (dot) org)
192    Date: 2003/08/28 17:00:00 CDT
193    Tag: blume-20030828-intinf
194    Description:
195    
196    This is a major update which comes with a version number bump
197    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
198    and an implementation of IntInf in the Basis.
199    
200    There are a fairly large number of related changes and updates throughout
201    the system:
202    
203    Basis:
204            - Implemented IntInf.
205            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
206            - Added some missing Real64 operations, most notably Real.toLargeInt.
207            - Added FixedInt as a synonym for Int32.
208    
209    compiler:
210            * Added support for a built-in intinf type.
211              - literals
212              - pattern matching
213              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
214              - overloading on literals and operations
215    
216            This required adding a primitive type intinf, some additional
217            primops, and implementations for several non-trivial intinf
218            operations in Core.  (The intinf type is completely abstract
219            to the compiler; all operations get delegated back to the Core.)
220    
221            * Intinf equality is handled by polyequal.  However, the compiler
222            does not print its usual warning in this case (since polyequal
223            is the right thing to do there).
224    
225            * Improved the organization of structure InlineT.
226    
227            * A word about conversion primops:
228            If conversions involving intinf do not cancel out during
229            CPS contract, then the compiler must insert calls to Core functions.
230            Since all core access must be resolved already during the FLINT
231            translate phase, it would be too late a the time of CPS contract
232            to add new Core calls.  For this reason, conversion primops
233            for intinf carry two arguments:  1. the numeric argument that
234            they are supposed to convert, and 2. the Core function that
235            can help with this conversion if necessary.  If CPS contract
236            eliminates a primop, then the associated Core function becomes
237            dead and goes away.  Intinf conversion primops that do not get
238            eliminated by CPS contract get rewritten into calls of their
239            core functions by a separate, new phase.
240    
241    interactive system:
242            - Control.Print.intinfDepth controls max length of intinf constants
243            being printed. (Analogous to Control.Print.stringDepth.)
244            - Cleanup in printutil and pputil: got rid of unused stuff and
245            duplicates; replaced some of the code with code that makes better
246            use of library functionality.
247    
248    CM:
249            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
250            are no longer silent.
251    
252    CKIT:
253            Fixed mismatched uses of Int32 and LargeInt.  I always decided
254            in favor of LargeInt -- which is now the same as IntInf.
255            CKIT-knowledgable people should check whether this is what's
256            intended and otherwise change things back to using Int32 or
257            FixedInt.
258    
259    Throughout the code:
260            Started using IntInf.int literals and built-in operations
261            (e.g., comparison with 0) where this seems appropriate.
262    
263    
264    ----------------------------------------------------------------------
265    Name: Dave MacQueen (dbm@cs.uchicago.edu)
266    Date: 2003/08/13 11:36:00 CDT
267    Tag: dbm-20030813-mcz-merge1
268    Description:
269    
270    Merging changes from the mcz-branch development branch into trunk.
271    These changes involve replacement of the emulated old prettyprinter
272    interface with direct use of the SML/NJ Lib PP library, and fixing
273    of a couple of bugs (895, 1186) relating to error messages.  A new
274    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
275    has been added.
276    
277    ----------------------------------------------------------------------
278    Name: Matthias Blume (blume (at) tti - c (dot) org)
279    Date: 2003/08/11 15:45:00 CDT
280    Tag: blume-20030811-windows
281    Description:
282    
283    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
284    
285        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
286    
287    This patch restores SML/NJ's ability to run under win32.  There are a
288    number of changes, including fixes for several bugs that had gone
289    unnoticed until now:
290    
291      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
292        trying to run the win32 version from within cygwin.)
293    
294      - There are a number of simple .bat scripts that substitute for their
295        corresponding Unix shell-scripts.  (See below.)
296    
297      - The internals of ml-build have been modified slightly.  The main
298        difference is that instead of calling ".link-sml" (or link-sml.bat)
299        using OS.Process.system, the ML process delegates this task back
300        to the script.  Otherwise problems arise in mixed environments such
301        as Cygwin where scripts look and work like Unix scripts, but
302        where OS.Process.system cannot run them.
303    
304      - In CM, the srcpath pickler used native pathname syntax -- which
305        is incorrect in the case of cross-compilation.  The new pickle format
306        is independent of platform-specific naming conventions.
307    
308      - Path configuration files (such as lib/pathconfig) can now choose
309        between native and standard syntax.  Placing a line of the form
310    
311            standard!
312    
313        into the file causes all subsequent paths to be interpreted using
314        CM standard pathname syntax (= Unix conventions); a line
315    
316            native!
317    
318        switches back to native style.  This was needed so that
319        path config files can be written portably, see src/system/pathconfig.
320    
321      - Runtime system:
322    
323          - win32-filesys.c:  get_file_time and set_file_time now
324            access modification time, not creation time.
325    
326          - I/O code made aware of new array representation.
327    
328          - Bug fixes in X86.prim.masm.
329    
330       - src/system/makeml made aware of win32. (For use under cygwin
331            and other Unix-environments for windows.)
332    
333       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
334         which caused BinIO.inputAll to fail consistently.
335    
336    .bat scripts:
337    
338       Windows .bat scripts assume that SMLNJ_HOME is defined.
339    
340       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
341         applications (sml, ml-yacc, ml-lex).
342       - ml-build.bat: analogous to ml-build.
343       - config\install.bat:  Analogous to config/install.sh.  This requires
344         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
345         (nmake etc. must be on the path, and vcvars32 must have been run.)
346         Moreover, sources for ml-lex and ml-yacc need to exist under src,
347         and the bootfile hierarchy must have been unpacked under
348         sml.boot.x86-win32.
349         The script is very primitive and does a poor job at error checking.
350         It only installs the base system, ml-lex, and ml-yacc.  No other
351         libraries are being installed (i.e., you get only those that
352         are part of the compiler.)
353       - link-sml.bat:  analogous to .link-sml, but not currently used
354    
355    Unrelated bug fixes:
356    
357      - ml-nlffigen now exports structures ST_* corresponding to incomplete
358        types.
359      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
360        otherwise.)
361    
362    ----------------------------------------------------------------------
363    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
364    Date: 2003/06/17
365    Tag: macqueen-20030617-bug895
366    Description:
367    
368    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
369    Tag will be used for new development branch (mcz-branch) for
370    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
371    
372    ----------------------------------------------------------------------
373    Name: Matthias Blume (blume (at) tti - c (dot) org)
374    Date: 2003/05/27 16:55:00 CDT
375    Tag: blume-20030527-polyeq
376    Description:
377    
378    Tried to eliminated most cases of polymorphic equality.
379    
380    ----------------------------------------------------------------------
381    Name: Matthias Blume (blume (at) tti - c (dot) org)
382    Date: 2003/05/21 17:45:00 CDT
383    Tag: blume-20030517-complete
384    Description:
385    
386    Two changes:
387    
388      1. Added a flag for controlling whether non-exhaustive bindings will
389         be treated as errors (default is false).
390      2. Cleaned up the *entire* source tree so that CMB.make goes through
391         without a single non-exhaustive match- or bind warning.
392    
393    ----------------------------------------------------------------------
394    Name: Matthias Blume (blume (at) tti - c (dot) org)
395    Date: 2003/05/17 10:20:00 CDT
396    Tag: blume-20030517-absyn
397    Description:
398    
399    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
400    
401       This mainly affects the quality of error messages.  However, some
402       of the code is now more straightforward than before. (Treatment of
403       the above four constructs in translate.sml is much simpler than
404       the "macro-expansion" that was going on before.  Plus, the mach-
405       compiler no longer gets invoked just to be able to compile an
406       if-expression.)
407    
408    2. The ErrorMsg.Error exception is now caught and absorbed by the
409       interactive loop.
410    
411    ----------------------------------------------------------------------
412    Name: Allen Leung
413    Date: 2003/05/16 13:05:00 CDT
414    Tag: leunga-20030516-cygwin-runtime
415    Description:
416    
417        Ported the runtime system to cygwin, which uses the unix
418    x86-unix bin files.  Missing/buggy features:
419    
420       o getnetbyname, getnetbyaddr: these functions seem to be missing in
421         the Cygwin library.
422       o Ctrl-C handling may be flaky.
423       o Windows system calls and Windows I/O are not supported.
424    
425       A new set of binfiles is located at:
426    
427         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
428    
429       This is only needed for bootstrapping the cygwin version of smlnj.
430       Other x86 versions can use the existing binfiles.
431    
432    ----------------------------------------------------------------------
433    Name: Matthias Blume
434    Date: 2003/04/08 15:42:00 CDT
435    Tag: blume-20030408-listpair
436    Description:
437    
438    1. Added a target 'mlrisc' to installer.
439    
440    2. Added missing elements to structure ListPair.
441    
442    ----------------------------------------------------------------------
443    Name: Allen Leung
444    Date: 2003/01/07 10:40:00 EST
445    Tag: leunga-20030107-int-rem
446    Description:
447    
448    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
449    The arguments to the SUBL instruction were swapped.
450    
451    ----------------------------------------------------------------------
452    Name: Matthias Blume
453    Date: 2002/12/12 16:25:00 EST
454    Tag: blume-20021212-risc-ra
455    Description:
456    
457    Fixed a serious bug in the rewrite code for FP spilling/reloading that
458    sent the RA into an infinite loop when floating point registers get
459    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
460    110.37 and 110.38.)
461    There was another set of potential problems related to the handling of
462    MLRISC annotations (but those did not yet cause real problems, apparently).
463    
464    ----------------------------------------------------------------------
465    Name: Matthias Blume
466    Date: 2002/12/06 22:40:00 EST
467    Tag: blume-20021206-cm-fileid
468    Description:
469    
470    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
471    This fixes the problem of CM getting confused by files that suddenly
472    change their identity (e.g., by getting unlinked and recreated by some
473    text editor such as vi).  There might be a better/cheaper/cleaner way
474    of doing this, but for now this will have to do.
475    
476    ----------------------------------------------------------------------
477    Name: Matthias Blume
478    Date: 2002/10/28 09:50:00 EST
479    Tag: blume-20021028-typecheck
480    Description:
481    
482    Exported structure Typecheck from $smlnj/viscomp/core.cm.
483    
484    ----------------------------------------------------------------------
485    Name: Matthias Blume
486    Date: 2002/10/17 09:10:00 EDT
487    Tag: Release_110_42
488    Description:
489    
490    In good old tradition, there has been a slight hiccup so that we have
491    to patch 110.42 after the fact.   The old release tag has been replaced
492    (see below).
493    
494    The change solves a problem with two competing approaches the
495    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
496    each other's way.
497    
498    This change only affects the runtime system code and the installer script.
499    (No new bootfiles.)
500    
501    ----------------------------------------------------------------------
502    Name: Matthias Blume
503    Date: 2002/10/16 12:00:00 EDT
504    Tag: Release_110_42_removed
505    Description:
506    
507    New working release.  New bootfiles.
508    
509    ----------------------------------------------------------------------
510    Name: Matthias Blume
511    Date: 2002/10/10 13:10:00 EDT
512    Tag: blume-20021010-ppc-divs
513    Description:
514    
515    The mltree operator DIVS must be implemented with an overflow check on
516    the PPC because the hardware indicates divide-by-zero using "overflow" as
517    well.
518    
519    ----------------------------------------------------------------------
520    Name: Matthias Blume
521    Date: 2002/07/23 11:45:00
522    Tag: blume-20020723-smlnj-home
523    Description:
524    
525    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
526    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
527    CM_PATHCONFIG is also set), the path configuration file is assumed
528    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
529    the entire tree to some other place and everything will "just work".
530    
531    (Companion commands such as ml-build and ml-makedepend also sense this
532    variable.)
533    
534    ----------------------------------------------------------------------
535    Name: Matthias Blume
536    Date: 2002/07/12 21:19:00 EDT
537    Tag: blume-20020712-liveness
538    Description:
539    
540    Exported two useful "step" functions from liveness module (MLRISC).
541    
542    ----------------------------------------------------------------------
543    Name: Matthias Blume
544    Date: 2002/07/05 16:00 EDT
545    Tag: Release_110_41
546    Description:
547    
548    New working release.  New bootfiles.
549    
550    ----------------------------------------------------------------------
551    Name: Matthias Blume
552    Date: 2002/07/05 10:25:00 EDT
553    Tag: blume-20020705-btimp
554    Description:
555    
556    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
557    other clients can set up backtracing support.
558    
559    ----------------------------------------------------------------------
560    Name: Matthias Blume
561    Date: 2002/06/25 17:23:00 EDT
562    Tag: blume-20020625-fpmax
563    Description:
564    
565    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
566    case.  (The sense of the isNaN test was reversed -- which made min and
567    max always return their first argument.)
568    
569    ----------------------------------------------------------------------
570    Name: Matthias Blume
571    Date: 2002/06/11
572    Tag: blume-20020611-unixpath
573    Description:
574    
575    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
576    
577    ----------------------------------------------------------------------
578    Name: Matthias Blume
579    Date: 2002/06/10 16:35:00 EDT
580    Tag: blume-20020610-ieeereal
581    Description:
582    
583    I back-ported my implementation of IEEEReal.fromString from the
584    idlbasis-devel branch so that we can test it.
585    
586    Another small change is that ppDec tries to give more information
587    than just "<sig>" in the case of functors.  However, this code is
588    broken in some mysterious way if the functor's body's signature
589    has not been declared by ascription but gets inferred from the
590    implementation.  This needs fixing...
591    
592    ----------------------------------------------------------------------
593    Name: Matthias Blume
594    Date: 2002/05/31
595    Tag: blume-20020531-btrace-mode
596    Description:
597    
598    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
599    the wayside when I switched over to using Controls everywhere.)
600    
601    ----------------------------------------------------------------------
602    Name: Lal George
603    Date: 2002/05/23 12:21:40 EDT
604    Tag: george-20020523-visual-labels
605    Description:
606    
607    Labels are now displayed in the graphical output to make
608    the fall-through and target blocks obvious.
609    
610    ----------------------------------------------------------------------
611    Name: Matthias Blume
612    Date: 2002/05/22 11:03:00 EDT
613    Tag: blume-20020522-shrink
614    Description:
615    
616    John tweaked yesterday's fix for 1131 to handle an out-of-memory
617    situation that comes up when allocating huge arrays.
618    
619    ----------------------------------------------------------------------
620    Name: Matthias Blume
621    Date: 2002/05/21 16:00:00 EDT
622    Tag: Release_110_40
623    Description:
624    
625    New working release (110.40).  New bootfiles.
626    
627    [Also: John Reppy fixed GC bug 1131.]
628    
629    ----------------------------------------------------------------------
630    Name: Matthias Blume
631    Date: 2002/05/21 12:35:00 EDT
632    Tag: blume-20020521-cmdoc
633    Description:
634    
635    CM documentation update.
636    
637    ----------------------------------------------------------------------
638    Name: Matthias Blume
639    Date: 2002/05/21 10:55:00 EDT
640    Tag: blume-20020521-misc
641    Description:
642    
643    - John tweaked runtime to be silent on heap export (except when
644      GC messages are on).
645    - I added a few more things (cross-compiling versions of CMB) to
646      config/preloads (as suggestions).
647    
648    ----------------------------------------------------------------------
649    Name: Matthias Blume
650    Date: 2002/05/20 22:25:00 EDT
651    Tag: blume-20020520-controls
652    Description:
653    
654    - Added ControlUtil structure to control-lib.cm.
655    - Use it throughout.
656    - Used Controls facility to define MLRISC controls (as opposed to
657      registering MLRISC control ref cells with Controls after the fact)
658    - Fixed messed-up controls priorities.
659    
660    * Removed again all the stuff from config/preloads that one wouldn't
661      be able to preload at the time the initial heap image is built.
662      (Many libraries, e.g., CML, do not exist yet at this time.  The
663       only libraries that can be preloaded via config/preloads are those
664       that come bundled with the bootfiles.)
665    
666    ----------------------------------------------------------------------
667    Name: Matthias Blume
668    Date: 2002/05/20 10:59:00 EDT
669    Tag: blume-20020520-preloads
670    Description:
671    
672    Added a lot of commented-out suggestions for things to be included
673    in config/preloads.
674    
675    ----------------------------------------------------------------------
676    Name: Allen Leung
677    Date: 2002/05/18 14:20:00 EDT
678    Tag: leunga-20020518-mdl
679    Description:
680    
681    o Made the mdl tool stuff compile and run again.
682    o I've disabled all the stuff that depends on RTL specifications; they
683      are all badly broken anyway.
684    
685    ----------------------------------------------------------------------
686    Name: Matthias Blume
687    Date: 2002/05/17 16:49:00 EDT
688    Tag: blume-20020517-controls
689    Description:
690    
691    0. John Reppy made several modifications to the SML/NJ library.
692       In particular, there is a shiny new controls-lib.cm.
693    
694    1. Pushed new controls interface through compiler so that everything
695       compiles again.
696    
697    2. Added FormatComb and FORMAT_COMB to the CML version of the
698       SML/NJ library (so that CML compiles again).
699    
700    3. Modified init scripts because XXX_DEFAULT environment variables
701       are no longer with us.  (Boot-time initialization is now done
702       using the same environment variables that are also used for
703       startup-time initialization of controls.)
704    
705    ----------------------------------------------------------------------
706    Name: Lal George
707    Date: 2002/05/15 09:20:10 EDT
708    Tag: george-20020515-pseudo-op-decls
709    Description:
710    
711            All pseudo-ops emitted before the first segment declaration
712    such as TEXT, DATA, and BSS directives are assumed to be global
713    declarations and are emitted first in the assembly file. This is
714    useful in a number of situations where one has pseudo-ops that are not
715    specific to any segment, and also works around the constraint that one
716    cannot have client pseudo-ops in the TEXT segment.
717    
718            Because no segment is associated with these declarations it is
719    an error to allocate any space or objects before the first segment
720    directive and an exception will be raised. However, we cannot make
721    this check for client pseudo-ops.
722    
723            These top level declarations are a field in the CFG graph_info.
724    In theory you can continue to add to this field after the CFG has been
725    built -- provided you know what you are doing;-)
726    
727    ----------------------------------------------------------------------
728    Name: Matthias Blume
729    Date: 2002/05/13 16:40:00 EDT
730    Tag: blume-20020513-pp-etc
731    Description:
732    
733    A few minor bugfixes:
734    
735      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
736        (Bogus printouts for redefined bindings still occur.  Compiler
737        bug should no longer occur now.  We need to redo the prettyprinter
738        from scratch.)
739    
740      - CM pathname printer now also adds escape sequences for ( and )
741    
742      - commend and docu fixes for ml-nlffi
743    
744    ----------------------------------------------------------------------
745    Name: Matthias Blume
746    Date: 2002/05/10 16:40:00 EDT
747    Tag: blume-20020510-erg-textio
748    Description:
749    
750    Applied the following bugfix provided by Emden Gansner:
751    
752        Output is corrupted when outputSubstr is used rather than output.
753    
754        The problem occurs when a substring
755    
756            ss = (s, dataStart, dataLen)
757    
758        where dataStart > 0, fills a stream buffer with avail bytes left.
759        avail bytes of s, starting at index dataStart, are copied into the
760        buffer, the buffer is flushed, and then the remaining dataLen-avail
761        bytes of ss are copied into the beginning of the buffer. Instead of
762        starting this copy at index dataStart+avail in s, the current code
763        starts the copy at index avail.
764    
765        Fix:
766        In text-io-fn.sml, change line 695 from
767             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
768        to
769             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
770    
771    ----------------------------------------------------------------------
772    Name: Matthias Blume
773    Date: 2002/04/12 13:55:00 EDT
774    Tag: blume-20020412-assyntax
775    Description:
776    
777    1. Grabbed newer assyntax.h from the XFree86 project.
778    2. Fiddled with how to compile X86.prim.asm without warnings.
779    3. (Very) Minor cleanup in CM.
780    
781    ----------------------------------------------------------------------
782    Name: Matthias Blume
783    Date: 2002/04/01 (no joke!) 17:07:00 EST
784    Tag: blume-20020401-x86div
785    Description:
786    
787    Added full support for div/mod/rem/quot on the x86, using the machine
788    instruction's two results (without clumsily recomputing the remainder)
789    directly where appropriate.
790    
791    Some more extensive power-of-two support was added to the x86 instruction
792    selector (avoiding expensive divs, mods, and muls where they can be
793    replaced with cheaper shifts and masks).  However, this sort of thing
794    ought to be done earlier, e.g., within the CPS optimizer so that
795    all architectures benefit from it.
796    
797    The compiler compiles to a fixed point, but changes might be somewhat
798    fragile nevertheless.  Please, report any strange things that you might
799    see wrt. div/mod/quot/rem...
800    
801    ----------------------------------------------------------------------
802    Name: Matthias Blume
803    Date: 2002/03/29 17:22:00
804    Tag: blume-20020329-div
805    Description:
806    
807    Fixed my broken div/mod logic.  Unfortunately, this means that the
808    inline code for div/mod now has one more comparison than before.
809    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
810    The problem was with quotient = 0, because that alone does not tell
811    us which way the rounding went.  One then has to look at whether
812    remainder and divisor have the same sign...  :(
813    
814    Anyway, I replaced the bootfiles with fresh ones...
815    
816    ----------------------------------------------------------------------
817    Name: Matthias Blume
818    Date: 2002/03/29 14:10:00 EST
819    Tag: blume-20020329-inlprims
820    Description:
821    
822    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
823    
824    Primops have changed. This means that the bin/boot-file formats have
825    changed as well.
826    
827    To make sure that there is no confusion, I made a new version.
828    
829    
830    CHANGES:
831    
832    * removed REMT from mltree (remainder should never overflow).
833    
834    * added primops to deal with divisions of all flavors to the frontend
835    
836    * handled these primops all the way through so they map to their respective
837      MLRISC support
838    
839    * used these primops in the implementation of Int, Int32, Word, Word32
840    
841    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
842    
843    * parameterized INLMIN, INLMAX, and INLABS by a numkind
844    
845    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
846      floating point
847    
848    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
849      and Real (but Real.abs maps to a separate floating-point-only primop)
850    
851    
852    TODO items:
853    
854    * Hacked Alpha32 instruction selection, disabling the selection of REMx
855      instructions because the machine instruction encoder cannot handle
856      them.  (Hppa, PPC, and Sparc instruction selection did not handle
857      REM in the first place, and REM is supported by the x86 machine coder.)
858    
859    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
860      selection phase.  (The two can be streamlined because the hardware
861      delivers both quotient and remainder at the same time anyway.)
862    
863    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
864      (Currently the behavior is inconsistent both across architectures and
865      wrt. the draft Basis spec.)
866    
867    * Word8 should eventually be handled natively, too.
868    
869    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
870      as if there currently is no execution path that could trigger it in
871      SML/NJ.  (The assumptions underlying functions arith and promotable do not
872      hold for things like multiplication and division.)
873    
874    ----------------------------------------------------------------------
875    Name: Matthias Blume
876    Date: 2002/03/27 16:27:00 EST
877    Tag: blume-20020327-mlrisc-divisions
878    Description:
879    
880    Added support for all four division operations (ML's div, mod, quot,
881    and rem) to MLRISC.  In the course of doing so, I also rationalized
882    the naming (no more annoying switch-around of DIV and QUOT), by
883    parameterizing the operation by div_rounding_mode (which can be either
884    DIV_TO_ZERO or DIV_TO_NEGINF).
885    
886    The generic MLTreeGen functor takes care of compiling all four
887    operations down to only round-to-zero div.
888    
889    Missing pieces:
890    
891      * Doing something smarter than relying on MLTreeGen on architectures
892        like, e.g., the x86 where hardware division delivers both quotient and
893        remainder at the same time.  With this, the implementation of the
894        round-to-neginf operations could be further streamlined.
895    
896      * Remove inlining support for div/mod/rem from the frontend and replace it
897        with primops that get carried through to the backend.  Do this for all
898        int and word types.
899    
900    ----------------------------------------------------------------------
901    Name: Matthias Blume
902    Date: 2002/03/25 17:25:00 EST
903    Tag: blume-20020325-divmod
904    Description:
905    
906    I improved (hopefully without breaking them) the implementation of Int.div,
907    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
908    advantage of the following observations:
909    
910      Let  q = x quot y      r = x rem y
911           d = x div  y      m = x mod y
912    
913    where "quot" is the round-to-zero version of integer division that
914    hardware usually provides.  Then we have:
915    
916         r = x - q * y        where neither the * nor the - will overflow
917         d = if q >= 0 orelse x = q * y then q else q - 1
918                              where neither the * nor the - will overflow
919         m = if q >= 0 orelse r = 0 then r else r + y
920                              where the + will not overflow
921    
922    This results in substantial simplification of the generated code.
923    The following table shows the number of CFG nodes and edges generated
924    for
925            fun f (x, y) = x OPER y
926            (* with OPER \in div, mod, quot, rem *)
927    
928    
929        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
930        --------------------------------------------------------
931         div |         24 |         39 |         12 |         16
932         mod |         41 |         71 |         12 |         16
933        quot |          8 |         10 |          8 |         10
934         rem |         10 |         14 |          8 |         10
935    
936    
937    ----------------------------------------------------------------------
938    Name: Matthias Blume
939    Date: 2002/03/25 22:06:00 EST
940    Tag: blume-20020325-cprotobug
941    Description:
942    
943    Fixed a bug in cproto (c prototype decoder).
944    
945    ----------------------------------------------------------------------
946    Name: Matthias Blume
947    Date: 2002/03/25 16:00:00 EST
948    Tag: blume-20020325-raw-primops
949    Description:
950    
951    I did some cleanup to Allen's new primop code and
952    replaced yesterday's bootfiles with new ones.
953    (But they are stored in the same place.)
954    
955    ----------------------------------------------------------------------
956    Name: Matthias Blume
957    Date: 2002/03/24 22:40:00 EST
958    Tag: blume-20020324-bootfiles
959    Description:
960    
961    Made the bootfiles that Allen asked for.
962    
963    ----------------------------------------------------------------------
964    Name: Allen Leung
965    Date: 2002/03/23 15:50:00 EST
966    Tag: leunga-20020323-flint-cps-rcc-primops
967    Description:
968    
969      1. Changes to FLINT primops:
970    
971        (* make a call to a C-function;
972         * The primop carries C function prototype information and specifies
973         * which of its (ML-) arguments are floating point. C prototype
974         * information is for use by the backend, ML information is for
975         * use by the CPS converter. *)
976      | RAW_CCALL of { c_proto: CTypes.c_proto,
977                       ml_args: ccall_type list,
978                       ml_res_opt: ccall_type option,
979                       reentrant : bool
980                     } option
981       (* Allocate uninitialized storage on the heap.
982        * The record is meant to hold short-lived C objects, i.e., they
983        * are not ML pointers.  With the tag, the representation is
984        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
985        *)
986      | RAW_RECORD of {tag:bool,sz:int}
987      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
988    
989      2.  These CPS primops are now overloaded:
990    
991           rawload of {kind:numkind}
992           rawstore of {kind:numkind}
993    
994          The one argument form is:
995    
996             rawload {kind} address
997    
998          The two argument form is:
999    
1000             rawload {kind} [ml object, byte-offset]
1001    
1002      3. RAW_CCALL/RCC now takes two extra arguments:
1003    
1004         a. The first is whether the C call is reentrant, i.e., whether
1005            ML state should be saved and restored.
1006         b. The second argument is a string argument specifying the name of
1007            library and the C function.
1008    
1009         These things are currently not handled in the code generator, yet.
1010    
1011      4. In CProto,
1012    
1013         An encoding type of "bool" means "ml object" and is mapped into
1014         C prototype of PTR.  Note that "bool" is different than "string",
1015         even though "string" is also mapped into PTR, because "bool"
1016         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1017    
1018      5. Pickler/unpicker
1019    
1020         Changed to handle RAW_RECORD and newest RAW_CCALL
1021    
1022      6. MLRiscGen,
1023    
1024         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1025         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1026            in the file CodeGen/cpscompile/cps-c-calls.sml
1027    
1028      7. Added the conditional move operator
1029    
1030             condmove of branch
1031    
1032         to cps.  Generation of this is still buggy so it is currently
1033         disabled.
1034    
1035    ----------------------------------------------------------------------
1036    Name: Lal George
1037    Date: 2002/03/22 14:18:25 EST
1038    Tag: george-20020322-cps-branch-prob
1039    Description:
1040    
1041    Implemented the Ball-Larus branch prediction-heuristics, and
1042    incorporated graphical viewers for control flow graphs.
1043    
1044    Ball-Larus Heuristics:
1045    ---------------------
1046    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1047    
1048    By design it uses the Dempster-Shafer theory for combining
1049    probabilities.  For example, in the function:
1050    
1051        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1052    
1053    the ball-larus heuristics predicts that the n=0 is unlikely
1054    (OH-heuristic), and the 'then' branch is unlikely because of the
1055    RH-heuristic -- giving the 'then' branch an even lower combined
1056    probability using the Dempster-Shafer theory.
1057    
1058    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1059    probability of the 'then' branch because of the loop in the else
1060    branch.
1061    
1062    
1063    Graphical Viewing:
1064    ------------------
1065    I merely plugged in Allen's graphical viewers into the compiler. The
1066    additional code is not much. At the top level, saying:
1067    
1068            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1069    
1070    will display the graphical view of the control flow graph just before
1071    back-patching.  daVinci must be in your path for this to work. If
1072    daVinci is not available, then the default viewer can be changed
1073    using:
1074    
1075            Control.MLRISC.getString "viewer"
1076    
1077    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1078    course, these viewers must be in your path.
1079    
1080    The above will display the compilation unit at the level of clusters,
1081    many of which are small, boring, and un-interesting. Also setting:
1082    
1083            Control.MLRISC.getInt "cfg-graphical-view_size"
1084    
1085    will display clusters that are larger than the value set by the above.
1086    
1087    
1088    ----------------------------------------------------------------------
1089    Name: Matthias Blume
1090    Date: 2002/03/21 22:20:00 EST
1091    Tag: blume-20020321-kmp-bugfix
1092    Description:
1093    
1094    Changed the interface to the KMP routine in PreString and fixed
1095    a minor bug in one place where it was used.
1096    
1097    ----------------------------------------------------------------------
1098    Name: Allen Leung
1099    Date: 2002/03/21 20:30:00 EST
1100    Tag: leunga-20020321-cfg
1101    Description:
1102    
1103      Fixed a potential problem in cfg edge splitting.
1104    
1105    ----------------------------------------------------------------------
1106    Name: Allen Leung
1107    Date: 2002/03/21 17:15:00 EST
1108    Tag: leunga-20020321-x86-fp-cfg
1109    Description:
1110    
1111      1. Recoded the buggy parts of x86-fp.
1112    
1113         a. All the block reordering code has been removed.
1114            We now depend on the block placement phases to do this work.
1115    
1116         b. Critical edge splitting code has been simplified and moved into the
1117            CFG modules, as where they belong.
1118    
1119         Both of these were quite buggy and complex.  The code is now much, much
1120         simpler.
1121    
1122      2. X86 backend.
1123    
1124         a. Added instructions for 64-bit support.  Instruction selection for
1125            64-bit has not been committed, however, since that
1126            requires changes to MLTREE which haven't been approved by
1127            Lal and John.
1128    
1129         b. Added support for FUCOMI and FUCOMIP when generating code for
1130            PentiumPro and above.  We only generate these instructions in
1131            the fast-fp mode.
1132    
1133         c. Added cases for JP and JNP in X86FreqProps.
1134    
1135      3. CFG
1136    
1137         CFG now has a bunch of methods for edge splitting and merging.
1138    
1139      4. Machine description.
1140    
1141         John's simplification of MLTREE_BASIS.fcond broke a few machine
1142         description things:
1143    
1144         rtl-build.{sig,sml} and hppa.mdl fixed.
1145    
1146         NOTE: the machine description stuff in the repository is still broken.
1147               Again, I can't put my fixes in because that involves
1148               changes to MLTREE.
1149    
1150    ----------------------------------------------------------------------
1151    Name: Matthias Blume
1152    Date: 2002/03/20 15:55:00 EST
1153    Tag: blume-20020320-kmp
1154    Description:
1155    
1156    Implemented Knuth-Morris-Pratt string matching in PreString and used
1157    it for String.isSubstring, Substring.isSubstring, and
1158    Substring.position.
1159    
1160    (Might need some stress-testing.  Simple examples worked fine.)
1161    
1162    ----------------------------------------------------------------------
1163    Name: Matthias Blume
1164    Date: 2002/03/19 16:37:00 EST
1165    Tag: blume-20020319-witnesses
1166    Description:
1167    
1168    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1169    
1170    This implements a generic mechanism for changing constness qualifiers
1171    anywhere within big C types without resorting to outright "casts".
1172    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1173    modify the constness at the outermost level.)
1174    The implementation of "convert" is based on the idea of "witness"
1175    values -- values that are not used by the operation but whose types
1176    "testify" to their applicability.  On the implementation side, "convert"
1177    is simply a projection (returning its second curried argument).  With
1178    cross-module inlining, it should not result in any machine code being
1179    generated.
1180    
1181    ----------------------------------------------------------------------
1182    Name: Matthias Blume
1183    Date: 2002/03/15 16:40:00 EST
1184    Tag: blume-20020315-basis
1185    Description:
1186    
1187    Provided (preliminary?) implementations for
1188    
1189      {String,Substring}.{concatWith,isSuffix,isSubstring}
1190    
1191    and
1192    
1193      Substring.full
1194    
1195    Those are in the Basis spec but they were missing in SML/NJ.
1196    
1197    ----------------------------------------------------------------------
1198    Name: Matthias Blume
1199    Date: 2002/03/14 21:30:00 EST
1200    Tag: blume-20020314-controls
1201    Description:
1202    
1203    Controls:
1204    ---------
1205    
1206    1. Factored out the recently-added Controls : CONTROLS stuff and put
1207       it into its own library $/controls-lib.cm.  The source tree for
1208       this is under src/smlnj-lib/Controls.
1209    
1210    2. Changed the names of types and functions in this interface, so they
1211       make a bit more "sense":
1212    
1213          module -> registry
1214          'a registry -> 'a group
1215    
1216    3. The interface now deals in ref cells only.  The getter/setter interface
1217       is (mostly) gone.
1218    
1219    4. Added a function that lets one register an already-existing ref cell.
1220    
1221    5. Made the corresponding modifications to the rest of the code so that
1222       everything compiles again.
1223    
1224    6. Changed the implementation of Controls.MLRISC back to something closer
1225       to the original.  In particular, this module (and therefore MLRISC)
1226       does not depend on Controls.  There now is some link-time code in
1227       int-sys.sml that registers the MLRISC controls with the Controls
1228       module.
1229    
1230    CM:
1231    ---
1232    
1233      * One can now specify the lambda-split aggressiveness in init.cmi.
1234    
1235    ----------------------------------------------------------------------
1236    Name: Allen Leung
1237    Date: 2002/03/13 17:30:00 EST
1238    Tag: leunga-20020313-x86-fp-unary
1239    Description:
1240    
1241    Bug fix for:
1242    
1243    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1244    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1245    > - fun f(x,(y,z)) = Real.~ y;
1246    > [autoloading]
1247    > [autoloading done]
1248    >       fchsl   (%eax), 184(%esp)
1249    > Error: MLRisc bug: X86MCEmitter.emitInstr
1250    >
1251    > uncaught exception Error
1252    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1253    
1254    The problem was that the code generator did not generate any fp registers
1255    in this case, and the ra didn't know that it needed to run the X86FP phase to
1256    translate the pseudo fp instruction.   This only happened with unary fp
1257    operators in certain situations.
1258    
1259    ----------------------------------------------------------------------
1260    Name: Matthias Blume
1261    Date: 2002/03/13 14:00:00 EST
1262    Tag: blume-20020313-overload-etc
1263    Description:
1264    
1265    1. Added _overload as a synonym for overload for backward compatibility.
1266       (Control.overloadKW must be true for either version to be accepted.)
1267    
1268    2. Fixed bug in install script that caused more things to be installed
1269       than what was requested in config/targets.
1270    
1271    3. Made CM aware of the (_)overload construct so that autoloading
1272       works.
1273    
1274    ----------------------------------------------------------------------
1275    Name: Matthias Blume
1276    Date: 2002/03/12 22:03:00 EST
1277    Tag: blume-20020312-url
1278    Description:
1279    
1280    Forgot to update BOOT and srcarchiveurl.
1281    
1282    ----------------------------------------------------------------------
1283    Name: Matthias Blume
1284    Date: 2002/03/12 17:30:00 EST
1285    Tag: blume-20020312-version110392
1286    Description:
1287    
1288    Yet another version number bump (because of small changes to the
1289    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1290    
1291    Changes:
1292    
1293      The new pid generation scheme described a few weeks ago was overly
1294      complicated.  I implemented a new mechanism that is simpler and
1295      provides a bit more "stability":  Once CM has seen a compilation
1296      unit, it keeps its identity constant (as long as you do not delete
1297      those crucial CM/GUID/* files).  This means that when you change
1298      an interface, compile, then go back to the old interface, and
1299      compile again, you arrive at the original pid.
1300    
1301      There now also is a mechanism that instructs CM to use the plain
1302      environment hash as a module's pid (effectively making its GUID
1303      the empty string).  For this, "noguid" must be specified as an
1304      option to the .sml file in question within its .cm file.
1305      This is most useful for code that is being generated by tools such
1306      as ml-nlffigen (because during development programmers tend to
1307      erase the tool's entire output directory tree including CM's cached
1308      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1309      revert to the old, broken behavior of SML/NJ, but in specific cases
1310      where there is no danger of interface confusion, its use is ok
1311      (I think).
1312    
1313      ml-nlffigen by default generates "noguid" annotations.  They can be
1314      turned off by specifying -guid in its command line.
1315    
1316    ----------------------------------------------------------------------
1317    Name: Lal George
1318    Date: 2002/03/12 12 14:42:36 EST
1319    Tag: george-20020312-frequency-computation
1320    Description:
1321    
1322    Integrated jump chaining and static block frequency into the
1323    compiler. More details and numbers later.
1324    
1325    ----------------------------------------------------------------------
1326    Name: Lal George
1327    Date: 2002/03/11 11 22:38:53 EST
1328    Tag: george-20020311-jump-chain-elim
1329    Description:
1330    
1331    Tested the jump chain elimination on all architectures (except the
1332    hppa).  This is on by default right now and is profitable for the
1333    alpha and x86, however, it may not be profitable for the sparc and ppc
1334    when compiling the compiler.
1335    
1336    The gc test will typically jump to a label at the end of the cluster,
1337    where there is another jump to an external cluster containing the actual
1338    code to invoke gc. This is to allow factoring of common gc invocation
1339    sequences. That is to say, we generate:
1340    
1341            f:
1342               testgc
1343               ja   L1      % jump if above to L1
1344    
1345            L1:
1346               jmp L2
1347    
1348    
1349    After jump chain elimination the 'ja L1' instructions is converted to
1350    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1351    up being implemented in their long form (if L2 is far away) using:
1352    
1353            jbe     L3      % jump if below or equal to L3
1354            jmp     L2
1355         L3:
1356            ...
1357    
1358    
1359    For large compilation units L2  may be far away.
1360    
1361    
1362    ----------------------------------------------------------------------
1363    Name: Matthias Blume
1364    Date: 2002/03/11 13:30:00 EST
1365    Tag: blume-20020311-mltreeeval
1366    Description:
1367    
1368    A functor parameter was missing.
1369    
1370    ----------------------------------------------------------------------
1371    Name: Allen Leung
1372    Date: 2002/03/11 10:30:00 EST
1373    Tag: leunga-20020311-runtime-string0
1374    Description:
1375    
1376       The representation of the empty string now points to a
1377    legal null terminated C string instead of unit.  It is now possible
1378    to convert an ML string into C string with InlineT.CharVector.getData.
1379    This compiles into one single machine instruction.
1380    
1381    ----------------------------------------------------------------------
1382    Name: Allen Leung
1383    Date: 2002/03/10 23:55:00 EST
1384    Tag: leunga-20020310-x86-call
1385    Description:
1386    
1387       Added machine generation for CALL instruction (relative displacement mode)
1388    
1389    ----------------------------------------------------------------------
1390    Name: Matthias Blume
1391    Date: 2002/03/08 16:05:00
1392    Tag: blume-20020308-entrypoints
1393    Description:
1394    
1395    Version number bumped to 110.39.1.  NEW BOOTFILES!
1396    
1397    Entrypoints: non-zero offset into a code object where execution should begin.
1398    
1399    - Added the notion of an entrypoint to CodeObj.
1400    - Added reading/writing of entrypoint info to Binfile.
1401    - Made runtime system bootloader aware of entrypoints.
1402    - Use the address of the label of the first function given to mlriscGen
1403      as the entrypoint.  This address is currently always 0, but it will
1404      not be 0 once we turn on block placement.
1405    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
1406      with entry points) from mlriscGen.
1407    
1408    ----------------------------------------------------------------------
1409    Name: Allen Leung
1410    Date: 2002/03/07 20:45:00 EST
1411    Tag: leunga-20020307-x86-cmov
1412    Description:
1413    
1414       Bug fixes for CMOVcc on x86.
1415    
1416       1. Added machine code generation for CMOVcc
1417       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1418       3. CMOVcc cannot have an immediate operand as argument.
1419    
1420    ----------------------------------------------------------------------
1421    Name: Matthias Blume
1422    Date: 2002/03/07 16:15:00 EST
1423    Tag: blume-20020307-controls
1424    Description:
1425    
1426    This is a very large but mostly boring patch which makes (almost)
1427    every tuneable compiler knob (i.e., pretty much everything under
1428    Control.* plus a few other things) configurable via both the command
1429    line and environment variables in the style CM did its configuration
1430    until now.
1431    
1432    Try starting sml with '-h' (or, if you are brave, '-H')
1433    
1434    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1435    implements the underlying generic mechanism.
1436    
1437    The interface to some of the existing such facilities has changed somewhat.
1438    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1439    (The getFoo interface is still there for backward-compatibility, but its
1440    use is deprecated.)
1441    
1442    The ml-build script passes -Cxxx=yyy command-line arguments through so
1443    that one can now twiddle the compiler settings when using this "batch"
1444    compiler.
1445    
1446    TODO items:
1447    
1448    We should go through and throw out all controls that are no longer
1449    connected to anything.  Moreover, we should go through and provide
1450    meaningful (and correct!) documentation strings for those controls
1451    that still are connected.
1452    
1453    Currently, multiple calls to Controls.new are accepted (only the first
1454    has any effect).  Eventually we should make sure that every control
1455    is being made (via Controls.new) exactly once.  Future access can then
1456    be done using Controls.acc.
1457    
1458    Finally, it would probably be a good idea to use the getter-setter
1459    interface to controls rather than ref cells.  For the time being, both
1460    styles are provided by the Controls module, but getter-setter pairs are
1461    better if thread-safety is of any concern because they can be wrapped.
1462    
1463    *****************************************
1464    
1465    One bug fix: The function blockPlacement in three of the MLRISC
1466    backpatch files used to be hard-wired to one of two possibilities at
1467    link time (according to the value of the placementFlag).  But (I
1468    think) it should rather sense the flag every time.
1469    
1470    *****************************************
1471    
1472    Other assorted changes (by other people who did not supply a HISTORY entry):
1473    
1474    1. the cross-module inliner now works much better (Monnier)
1475    2. representation of weights, frequencies, and probabilities in MLRISC
1476       changed in preparation of using those for weighted block placement
1477       (Reppy, George)
1478    
1479    ----------------------------------------------------------------------
1480    Name: Lal George
1481    Date: 2002/03/07 14:44:24 EST 2002
1482    Tag: george-20020307-weighted-block-placement
1483    
1484    Tested the weighted block placement optimization on all architectures
1485    (except the hppa) using AMPL to generate the block and edge frequencies.
1486    Changes were required in the machine properties to correctly
1487    categorize trap instructions. There is an MLRISC flag
1488    "weighted-block-placement" that can be used to enable weighted block
1489    placement, but this will be ineffective without block/edge
1490    frequencies (coming soon).
1491    
1492    
1493    ----------------------------------------------------------------------
1494    Name: Lal George
1495    Date: 2002/03/05 17:24:48 EST
1496    Tag: george-20020305-linkage-cluster
1497    
1498    In order to support the block placement optimization, a new cluster
1499    is generated as the very first cluster (called the linkage cluster).
1500    It contains a single jump to the 'real' entry point for the compilation
1501    unit. Block placement has no effect on the linkage cluster itself, but
1502    all the other clusters  have full freedom in the manner in which they
1503    reorder blocks or functions.
1504    
1505    On the x86 the typical linkage code that is generated is:
1506       ----------------------
1507            .align 2
1508       L0:
1509            addl    $L1-L0, 72(%esp)
1510            jmp     L1
1511    
1512    
1513            .align  2
1514       L1:
1515       ----------------------
1516    
1517    72(%esp) is the memory location for the stdlink register. This
1518    must contain the address of the CPS function being called. In the
1519    above example, it contains the address of  L0; before
1520    calling L1 (the real entry point for the compilation unit), it
1521    must contain the address for L1, and hence
1522    
1523            addl $L1-L0, 72(%esp)
1524    
1525    I have tested this on all architectures except the hppa.The increase
1526    in code size is of course negligible
1527    
1528    ----------------------------------------------------------------------
1529    Name: Allen Leung
1530    Date: 2002/03/03 13:20:00 EST
1531    Tag: leunga-20020303-mlrisc-tools
1532    
1533      Added #[ ... ] expressions to mlrisc tools
1534    
1535    ----------------------------------------------------------------------
1536    Name: Matthias Blume
1537    Date: 2002/02/27 12:29:00 EST
1538    Tag: blume-20020227-cdebug
1539    Description:
1540    
1541    - made types in structure C and C_Debug to be equal
1542    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1543    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1544    
1545    ----------------------------------------------------------------------
1546    Name: Matthias Blume
1547    Date: 2002/02/26 12:00:00 EST
1548    Tag: blume-20020226-ffi
1549    Description:
1550    
1551    1. Fixed a minor bug in CM's "noweb" tool:
1552       If numbering is turned off, then truly don't number (i.e., do not
1553       supply the -L option to noweb).  The previous behavior was to supply
1554       -L'' -- which caused noweb to use the "default" line numbering scheme.
1555       Thanks to Chris Richards for pointing this out (and supplying the fix).
1556    
1557    2. Once again, I reworked some aspects of the FFI:
1558    
1559       A. The incomplete/complete type business:
1560    
1561       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
1562         gone!
1563       - ML types representing an incomplete type are now *equal* to
1564         ML types representing their corresponding complete types (just like
1565         in C).  This is still safe because ml-nlffigen will not generate
1566         RTTI for incomplete types, nor will it generate functions that
1567         require access to such RTTI.   But when ML code generated from both
1568         incomplete and complete versions of the C type meet, the ML types
1569         are trivially interoperable.
1570    
1571         NOTE:  These changes restore the full generality of the translation
1572         (which was previously lost when I eliminated functorization)!
1573    
1574       B. Enum types:
1575    
1576       - Structure C now has a type constructor "enum" that is similar to
1577         how the "su" constructor works.  However, "enum" is not a phantom
1578         type because each "T enum" has values (and is isomorphic to
1579         MLRep.Signed.int).
1580       - There are generic access operations for enum objects (using
1581         MLRep.Signed.int).
1582       - ml-nlffigen will generate a structure E_foo for each "enum foo".
1583         * The structure contains the definition of type "mlrep" (the ML-side
1584         representation type of the enum).  Normally, mlrep is the same
1585         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
1586         then mlrep will be defined as a datatype -- thus facilitating
1587         pattern matching on mlrep values.
1588         ("-ec" will be suppressed if there are duplicate values in an
1589          enumeration.)
1590         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
1591         will be generated for each C enum constant xxx.
1592         * Conversion functions m2i and i2m convert between mlrep and
1593         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
1594         * Coversion functions c and ml convert between mlrep and "tag enum".
1595         * Access functions (get/set) fetch and store mlrep values.
1596       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
1597         enumerations are merged into one single enumeration represented by
1598         structure E_'.
1599    
1600    ----------------------------------------------------------------------
1601    Name: Allen Leung
1602    Date: 2002/02/25 04:45:00 EST
1603    Tag: leunga-20020225-cps-spill
1604    
1605    This is a new implementation of the CPS spill phase.
1606    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
1607    In case of problems, replace it with the old file spill.sml
1608    
1609    The current compiler runs into some serious performance problems when
1610    constructing a large record.  This can happen when we try to compile a
1611    structure with many items.  Even a very simple structure like the following
1612    makes the compiler slow down.
1613    
1614        structure Foo = struct
1615           val x_1 = 0w1 : Word32.int
1616           val x_2 = 0w2 : Word32.int
1617           val x_3 = 0w3 : Word32.int
1618           ...
1619           val x_N = 0wN : Word32.int
1620        end
1621    
1622    The following table shows the compile time, from N=1000 to N=4000,
1623    with the old compiler:
1624    
1625    N
1626    1000   CPS 100 spill                           0.04u  0.00s  0.00g
1627           MLRISC ra                               0.06u  0.00s  0.05g
1628              (spills = 0 reloads = 0)
1629           TOTAL                                   0.63u  0.07s  0.21g
1630    
1631    1100   CPS 100 spill                           8.25u  0.32s  0.64g
1632           MLRISC ra                               5.68u  0.59s  3.93g
1633              (spills = 0 reloads = 0)
1634           TOTAL                                   14.71u  0.99s  4.81g
1635    
1636    1500   CPS 100 spill                           58.55u  2.34s  1.74g
1637           MLRISC ra                               5.54u  0.65s  3.91g
1638              (spills = 543 reloads = 1082)
1639           TOTAL                                   65.40u  3.13s  6.00g
1640    
1641    2000   CPS 100 spill                           126.69u  4.84s  3.08g
1642           MLRISC ra                               0.80u  0.10s  0.55g
1643              (spills = 42 reloads = 84)
1644           TOTAL                                   129.42u  5.10s  4.13g
1645    
1646    3000   CPS 100 spill                           675.59u  19.03s  11.64g
1647           MLRISC ra                               2.69u  0.27s  1.38g
1648              (spills = 62 reloads = 124)
1649           TOTAL                                   682.48u  19.61s  13.99g
1650    
1651    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
1652           MLRISC ra                               4.96u  0.27s  2.72g
1653              (spills = 85 reloads = 170)
1654           TOTAL                                   2375.26u  57.21s  48.00g
1655    
1656    As you can see the old cps spill module suffers from some serious
1657    performance problem.  But since I cannot decipher the old code fully,
1658    instead of patching the problems up, I'm reimplementing it
1659    with a different algorithm.  The new code is more modular,
1660    smaller when compiled, and substantially faster
1661    (O(n log n) time and O(n) space).  Timing of the new spill module:
1662    
1663    4000  CPS 100 spill                           0.02u  0.00s  0.00g
1664          MLRISC ra                               0.25u  0.02s  0.15g
1665             (spills=1 reloads=3)
1666          TOTAL                                   7.74u  0.34s  1.62g
1667    
1668    Implementation details:
1669    
1670    As far as I can tell, the purpose of the CPS spill module is to make sure the
1671    number of live variables at any program point (the bandwidth)
1672    does not exceed a certain limit, which is determined by the
1673    size of the spill area.
1674    
1675    When the bandwidth is too large, we decrease the register pressure by
1676    packing live variables into spill records.  How we achieve this is
1677    completely different than what we did in the old code.
1678    
1679    First, there is something about the MLRiscGen code generator
1680    that we should be aware of:
1681    
1682    o MLRiscGen performs code motion!
1683    
1684       In particular, it will move floating point computations and
1685       address computations involving only the heap pointer to
1686       their use sites (if there is only a single use).
1687       What this means is that if we have a CPS record construction
1688       statement
1689    
1690           RECORD(k,vl,w,e)
1691    
1692       we should never count the new record address w as live if w
1693       has only one use (which is often the case).
1694    
1695       We should do something similar to floating point, but the transformation
1696       there is much more complex, so I won't deal with that.
1697    
1698    Secondly, there are now two new cps primops at our disposal:
1699    
1700     1. rawrecord of record_kind option
1701        This pure operator allocates some uninitialized storage from the heap.
1702        There are two forms:
1703    
1704         rawrecord NONE [INT n]  allocates a tagless record of length n
1705         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
1706                                     and initializes the tag.
1707    
1708     2. rawupdate of cty
1709          rawupdate cty (v,i,x)
1710          Assigns to x to the ith component of record v.
1711          The storelist is not updated.
1712    
1713    We use these new primops for both spilling and increment record construction.
1714    
1715     1. Spilling.
1716    
1717        This is implemented with a linear scan algorithm (but generalized
1718        to trees).  The algorithm will create a single spill record at the
1719        beginning of the cps function and use rawupdate to spill to it,
1720        and SELECT or SELp to reload from it.  So both spills and reloads
1721        are fine-grain operations.  In contrast, in the old algorithm
1722        "spills" have to be bundled together in records.
1723    
1724        Ideally, we should sink the spill record construction to where
1725        it is needed.  We can even split the spill record into multiple ones
1726        at the places where they are needed.  But CPS is not a good
1727        representation for global code motion, so I'll keep it simple and
1728        am not attempting this.
1729    
1730     2. Incremental record construction (aka record splitting).
1731    
1732        Long records with many component values which are simulatenously live
1733        (recall that single use record addresses are not considered to
1734         be live) are constructed with rawrecord and rawupdate.
1735        We allocate space on the heap with rawrecord first, then gradually
1736        fill it in with rawupdate.  This is the technique suggested to me
1737        by Matthias.
1738    
1739        Some restrictions on when this is applicable:
1740        1. It is not a VECTOR record.  The code generator currently does not handle
1741           this case. VECTOR record uses double indirection like arrays.
1742        2. All the record component values are defined in the same "basic block"
1743           as the record constructor.  This is to prevent speculative
1744           record construction.
1745    
1746    ----------------------------------------------------------------------
1747    Name: Allen Leung
1748    Date: 2002/02/22 01:02:00 EST
1749    Tag: leunga-20020222-mlrisc-tools
1750    
1751    Minor bug fixes in the parser and rewriter
1752    
1753    ----------------------------------------------------------------------
1754    Name: Allen Leung
1755    Date: 2002/02/21 20:20:00 EST
1756    Tag: leunga-20020221-peephole
1757    
1758    Regenerated the peephole files.  Some contained typos in the specification
1759    and some didn't compile because of pretty printing bugs in the old version
1760    of 'nowhere'.
1761    
1762    ----------------------------------------------------------------------
1763    Name: Allen Leung
1764    Date: 2002/02/19 20:20:00 EST
1765    Tag: leunga-20020219-mlrisc-tools
1766    Description:
1767    
1768       Minor bug fixes to the mlrisc-tools library:
1769    
1770       1.  Fixed up parsing colon suffixed keywords
1771       2.  Added the ability to shut the error messages up
1772       3.  Reimplemented the pretty printer and fixed up/improved
1773           the pretty printing of handle and -> types.
1774       4.  Fixed up generation of literal symbols in the nowhere tool.
1775       5.  Added some SML keywords to to sml.sty
1776    
1777    ----------------------------------------------------------------------
1778    Name: Matthias Blume
1779    Date: 2002/02/19 16:20:00 EST
1780    Tag: blume-20020219-cmffi
1781    Description:
1782    
1783    A wild mix of changes, some minor, some major:
1784    
1785    * All C FFI-related libraries are now anchored under $c:
1786        $/c.cm      --> $c/c.cm
1787        $/c-int.cm  --> $c/internals/c-int.cm
1788        $/memory.cm --> $c/memory/memory.cm
1789    
1790    * "make" tool (in CM) now treats its argument pathname slightly
1791      differently:
1792        1. If the native expansion is an absolute name, then before invoking
1793           the "make" command on it, CM will apply OS.Path.mkRelative
1794           (with relativeTo = OS.FileSys.getDir()) to it.
1795        2. The argument will be passed through to subsequent phases of CM
1796           processing without "going native".  In particular, if the argument
1797           was an anchored path, then "make" will not lose track of that anchor.
1798    
1799    * Compiler backends now "know" their respective C calling conventions
1800      instead of having to be told about it by ml-nlffigen.  This relieves
1801      ml-nlffigen from one of its burdens.
1802    
1803    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
1804    
1805    * Export C_DEBUG and C_Debug from $c/c.cm.
1806    
1807    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
1808      subtyping relationship between incomplete pointers and their complete
1809      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
1810      with the convention of instantiating 'o with ('t, 'c) obj whenever
1811      the pointer target type is complete.  In the incomplete case, 'o
1812      will be instantiated with some "'c iobj" -- a type obtained by
1813      using one of the functors PointerToIncompleteType or PointerToCompleteType.
1814    
1815      Operations that work on both incomplete and complete pointer types are
1816      typed as taking an 'o ptr while operations that require the target to
1817      be known are typed as taking some ('t, 'c) obj ptr.
1818    
1819      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
1820      where void is an eqtype without any values.  This makes it possible
1821      to work on voidptr values using functions meant to operate on light
1822      incomplete pointers.
1823    
1824    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
1825      been vastly simplified.
1826    
1827    ----------------------------------------------------------------------
1828    Name: Matthias Blume
1829    Date: 2002/02/19 10:48:00 EST
1830    Tag: blume-20020219-pqfix
1831    Description:
1832    
1833    Applied Chris Okasaki's bug fix for priority queues.
1834    
1835    ----------------------------------------------------------------------
1836    Name: Matthias Blume
1837    Date: 2002/02/15 17:05:00
1838    Tag: Release_110_39
1839    Description:
1840    
1841    Last-minute retagging is becoming a tradition... :-(
1842    
1843    This is the working release 110.39.
1844    
1845    ----------------------------------------------------------------------
1846    Name: Matthias Blume
1847    Date: 2002/02/15 16:00:00 EST
1848    Tag: Release_110_39-orig
1849    Description:
1850    
1851    Working release 110.39.  New bootfiles.
1852    
1853    (Update: There was a small bug in the installer so it wouldn't work
1854    with all shells.  So I retagged. -Matthias)
1855    
1856    ----------------------------------------------------------------------
1857    Name: Matthias Blume
1858    Date: 2002/02/15 14:17:00 EST
1859    Tag: blume-20020215-showbindings
1860    Description:
1861    
1862    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
1863    the latter can be useful for exploring what bindings are available at
1864    the interactive prompt.  (The first function returns only the list
1865    of symbols that are really bound, the second prints those but also the
1866    ones that CM's autoloading mechanism knows about.)
1867    
1868    ----------------------------------------------------------------------
1869    Name: Matthias Blume
1870    Date: 2002/02/15 12:08:00 EST
1871    Tag: blume-20020215-iptrs
1872    Description:
1873    
1874    Two improvements to ml-nlffigen:
1875    
1876      1. Write files only if they do not exist or if their current contents
1877         do not coincide with what's being written.  (That is, avoid messing
1878         with the time stamps unless absolutely necessary.)
1879    
1880      2. Implement a "repository" mechanism for generated files related
1881         to "incomplete pointer types".   See the README file for details.
1882    
1883    ----------------------------------------------------------------------
1884    Name: Matthias Blume
1885    Date: 2002/02/14 11:50:00 EST
1886    Tag: blume-20020214-quote
1887    Description:
1888    
1889    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
1890    because of the new and improved tag generation scheme.  (Thanks to Allen
1891    Leung for pointing it out.)
1892    
1893    ----------------------------------------------------------------------
1894    Name: Lal George
1895    Date: 2002/02/14 09:55:27 EST 2002
1896    Tag: george-20020214-isabelle-bug
1897    Description:
1898    
1899    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
1900    of Isabelle on the x86.
1901    
1902    From Allen:
1903    -----------
1904     I've found the problem:
1905    
1906         in ra-core.sml, I use the counter "blocked" to keep track of the
1907         true number of elements in the freeze queue.  When the counter goes
1908         to zero, I skip examining the queue.  But I've messed up the
1909         bookkeeping in combine():
1910    
1911             else ();
1912             case !ucol of
1913               PSEUDO => (if !cntv > 0 then
1914                     (if !cntu > 0 then blocked := !blocked - 1 else ();
1915                                        ^^^^^^^^^^^^^^^^^^^^^^^
1916                      moveu := mergeMoveList(!movev, !moveu)
1917                     )
1918                  else ();
1919    
1920         combine() is called to coalesce two nodes u and v.
1921         I think I was thinking that if the move counts of u and v are both
1922         greater than zero then after they are coalesced then one node is
1923         removed from the freeze queue.  Apparently I was thinking that
1924         both u and v are of low degree, but that's clearly not necessarily true.
1925    
1926    
1927    02/12/2002:
1928        Here's the patch.  HOL now compiles.
1929    
1930        I don't know how this impact on performance (compile
1931        time or runtime).  This bug caused the RA (especially on the x86)
1932        to go thru the potential spill phase when there are still nodes on the
1933        freeze queue.
1934    
1935    
1936    
1937    
1938    ----------------------------------------------------------------------
1939    Name: Matthias Blume
1940    Date: 2002/02/13 22:40:00 EST
1941    Tag: blume-20020213-fptr-rtti
1942    Description:
1943    
1944    Fixed a bug in ml-nlffigen that was introduced with one of the previous
1945    updates.
1946    
1947    ----------------------------------------------------------------------
1948    Name: Matthias Blume
1949    Date: 2002/02/13 16:41:00 EST
1950    Tag: blume-20020213-cmlpq
1951    Description:
1952    
1953    Added new priority queue export symbols (which have just been added to
1954    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
1955    not compile and the installer would choke.)
1956    
1957    ----------------------------------------------------------------------
1958    Name: Matthias Blume
1959    Date: 2002/02/13 16:15:00 EST
1960    Tag: blume-20020213-various
1961    Description:
1962    
1963    1. More tweaks to ml-nlffigen:
1964    
1965       - better internal datastructures (resulting in slight speedup)
1966       - "-match" option requires exact match
1967       - "localized" gensym counters (untagged structs/unions nested within
1968         other structs/unions or within typedefs get a fresh counter; their
1969         tag will be prefixed by a concatenation of their parents' tags)
1970       - bug fixes (related to calculation of transitive closure of types
1971         to be included in the output)
1972    
1973    2. Minor Basis updates:
1974    
1975       - added implementations for List.collate and Option.app
1976    
1977    ----------------------------------------------------------------------
1978    Name: Matthias Blume
1979    Date: 2002/02/11 15:55:00 EST
1980    Tag: blume-20020211-gensym
1981    Description:
1982    
1983    Added a "-gensym" option to command line of ml-nlffigen.  This can be
1984    used to specify a "stem" -- a string that is inserted in all "gensym'd"
1985    names (ML structure names that correspond to unnamed C structs, unions,
1986    and enums), so that separate runs of ml-nlffigen do not clash.
1987    
1988    ----------------------------------------------------------------------
1989    Name: Matthias Blume
1990    Date: 2002/02/11 12:05:00 EST
1991    Tag: blume-20020211-gensml
1992    Description:
1993    
1994    A quick fix for a problem with GenSML (in the pgraph-util library):
1995    Make generation of toplevel "local" optional.  (Strictly speaking,
1996    signature definitions within "local" are not legal SML.)
1997    
1998    Other than that: updates to INSTALL and cm/TODO.
1999    
2000    ----------------------------------------------------------------------
2001    Name: Matthias Blume
2002    Date: 2002/02/08 15:00:00 EST
2003    Tag: blume-20020208-uniquepid
2004    Description:
2005    
2006    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2007    
2008    1. The installer (config/install.sh) has gotten smarter:
2009    
2010         - Configuration options are a bit easier to specify now
2011           (in config/targets).
2012         - Bug in recognizing .tar.bz2 files fixed.
2013         - Installer automatically resolves dependencies between
2014           configuration options (e.g., if you ask for eXene, you will
2015           also get cml -- regardless whether you asked for it or not).
2016         - Installer can run in "quieter mode" by setting the environment
2017           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2018           "completely silent", though.
2019         - Build HashCons library as part of smlnj-lib.
2020    
2021    2. A new scheme for assigning persistent identifiers to compilation
2022       units (and, by extension, to types etc.) has been put into place.
2023       This fixes a long-standing bug where types and even dynamic values
2024       can get internally confused, thereby compromising type safety
2025       (abstraction) and dynamic correctness.  See
2026    
2027         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2028    
2029       for an example of how things could go wrong until now.
2030    
2031       The downside of the new scheme is that pids are not quite as
2032       persistent as they used to be: CM will generate a fresh pid
2033       for every compilation unit that it thinks it sees for the first
2034       time.  That means that if you compile starting from a clean, fresh
2035       source tree at two different times, you end up with different
2036       binaries.
2037    
2038       Cutoff recompilation, however, has not been compromised because
2039       CM keeps pid information in special caches between runs.
2040    
2041    ----------------------------------------------------------------------
2042    Name: Lal George
2043    Date: 2002/02/07 15:34:13 EST 2002
2044    Tag: <none>
2045    Description:
2046    
2047    Compilers that generate assembly code may produce  global labels
2048    whose value is resolved at link time. The various peephole optimization
2049    modules did not take this in account.
2050    
2051    TODO. The Labels.addrOf function should really return an option
2052    type so that clients are forced to deal with this issue, rather
2053    than an exception being raised.
2054    
2055    ----------------------------------------------------------------------
2056    Name: Lal George
2057    Date: 2002/02/06 13:55:02 EST
2058    Tag: george-20020206-ra-breakup
2059    Description:
2060    
2061    1. A bug fix from Allen.
2062    
2063        A typo causes extra fstp %st(0)'s to be generated at compensation
2064        edges, which might cause stack underflow traps at runtime.  This
2065        occurs in fft where there are extraneous fstps right before the 'into'
2066        trap instruction (in this case they are harmless since none of the
2067        integers overflow.)
2068    
2069    2. Pulled out various utility modules that were embedded in the modules
2070       of the register allocator. I need these modules for other purposes, but
2071       they are not complete enough to put into a library (just yet).
2072    ----------------------------------------------------------------------
2073    Name: Matthias Blume
2074    Date: 2002/01/31 16:05:00 EST
2075    Tag: blume-20020131-sparc-ccalls
2076    Description:
2077    
2078    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2079       of extra stack space by mistake.  Fixed.
2080    
2081    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2082    
2083    ----------------------------------------------------------------------
2084    Name: Allen Leung
2085    Date: 2002/01/30
2086    Tag: leunga-20020130-nowhere-bug-fix
2087    Description:
2088    
2089       MLRISC bug fixes:
2090       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2091       2. MachineInt.fromString was negating its value.
2092    
2093    ----------------------------------------------------------------------
2094    Name: Matthias Blume
2095    Date: 2002/01/29
2096    Tag: blume-20020129-INSTALL
2097    Description:
2098    
2099    - Added somewhat detailed installation instructions (file INSTALL).
2100    - Fixed curl-detection bug in config/install.sh.
2101    - It is now possible to select the URL getter using the URLGETTER
2102      environment variable:
2103    
2104          not set / "unknown"      --> automatic detection (script tries wget,
2105                                       curl, and lynx)
2106          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2107                                       how to properly invoke them)
2108          other                    --> use $URLGETTER directly, it must take
2109                                       precisely two command-line arguments
2110                                       (source URL and destination file name)
2111    
2112    ----------------------------------------------------------------------
2113    Name: Matthias Blume
2114    Date: 2002/01/28
2115    Tag: blume-20020128-sparc-ccalls
2116    Description:
2117    
2118    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2119    - Make use of the allocParam argument in sparc-c-calls.
2120    
2121    ----------------------------------------------------------------------
2122    Name: Matthias Blume
2123    Date: 2002/01/28
2124    Tag: blume-20020128-allocParam
2125    Description:
2126    
2127    John Reppy:  Changes c-calls API to accept client-callback for
2128    allocating extra stack space.
2129    me: Corresponding changes to mlriscGen (using a dummy argument that
2130        does not change the current behavior).
2131    
2132    ----------------------------------------------------------------------
2133    Name: Matthias Blume
2134    Date: 2002/01/28 12:00:00
2135    Tag: Release_110_38
2136    Description:
2137    
2138    This time for real!!!
2139    
2140    ----------------------------------------------------------------------
2141    Name: Matthias Blume
2142    Date: 2002/01/28 10:56:00 EST
2143    Tag: blume-20020128-retraction
2144    Description:
2145    
2146    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
2147       with blume-Release_110_38-retracted.)
2148    
2149    1. Fixed a problem with incorrect rounding modes in real64.sml.
2150       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2151    
2152    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2153       and enums fixed.  The naming of corresponding ML identifiers should
2154       now be consistent again.
2155    
2156    ----------------------------------------------------------------------
2157    Name: Allen Leung
2158    Date: 2002/01/27
2159    Tag: leunga-20020127-nowhere
2160    Description:
2161    
2162       Added a target called nowhere in the configuration scripts.
2163       Enabling this will build the MLRISC 'nowhere' tool (for translating
2164       programs with where-clauses into legal SML code) during installation.
2165    
2166    ----------------------------------------------------------------------
2167    Name: Matthias Blume
2168    Date: 2002/01/25 21:27:00 EST
2169    Tag: blume-Release_110_38-retracted
2170    Description:
2171    
2172    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
2173    
2174    README will be added later.
2175    
2176    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
2177    (Release_110_38) removed.  Reason: Last-minute bug fixes.
2178    
2179    ----------------------------------------------------------------------
2180    Name: Matthias Blume
2181    Date: 2002/01/25
2182    Tag: blume-20020125-ffi
2183    Description:
2184    
2185    A large number of tweaks and improvements to ml-nlffi-lib and
2186    ml-nlffigen:
2187    
2188       - ML represenation types have been streamlined
2189       - getter and setter functions work with concrete values, not abstract
2190         ones where possible
2191       - ml-nlffigen command line more flexible (see README file there)
2192       - some bugs have been fixed (hopefully)
2193    
2194    ----------------------------------------------------------------------
2195    Name: Lal George
2196    Date: 2002/01/24
2197    Tag: george-20020124-risc-ra-interface
2198    Description:
2199    
2200       There is a dramatic simplification in the interface to the
2201       register allocator for RISC architectures as a result of making
2202       parallel copy instructions explicit.
2203    
2204    ----------------------------------------------------------------------
2205    Name: Matthias Blume
2206    Date: 2002/01/22
2207    Tag: blume-20020122-x86-ccalls
2208    Description:
2209    
2210    Bug fix for c-calls on x86 (having to do with how char- and
2211    short-arguments are being handled).
2212    
2213    ----------------------------------------------------------------------
2214    Name: Matthias Blume
2215    Date: 2002/01/21
2216    Tag: blume-20020121-ff
2217    Description:
2218    
2219    Another day of fiddling with the FFI...
2220    
2221    1. Bug fix/workaround:  CKIT does not complain about negative array
2222       dimensions, so ml-nlffigen has to guard itself against this possibility.
2223       (Otherwise a negative dimension would send it into an infinite loop.)
2224    
2225    2. Some of the abstract types (light objects, light pointers, most "base"
2226       types) in structure C are now eqtypes.
2227    
2228    3. Added constructors and test functions for NULL function pointers.
2229    
2230    ----------------------------------------------------------------------
2231    Name: Matthias Blume
2232    Date: 2002/01/18
2233    Tag: blume-20020118-ready-for-new-release
2234    Description:
2235    
2236    Made config/srcarchiveurl point to a new place.  (Will provide boot
2237    files shortly.)
2238    
2239    Maybe we christen this to be 110.38?
2240    
2241    ----------------------------------------------------------------------
2242    Name: Matthias Blume
2243    Date: 2002/01/18
2244    Tag: blume-20020118-more-ffifiddle
2245    Description:
2246    
2247    Today's FFI fiddling:
2248    
2249      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
2250        C.Set.* that use concrete (MLRep.*) arguments and results instead
2251        of abstract ones.
2252    
2253      - Provided word-style bit operations etc. for "int" representation
2254        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
2255        and Long.
2256    
2257    ----------------------------------------------------------------------
2258    Name: Matthias Blume
2259    Date: 2002/01/18
2260    Tag: blume-20020118-use-x86-fp
2261    Description:
2262    
2263    Now that x86-fast-fp seems to be working, I turned it back on again
2264    by default.  (Seems to work fine now, even with the FFI.)
2265    
2266    Other than that, I added some documentation about the FFI to
2267    src/ml-nlffigen/README and updated the FFI test examples in
2268    src/ml-nlffi-lib/Tests/*.
2269    
2270    ----------------------------------------------------------------------
2271    Name: Allen Leung
2272    Date: 2002/01/17
2273    Tag: leunga-20020117-x86-fast-fp-call
2274    Description:
2275    
2276       1. Fixed a problem with handling return fp values when x86's fast fp
2277          mode is turned on.
2278    
2279       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
2280          of %f32.
2281    
2282       3. Added a constructor INT32lit to the ast of MLRISC tools.
2283    
2284    ----------------------------------------------------------------------
2285    Name: Matthias Blume
2286    Date: 2002/01/16
2287    Tag: blume-20020116-ffifiddle
2288    Description:
2289    
2290    More fiddling with the FFI interface:
2291    
2292     - Make constness 'c instead of rw wherever possible.  This eliminates
2293       the need for certain explicit coercions.  (However, due to ML's
2294       value polymorphism, there will still be many cases where explicit
2295       coercions are necessary.  Phantom types are not the whole answer
2296       to modeling a subtyping relationship in ML.)
2297    
2298     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
2299    
2300     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
2301       clumsy workaround for varargs problem.)
2302    
2303    ----------------------------------------------------------------------
2304    Name: Lal George
2305    Date: 2002/01/15
2306    Tag: <none>
2307    Description:
2308    
2309    1. Since COPY instructions are no longer native to the architecture,
2310       a generic functor can be used to implement the expandCopies function.
2311    
2312    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
2313       TEXT segment.
2314    
2315    ----------------------------------------------------------------------
2316    Name: Matthias Blume
2317    Date: 2002/01/15
2318    Tag: blume-20020115-ffiupdates
2319    Description:
2320    
2321    1. Fix for bug resulting in single-precision float values being returned
2322       incorrectly from FFI calls.
2323    
2324    2. Small modifications to C FFI API:
2325    
2326        - memory-allocation routines return straight objects (no options)
2327          and raise an exception in out-of-memory situations
2328        - unsafe extensions to cast between function pointers and pointers
2329          from/to ints
2330        - added structure C_Debug as an alternative to structure C where
2331          pointer-dereferencing (|*| and |*!) always check for null-pointers
2332        - added open_lib' to DynLinkage;  open_lib' works like open_lib
2333          but also takes a (possibly empty) list of existing library handles
2334          that the current library depends on
2335    
2336    ----------------------------------------------------------------------
2337    Name: Matthias Blume
2338    Date: 2002/01/10
2339    Tag: blume-20020110-newffigen
2340    Description:
2341    
2342    1. Updates to portable graph code.
2343    
2344    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
2345       more scalable now so that even huge interfaces such as the one
2346       for GTK compile in finite time and space. :-)
2347       See src/ml-nlffigen/README for details on what's new.
2348    
2349    ----------------------------------------------------------------------
2350    Name: Lal George
2351    Date: 2001/01/09 14:31:35 EST 2002
2352    Tag: george-20011206-rm-native-copy
2353    Description:
2354    
2355            Removed the native COPY and FCOPY instructions
2356            from all the architectures and replaced it with the
2357            explicit COPY instruction from the previous commit.
2358    
2359            It is now possible to simplify many of the optimizations
2360            modules that manipulate copies. This has not been
2361            done in this change.
2362    
2363    ----------------------------------------------------------------------
2364    Name: Lal George
2365    Date: 2001/12/06 16:50:13 EST 2001
2366    Tag: george-20011206-mlrisc-instruction
2367    Description:
2368    
2369    Changed the representation of instructions from being fully abstract
2370    to being partially concrete. That is to say:
2371    
2372      from
2373            type instruction
2374    
2375      to
2376            type instr                              (* machine instruction *)
2377    
2378            datatype instruction =
2379                LIVE of {regs: C.cellset, spilled: C.cellset}
2380              | KILL of {regs: C.cellset, spilled: C.cellset}
2381              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
2382              | ANNOTATION of {i: instruction, a: Annotations.annotation}
2383              | INSTR of instr
2384    
2385    This makes the handling of certain special instructions that appear on
2386    all architectures easier and uniform.
2387    
2388    LIVE and KILL say that a list of registers are live or killed at the
2389    program point where they appear. No spill code is generated when an
2390    element of the 'regs' field is spilled, but the register is moved to
2391    the 'spilled' (which is present, more for debugging than anything else).
2392    
2393    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
2394    We used to generate:
2395    
2396            DEFFREG f1
2397            f1 := f2 + f3
2398            trapb
2399    
2400    but now generate:
2401    
2402            f1 := f2 + f3
2403            trapb
2404            LIVE {regs=[f1,f2,f3], spilled=[]}
2405    
2406    Furthermore, the DEFFREG (hack) required that all floating point instruction
2407    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
2408    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
2409    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
2410    this architecture).
2411    
2412    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
2413    all the architectures. This will result in further simplification of the
2414    register allocator that must be aware of them for coalescing purposes, and
2415    will also simplify certain aspects of the machine description that provides
2416    callbacks related to parallel copies.
2417    
2418    ANNOTATION should be obvious, and now INSTR represents the honest to God
2419    machine instruction set!
2420    
2421    The <arch>/instructions/<arch>Instr.sml files define certain utility
2422    functions for making porting easier -- essentially converting upper case
2423    to lower case. All machine instructions (of type instr) are in upper case,
2424    and the lower case form generates an MLRISC instruction. For example on
2425    the alpha we have:
2426    
2427      datatype instr =
2428         LDA of {r:cell, b:cell, d:operand}
2429       | ...
2430    
2431      val lda : {r:cell, b:cell, d:operand} -> instruction
2432        ...
2433    
2434    where lda is just (INSTR o LDA), etc.
2435    
2436    ----------------------------------------------------------------------
2437    Name: Matthias Blume
2438    Date: 2001/11/22 21:40:00 EST
2439    Tag: Release_110_37
2440    Description:
2441    
2442    Release 110.37.  This time for real.
2443    
2444    ----------------------------------------------------------------------
2445    Name: Matthias Blume
2446    Date: 2001/11/21 16:35:00 EST
2447    Tag: blume-20011121-foot-in-mouth
2448    Description:
2449    
2450    Removed the "Release_110_37" tag because of a serious bug.
2451    This will be re-tagged once the bug is fixed.
2452    
2453    ----------------------------------------------------------------------
2454    Name: Matthias Blume
2455    Date: 2001/11/21 16:14:00 EST
2456    Tag: blume-20011121-forgottenfile
2457    Description:
2458    
2459    Forgot to add a file.  (Just a .tex-file -- part of
2460    the CM manual source.)
2461    
2462    ----------------------------------------------------------------------
2463    Name: Matthias Blume
2464    Date: 2001/11/21 16:10:00 EST
2465    Tag: blume-20011121-invalid_110_37
2466    Description:
2467    
2468    Note: I removed the original tag "Release_110_37" from this commit
2469          because we found a serious bug in all non-x86 backends.
2470          - Matthias
2471    
2472    1. Modifications to the SML/NJ code generator and to the runtime system
2473       so that code object name strings are directly inserted into code
2474       objects at code generation time.  The only business the runtime system
2475       has with this is now to read the name strings on occasions.
2476       (The encoding of the name string has also changed somewhat.)
2477    
2478    2. CM now implements a simple "set calculus" for specifying export lists.
2479       In particular, it is now possible to refer to the export lists of
2480       other libraries/groups/sources and form unions as well as differences.
2481       See the latest CM manual for details.
2482    
2483    3. An separate notion of "proxy" libraries has again be eliminated from
2484       CM's model.  (Proxy libraries are now simply a special case of using
2485       the export list calculus.)
2486    
2487    4. Some of the existing libraries now take advantage of the new set
2488       calculus.
2489       (Notice that not all libraries have been converted because some
2490       of the existing .cm-files are supposed to be backward compatible
2491       with 110.0.x.)
2492    
2493    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
2494       "General.exnMessage"!  The former relies on a certain hook to be
2495       initialized, and that often does not happen in the stand-alone case.)
2496    
2497    ----------------------------------------------------------------------
2498    Name: Lal George
2499    Date: 2001/11/21  13:56:18 EST
2500    Tag: george-2001121-pseudo-ops
2501    Description:
2502    
2503      Implemented a complete redesign of MLRISC pseudo-ops. Now there
2504      ought to never be any question of incompatabilities with
2505      pseudo-op syntax expected by host assemblers.
2506    
2507      For now, only modules supporting GAS syntax are implemented
2508      but more should follow, such as MASM, and vendor assembler
2509      syntax, e.g. IBM as, Sun as, etc.
2510    
2511    ----------------------------------------------------------------------
2512    Name: Matthias Blume
2513    Date: 2001/11/14 11:52:00 EST
2514    Tag: blume-20011114-srcname
2515    Description:
2516    
2517    1. Routed the name of the current source file to mlriscgen where it
2518       should be directly emitted into the code object.  (This last part
2519       is yet to be done.)
2520    
2521    2. Some cleanup of the pgraph code to make it match the proposal that
2522       I put out the other day.  (The proposal notwithstanding, things are
2523       still in flux here.)
2524    
2525    ----------------------------------------------------------------------
2526    Name: Lal George
2527    Date: 2001/11/14 09:44:04 EST
2528    Tag:
2529    Description:
2530    
2531      Fix for a backpatching bug reported by Allen.
2532    
2533      Because the boundary between short and long span-dependent
2534      instructions is +/- 128, there are an astounding number of
2535      span-dependent instructions whose size is over estimated.
2536    
2537      Allen came up with the idea of letting the size of span
2538      dependent instructions be non-monotonic, for a maxIter
2539      number of times, after which the size must be monotonically
2540      increasing.
2541    
2542      This table shows the number of span-dependent instructions
2543      whose size was over-estimated as a function of maxIter, for the
2544      file Parse/parse/ml.grm.sml:
2545    
2546         maxIter            # of instructions:
2547            10                      687
2548            20                      438
2549            30                      198
2550            40                        0
2551    
2552      In compiling the compiler, there is no significant difference in
2553      compilation speed between maxIter=10 and maxIter=40. Actually,
2554      my measurements showed that maxIter=40 was a tad faster than
2555      maxIter=10! Also 96% of the  files in the compiler reach a fix
2556      point within 13 iterations, so fixing maxIter at 40, while high,
2557      is okay.
2558    
2559    ----------------------------------------------------------------------
2560    Name: Matthias Blume
2561    Date: 2001/10/31 15:25:00 EST
2562    Tag: blume-20011031-pgraph
2563    Description:
2564    
2565    CKIT:
2566    * Changed the "Function" constructor of type Ast.ctype to carry optional
2567      argument identifiers.
2568    * Changed the return type of TypeUtil.getFunction accordingly.
2569    * Type equality ignores the argument names.
2570    * TypeUtil.composite tries to preserve argument names but gives up quickly
2571      if there is a mismatch.
2572    
2573    installation script:
2574    * attempts to use "curl" if available (unless "wget" is available as well)
2575    
2576    CM:
2577    * has an experimental implementation of "portable graphs" which I will
2578      soon propose as an implementation-independent library format
2579    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
2580    
2581    NLFFI-LIB:
2582    * some cleanup (all cosmetic)
2583    
2584    NLFFIGEN:
2585    * temporarily disabled the mechanism that suppresses ML output for
2586      C definitions whose identifiers start with an underscore character
2587    * generate val bindings for enum constants
2588    * user can request that only one style (light or heavy) is being used;
2589      default is to use both (command-line arguments: -heavy and -light)
2590    * fixed bug in handling of function types involving incomplete pointers
2591    * generate ML entry points that take record arguments (i.e., using
2592      named arguments) for C functions that have a prototype with named
2593      arguments
2594      (see changes to CKIT)
2595    
2596    ----------------------------------------------------------------------
2597    Name: Allen Leung
2598    Date: 2001/10/27 20:34:00 EDT
2599    Tag: leunga-20011027-x86-fast-fp-call
2600    Description:
2601    
2602       Fixed the bug described in blume-20010920-slowfp.
2603    
2604       The fix involves
2605          1. generating FCOPYs in FSTP in ia32-svid
2606          2. marking a CALL with the appropriate annotation
2607    
2608    ----------------------------------------------------------------------
2609    Name: Matthias Blume
2610    Date: 2001/10/16 11:32:00 EDT
2611    Tag: blume-20011016-netbsd
2612    Description:
2613    
2614    Underscore patch from Chris Richards (fixing problem with compiling
2615    runtime system under recent NetBSD).
2616    
2617    ----------------------------------------------------------------------
2618    Name: Allen Leung
2619    Date: 2001/10/12 17:18:32 EDT 2001
2620    Tag: leung-20011012-x86-printflowgraph
2621    Description:
2622    
2623    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
2624    
2625    ----------------------------------------------------------------------
2626    Name: Lal George
2627    Date: 2001/10/11 23:51:34 EDT
2628    Tag: george-20011011-too-many-instrs
2629    Description:
2630    
2631    The representation of a program point never expected to see more
2632    than 65536 instructions in a basic block!
2633    
2634    ----------------------------------------------------------------------
2635    Name: Lal George
2636    Date: 2001/10/09 09:41:37 EDT
2637    Tag: george-20011008-mlrisc-labels
2638    Description:
2639    
2640    Changed the machine description files to support printing of
2641    local and global labels in assembly code, based on host assembler
2642    conventions.
2643    
2644    ----------------------------------------------------------------------
2645    Name: Matthias Blume
2646    Date: 2001/09/25 15:25:00 EDT
2647    Tag: blume-20010925-exninfo
2648    Description:
2649    
2650    I provided a non-hook implementation of exnName (at the toplevel) and
2651    made the "dummy" implementation of exnMessage (at the toplevel) more
2652    useful: if nothing gets "hooked in", then at least you are going to
2653    see the exception name and a message indicating why you don't see more.
2654    
2655    [For the time being, programs that need exnMessage and want to use
2656    ml-build should either use General.exnMessage (strongly recommended) or
2657    refer to structure General at some other point so that CM sees a
2658    static dependency.]
2659    
2660    [Similar remarks go for "print" and "use":  If you want to use their
2661    functionality in stand-alone programs generated by ml-build, then use
2662    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
2663    
2664    ----------------------------------------------------------------------
2665    Name: Matthias Blume
2666    Date: 2001/09/20 17:28:00 EDT
2667    Tag: blume-20010920-slowfp
2668    Description:
2669    
2670    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
2671    
2672    ----------------------------------------------------------------------
2673    Name: Matthias Blume
2674    Date: 2001/09/20 17:20:00 EDT
2675    Tag: blume-20010920-canonicalpaths
2676    Description:
2677    
2678    0. Updated the BOOT file (something that I forgot to do earlier).
2679    
2680    1. Small internal change to CM so that it avoids "/../" in filenames
2681       as much as possible (but only where it is safe).
2682    
2683    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
2684       that arguments that contain delimiters are passed through correctly.
2685       This change also means that all "special" arguments of the form
2686       @SMLxxx... must come first.
2687    
2688    3. Changed install script to put relative anchor names for tool commands
2689       into pathconfig.
2690    
2691    ----------------------------------------------------------------------
2692    Name: Matthias Blume
2693    Date: 2001/09/18 15:35:00 EDT
2694    Tag: blume-20010918-readme11036
2695    Description:
2696    
2697    Added README files.
2698    
2699    ----------------------------------------------------------------------
2700    Name: Matthias Blume
2701    Date: 2001/09/18 11:45:00 EDT
2702    Tag: Release_110_36 (retag)
2703    Description:
2704    
2705    Fixed mistake in config/preloads. Retagged as 110.36.
2706    
2707    ----------------------------------------------------------------------
2708    Name: Matthias Blume
2709    Date: 2001/09/18 09:40:00 EDT
2710    Tag: Release_110_36_orig (tag changed)
2711    Description:
2712    
2713    New version (110.36).  New bootfiles.
2714    
2715    ----------------------------------------------------------------------
2716    Name: Matthias Blume
2717    Date: 2001/09/14 16:15:00 EDT
2718    Tag: blume-20010914-x86fastfp
2719    Description:
2720    
2721    John committed some changes that Allen made, in particular a (hopefully)
2722    correctly working version of the x86-fp module.
2723    
2724    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
2725    flag to "true".  Everything seems to compile to a fixpoint ok, and
2726    "mandelbrot" speeds up by about 15%.
2727    
2728    ----------------------------------------------------------------------
2729    Name: Matthias Blume
2730    Date: 2001/09/13 11:20:00 EDT
2731    Tag: blume-20010913-minimal
2732    Description:
2733    
2734    1. Stefan Monnier's patch to fix a miscompilation problem that
2735       was brought to light by John Reppy's work on Moby.
2736    
2737    2. Implemented a minimal "structure Compiler" that contains just
2738       "version" and "architecture".  The minimal version will be
2739       available when the full version is not.  This is for backward-
2740       compatibility with code that wants to test Compiler.version.
2741    
2742    ----------------------------------------------------------------------
2743    Name: Matthias Blume
2744    Date: 2001/08/28 14:03:00 EDT
2745    Tag: blume-20010828-ml-lex
2746    Description:
2747    
2748    Fix for bug 1581, received from Neophytos Michael.
2749    
2750    ----------------------------------------------------------------------
2751    Name: Matthias Blume
2752    Date: 2001/08/27 11:20:00 EDT
2753    Tag: blume-20010827-readme11035
2754    Description:
2755    
2756    Fleshed out the README file for 110.35.
2757    
2758    ----------------------------------------------------------------------
2759    Name: Matthias Blume
2760    Date: 2001/08/24 17:10:00 EDT
2761    Tag: Release_110_35
2762    Description:
2763    
2764    New version number (110.35).  New bootfiles.
2765    
2766    ----------------------------------------------------------------------
2767    Name: Lal George
2768    Date: 2001/08/24 13:47:18 EDT 2001
2769    Tag: george-20010824-MLRISC-graphs
2770    Description:
2771    
2772     removed clusters from MLRISC completely and replaced with graphs.
2773    
2774    ----------------------------------------------------------------------
2775    Name: Matthias Blume
2776    Date: 2001/08/23 17:50:00 EDT
2777    Tag: blume-20010823-toplevel
2778    Description:
2779    
2780    - some reorganization of the code that implements various kinds of
2781      environments in the compiler (static, dynamic, symbolic, combined)
2782    - re-implemented the EnvRef module so that evalStream works properly
2783      (if the stream contains references to "use", "CM.make", etc.)
2784    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
2785    
2786    ----------------------------------------------------------------------
2787    Name: Matthias Blume
2788    Date: 2001/08/20 15:50 EDT
2789    Tag: blume20010820-slipup
2790    Description:
2791    
2792    I forgot to commit a few files.  Here they are...
2793    
2794    ----------------------------------------------------------------------
2795    Name: Matthias Blume
2796    Date: 2001/08/20 15:35:00 EDT
2797    Tag: blume-20010820-debugprof
2798    Description:
2799    
2800    !!!! NEW BOOTFILES !!!!
2801    
2802    This is another round of reorganizing the compiler sources.  This
2803    time the main goal was to factor out all the "instrumentation"
2804    passes (for profiling and backtracing) into their own library.
2805    The difficulty was to do it in such a way that it does not depend
2806    on elaborate.cm but only on elabdata.cm.
2807    
2808    Therefore there have been further changes to both elaborate.cm and
2809    elabdata.cm -- more "generic" things have been moved from the former
2810    to the latter.  As a result, I was forced to split the assignment
2811    of numbers indicating "primtyc"s into two portions: SML-generic and
2812    SML/NJ-specific.  Since it would have been awkward to maintain,
2813    I bit the bullet and actually _changed_ the mapping between these
2814    numbers and primtycs.  The bottom line of this is that you need
2815    a new set of bin- and bootfiles.
2816    
2817    I have built new bootfiles for all architectures, so doing a fresh
2818    checkout and config/install.sh should be all you need.
2819    
2820    The newly created library's name is
2821    
2822        $smlnj/viscomp/debugprof.cm
2823    
2824    and its sources live under
2825    
2826        src/compiler/DebugProf
2827    
2828    ----------------------------------------------------------------------
2829    Name: Matthias Blume
2830    Date: 2001/08/15 17:15:00 EDT
2831    Tag: blume-20010815-compreorg
2832    Description:
2833    
2834    This is a first cut at reorganizing the CM libraries that make up the
2835    core of the compiler.  The idea is to separate out pieces that could
2836    be used independently by tools, e.g., the parser, the typechecker, etc.
2837    
2838    The current status is a step in this direction, but it is not quite
2839    satisfactory yet.  Expect more changes in the future.
2840    
2841    Here is the current (new) organization...
2842    
2843        What used to be $smlnj/viscomp/core.cm is now divided into
2844        six CM libraries:
2845    
2846             $smlnj/viscomp/basics.cm
2847                           /parser.cm
2848                           /elabdata.cm
2849                           /elaborate.cm
2850                           /execute.cm
2851                           /core.cm
2852    
2853        The CM files for these libraries live under src/system/smlnj/viscomp.
2854        All these libraries are proxy libraries that contain precisely
2855        one CM library component.  Here are the locations of the components
2856        (all within the src/compiler tree):
2857    
2858             Basics/basics.cm
2859             Parse/parser.cm
2860             ElabData/elabdata.cm
2861             Elaborator/elaborate.cm
2862             Execution/execute.cm
2863             core.cm
2864    
2865         [This organization is the same that has been used already
2866         for a while for the architecture-specific parts of the visible
2867         compiler and for the old version of core.cm.]
2868    
2869         As you will notice, many source files have been moved from their
2870         respective original locations to a new home in one of the above
2871         subtrees.
2872    
2873         The division of labor between the new libraries is the following:
2874    
2875             basics.cm:
2876                - Simple, basic definitions that pertain to many (or all) of
2877                  the other libraries.
2878             parser.cm:
2879                - The SML parser, producing output of type Ast.dec.
2880                - The type family for Ast is also defined and exported here.
2881             elabdata.cm:
2882                - The datatypes that describe input and output of the elaborator.
2883                  This includes types, absyn, and static environments.
2884             elaborator.cm:
2885                - The SML/NJ type checker and elaborator.
2886                  This maps an Ast.dec (with a given static environment) to
2887                  an Absyn.dec (with a new static environment).
2888                - This libraries implements certain modules that used to be
2889                  structures as functors (to remove dependencies on FLINT).
2890             execute.cm:
2891                - Everything having to do with executing binary code objects.
2892                - Dynamic environments.
2893             core.cm:
2894                - SML/NJ-specific instantiations of the elaborator and MLRISC.
2895                - Top-level modules.
2896                - FLINT (this should eventually become its own library)
2897    
2898    Notes:
2899    
2900    I am not 100% happy with the way I separated the elaborator (and its
2901    data structures) from FLINT.  Two instances of the same problem:
2902    
2903        1. Data structures contain certain fields that carry FLINT-specific
2904           information.  I hacked around this using exn and the property list
2905           module from smlnj-lib.  But the fact that there are middle-end
2906           specific fields around at all is a bit annoying.
2907    
2908        2. The elaborator calculates certain FLINT-related information.  I tried
2909           to make this as abstract as I could using functorization, but, again,
2910           the fact that the elaborator has to perform calculations on behalf
2911           of the middle-end at all is not nice.
2912    
2913        3. Having to used exn and property lists is unfortunate because it
2914           weakens type checking.  The other alternative (parameterizing
2915           nearly *everything*) is not appealing, though.
2916    
2917    I removed the "rebinding =" warning hack because due to the new organization
2918    it was awkward to maintain it.  As a result, the compiler now issues some of
2919    these warnings when compiling init.cmi during bootstrap compilation. On
2920    the plus side, you also get a warning when you do, for example:
2921       val op = = Int32.+
2922    which was not the case up to now.
2923    
2924    I placed "assign" and "deref" into the _Core structure so that the
2925    code that deals with the "lazy" keyword can find them there.  This
2926    removes the need for having access to the primitive environment
2927    during elaboration.
2928    
2929    ----------------------------------------------------------------------
2930    Name: Matthias Blume
2931    Date: 2001/08/13
2932    Tag: blume-20010813-closures
2933    Description:
2934    
2935    This fix was sent to us by Zhong Shao.  It is supposed to improve the
2936    performance of certain loops by avoiding needless closure allocation.
2937    
2938    ----------------------------------------------------------------------
2939    Name: Lal George
2940    Date: 2001/07/31 10:03:23 EDT 2001
2941    Tag: george-20010731-x86-fmalloc
2942    Description: Fixed bug in x86 calls
2943    
2944        There was a bug where call instructions would mysteriously
2945        vanish. The call instruction had to be one that returned
2946        a floating point value.
2947    
2948    ----------------------------------------------------------------------
2949    Name: Lal George
2950    Date: 2001/07/19 16:36:29 EDT 2001
2951    Tag: george-20010719-simple-cells
2952    Description:
2953    
2954    I have dramatically simplified the interface for CELLS in MLRISC.
2955    
2956    In summary, the cells interface is broken up into three parts:
2957    
2958      1. CellsBasis : CELLS_BASIS
2959    
2960            CellsBasis is a top level structure and common for all
2961            architectures.  it contains the definitions of basic datatypes
2962            and utility  functions over these types.
2963    
2964      2. functor Cells() : CELLS
2965    
2966            Cells generates an interface for CELLS that incorporates the
2967            specific resources on the target architecture, such as the
2968            presence of special register classes, their number and size,
2969            and various useful substructures.
2970    
2971      3. <ARCH>CELLS
2972    
2973            e.g. SparcCells: SPARCCELLS
2974    
2975            <ARCH>CELLS usually contains additional bindings for special
2976            registers  on the architecture, such as:
2977    
2978                    val r0 : cell           (* register zero *)
2979                    val y : cell            (* Y register *)
2980                    val psr : cell          (* processor status register *)
2981                    ...
2982    
2983            The structure returned by applying the Cells functor is opened
2984            in this interface.
2985    
2986    The main implication of all this is that the datatypes for cells is
2987    split between CellsBasis and CELLS -- a fairly simple change for user
2988    code.
2989    
2990    In the old scheme the CELLS interface had a definitional binding of
2991    the form:
2992    
2993            signature CELLS = sig
2994    
2995               structure CellsBasis = CellsBasis
2996    
2997               ...
2998    
2999            end
3000    
3001    With all the sharing constraints that goes on in MLRISC, this old
3002    design  quickly leads to errors such as:
3003    
3004            "structure definition spec inside of sharing ... "
3005    
3006    
3007    and appears to require an unacceptable amount of sharing and where
3008    constraint hackery.
3009    
3010    I think this error message (the interaction of definitional specs and
3011    sharing) requires more explanation on our web page.
3012    
3013    ----------------------------------------------------------------------
3014    Name: Matthias Blume
3015    Date: 2001/07/19 15:00:00 EDT
3016    Tag: blume-20010719-libreorg
3017    Description:
3018    
3019    This update puts together a fairly extensive but straightforward change
3020    to the way the libraries that implement the interactive system are
3021    organized:
3022    
3023       The biggest change is the elimination of structure Compiler.  As a
3024       replacement for this structure, there is now a CM library
3025       (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
3026       that exports all the substructures of the original structure Compiler
3027       directly.  So instead of saying Compiler.Foo.bar one now simply
3028       says Foo.bar.  (The CM libraries actually export a collection of
3029       structures that is richer than the collection of substructures of
3030       structure Compiler.)
3031    
3032       To make the transition smooth, there is a separate library called
3033       $smlnj/compiler/compiler.cm which puts together and exports the
3034       original structure Compiler (or at least something very close to it).
3035    
3036       There are five members of the original structure Compiler
3037       that are not exported directly but which instead became members
3038       of a new structure Backend (described by signature BACKEND).  These are:
3039       structure Profile (: PROFILE), structure Compile (: COMPILE), structure
3040       Interact (: INTERACT), structure Machine (: MACHINE), and val
3041       architecture (: string).
3042    
3043       Structure Compiler.Version has become structure CompilerVersion.
3044    
3045       Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
3046       by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
3047       or x86, respectively.
3048       Each of these exports the same frontend structures that
3049       $smlnj/compiler.cm exports.  But they do not have a structure Backend
3050       and instead export some structure <Arch>Backend where <Arch> is Alpha32,
3051       Hppa, PPC, Sparc, or X86, respectively.
3052    
3053       Library $smlnj/compiler/all.cm exports the union of the exports of
3054       $smlnj/compiler/<arch>.cm
3055    
3056       There are no structures <Arch>Compiler anymore, use
3057       $smlnj/compiler/<arch>.cm instead.
3058    
3059       Library host-compiler-0.cm is gone.  Instead, the internal library
3060       that instantiates CM is now called cm0.cm.  Selection of the host
3061       compiler (backend) is no longer done here but. (Responsibility for it
3062       now lies with $smlnj/compiler/current.cm.  This seems to be more
3063       logical.)
3064    
3065       Many individual files have been moved or renamed.  Some files have
3066       been split into multiple files, and some "dead" files have been deleted.
3067    
3068    Aside from these changes to library organization, there are also changes
3069    to the way the code itself is organized:
3070    
3071       Structure Binfile has been re-implemented in such a way that it no
3072       longer needs any knowledge of the compiler.  It exclusively deals
3073       with the details of binfile layout.  It no longer invokes the
3074       compiler (for the purpose of creating new prospective binfile
3075       content), and it no longer has any knowledge of how to interpret
3076       pickles.
3077    
3078       Structure Compile (: COMPILE) has been stripped down to the bare
3079       essentials of compilation.  It no longer deals with linking/execution.
3080       The interface has been cleaned up considerably.
3081    
3082       Utility routines for dealing with linking and execution have been
3083       moved into their own substructures.
3084    
3085       (The ultimate goal of these changes is to provide a light-weight
3086       binfile loader/linker (at least for, e.g., stable libraries) that
3087       does not require CM or the compiler to be present.)
3088    
3089    CM documentation has been updated to reflect the changes to library
3090    organization.
3091    
3092    ----------------------------------------------------------------------
3093    Name: Matthias Blume
3094    Date: 2001/07/10 17:30:00 EDT
3095    Tag: Release_110_34
3096    Description:
3097    
3098    Minor tweak to 110.34 (re-tagged):
3099    
3100      - README.html file added to CVS repository
3101      - runtime compiles properly under FreeBSD 3.X and 4.X
3102    
3103    ----------------------------------------------------------------------
3104    Name: Matthias Blume
3105    Date: 2001/07/10 17:30:00 EDT
3106    Tag: Release_110_34
3107    Description:
3108    
3109    New version number (110.34). New bootfiles.
3110    
3111    ----------------------------------------------------------------------
3112    Name: Matthias Blume
3113    Date: 2001/07/09 16:00:00 EDT
3114    Tag: blume-20010709-more-varargs
3115    Description:
3116    
3117    I changed the handling of varargs in ml-nlffigen again:
3118    The ellipsis ... will now simply be ignored (with an accompanying warning).
3119    
3120    The immediate effect is that you can actually call a varargs function
3121    from ML -- but you can't actually supply any arguments beyond the ones
3122    specified explicitly.  (For example, you can call printf with its format
3123    string, but you cannot pass additional arguments.)
3124    
3125    This behavior is only marginally more useful than the one before, but
3126    it has the advantage that a function or, more importantly, a function
3127    type never gets dropped on the floor, thus avoiding follow-up problems with
3128    other types that refer to the offending one.
3129    
3130    ----------------------------------------------------------------------
3131    Name: Matthias Blume
3132    Date: 2001/07/09 11:25:00 EDT
3133    Tag: blume-20010709-varargs
3134    Description:
3135    
3136    1. ckit-lib.cm now exports structure Error
3137    2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
3138       with a warning accompanied by a source location.  Moreover, it
3139       merely skips the offending function or type and proceeds with the
3140       rest of its work.u  As a result, one can safely feed C code containing
3141       "..." to ml-nlffigen.
3142    3. There are some internal improvements to CM, providing slightly
3143       more general string substitutions in the tools subsystem.
3144    
3145    ----------------------------------------------------------------------
3146    Name: Matthias Blume
3147    Date: 2001/06/27 15:10:00 EDT
3148    Tag: blume-20010627-concur
3149    Description:
3150    
3151    Fixed a small bug in CM's handling of parallel compilation.
3152    (You could observe the bug by Control-C-interrupting an ordinary
3153    CMB.make or CM.stabilize and then attaching some compile servers.
3154    The result was that all of a sudden the previously interrupted
3155    compilation would continue on its own.  This was because of
3156    an over-optimization: CM did not bother to clean out certain queues
3157    when no servers were attached "anyway", resulting in the contents
3158    of these queues to grab control when new servers did get attached.)
3159    
3160    There is also another minor update to the CM manual.
3161    
3162    ----------------------------------------------------------------------
3163    Name: Matthias Blume
3164    Date: 2001/06/26 16:15:00 EDT
3165    Tag: blume-20010626-cmdoc
3166    Description:
3167    
3168    Minor typo fixed in CM manual (syntax diagram for libraries).
3169    
3170    ----------------------------------------------------------------------
3171    Name: Matthias Blume
3172    Date: 2001/06/25 22:55:00 EDT
3173    Tag: blume-20010625-x86pc
3174    Description:
3175    
3176    Fixed a nasty bug in the X86 assembly code that caused signal
3177    handlers to fail (crash) randomly.
3178    
3179    ----------------------------------------------------------------------
3180    Name: Matthias Blume
3181    Date: 2001/06/25 12:05:00 EDT
3182    Tag: blume-20010625-nlffigen
3183    Description:
3184    
3185    This update fixes a number of minor bugs in ml-nlffigen as reported by
3186    Nick Carter <nbc@andrew.cmu.edu>.
3187    
3188      1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
3189      2. Default names for generated files are now derived from the name of
3190         the C file *without its directory*.  In particular, this causes generated
3191         files to be placed locally even if the C file is in some system directory.
3192      3. Default names for generated signatures and structures are also derived
3193         from the C file name without its directory.  This avoids silly things
3194         like "structure GL/GL".
3195         (Other silly names are still possible because ml-nlffigen does not do
3196          a thorough check of whether generated names are legal ML identifiers.
3197          When in doubt, use command line arguments to force particular names.)
3198    
3199    ----------------------------------------------------------------------
3200    Name: Matthias Blume
3201    Date: 2001/06/21 12:25:00 EDT
3202    Tag: blume-20010621-eXene
3203    Description:
3204    
3205    eXene now compiles and (sort of) works again.
3206    
3207    The library name (for version > 110.33) is $/eXene.cm.
3208    
3209    I also added an new example in src/eXene/examples/nbody.  See the
3210    README file there for details.
3211    
3212    ----------------------------------------------------------------------
3213    Name: Matthias Blume
3214    Date: 2001/06/20 16:40:00 EDT
3215    Tag: blume-20010620-cml
3216    Description:
3217    
3218    CML now compiles and works again.
3219    
3220    Libraries (for version > 110.33):
3221    
3222      $cml/cml.cm            Main CML library.
3223      $cml/basis.cm          CML's version of $/basis.cm.
3224      $cml/cml-internal.cm   Internal helper library.
3225      $cml/core-cml.cm       Internal helper library.
3226      $cml-lib/trace-cml.cm  Tracing facility.
3227      $cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
3228    
3229    The installer (config/install.sh) has been taught how to properly
3230    install this stuff.
3231    
3232    ----------------------------------------------------------------------
3233    Name: Matthias Blume
3234    Date: 2001/06/19 17:55:00 EDT
3235    Tag: blume-20010619-instantiate
3236    Description:
3237    
3238    This un-breaks the fix for bug 1432.
3239    (The bug was originally fixed in 110.9 but I broke it again some
3240    time after that.)
3241    
3242    ----------------------------------------------------------------------
3243    Name: Matthias Blume
3244    Date: 2001/06/19 17:25:00 EDT
3245    Tag: blume-20010619-signals
3246    Description:
3247    
3248    This should (hopefully) fix the long-standing signal handling bug.
3249    (The runtime system was constructing a continuation record with an
3250    incorrect descriptor which would cause the GC to drop data on the floor...)
3251    
3252    ----------------------------------------------------------------------
3253    Name: Matthias Blume
3254    Date: 2001/06/15 15:05:00 EDT
3255    Tag: blume-20010615-moresparc
3256    Description:
3257    
3258    Here is a short late-hour update related to Sparc c-calls:
3259    
3260     -- made handling of double-word arguments a bit smarter
3261    
3262     -- instruction selection phase tries to collapse certain clumsily
3263        constructed ML-Trees; typical example:
3264    
3265            ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
3266    
3267        This currently has no further impact on SML/NJ since mlriscGen does
3268        not seem to generate such patterns in the first place, and c-calls
3269        (which did generate them in the beginning) has meanwhile been fixed
3270        so as to avoid them as well.
3271    
3272    ----------------------------------------------------------------------
3273    Name: Matthias Blume
3274    Date: 2001/06/15 15:05:00 EDT
3275    Tag: blume-20010615-sparc
3276    Description:
3277    
3278    The purpose of this update is to provide an implementation of NLFFI
3279    on Sparc machines.
3280    
3281    Here are the changes in detail:
3282    
3283       * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
3284       the Sparc implementation of the c-calls API.
3285       * The Sparc backend of SML/NJ has been modified to uniformely use %fp
3286       for accessing the ML frame.  Thus, we have a real frame pointer and
3287       can freely modify %sp without need for an omit-frame-ptr phase.
3288       The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
3289       this case.
3290       * ml-nlffigen has been taught to produce code for different architectures
3291       and calling conventions.
3292       * In a way similar to what was done in the x86 case, the Sparc
3293       backend uses its own specific extension to mltree.  (For example,
3294       it needs to be able to generate UNIMP instructions which are part
3295       of the calling convention.)
3296       * ml-nlffi-lib was reorganized to make it more modular (in particular,
3297       to make it easier to plug in new machine- and os-dependent parts).
3298    
3299    There are some other fairly unrelated bug fixes and cleanups as well:
3300    
3301       * I further hacked the .cm files for MLRISC tools (like MDLGen) so
3302       that they properly share their libraries with existing SML/NJ libraries.
3303       * I fixed a minor cosmetic bug in CM, supressing certain spurious
3304       follow-up error messages.
3305       * Updates to CM/CMB documentation.
3306    
3307    TODO items:
3308    
3309       * MLRISC should use a different register as its asmTemp on the Sparc.
3310         (The current %o2 is a really bad choice because it is part of the
3311         calling conventions, so things might interfere in unexpected ways.)
3312    
3313    ----------------------------------------------------------------------
3314    Name: Matthias Blume
3315    Date: 2001/06/07
3316    Tag: blume-20010607-calls
3317    Description:
3318    
3319    A number of internal changes related to C calls and calling conventions:
3320    
3321    1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
3322       number of bytes popped implicitly (by the callee).  In most cases
3323       this field is 0 but on x86/win32 it is some non-zero value.  This
3324       is information provided for the benefit of the "omit-frameptr" pass.
3325    2. The CALL instruction on the x86 carries a similar "pops" field.
3326       The instruction selection phase copies its value from the ML-Tree
3327       CALL statement.
3328    3. On all other architectures, the instruction selection phase checks
3329       whether "pops=0" and complains if not.
3330    4. The c-calls implementation for x86 now accepts two calling conventions:
3331       "ccall" and "stdcall".  When "ccall" is selected, the caller cleans
3332       up after the call and pops is set to 0.  For "stdcall", the caller
3333       does nothing, leaving the cleanup to the callee; pops is set to
3334       the number of bytes that were pushed onto the stack.
3335    5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
3336       distinguish between "ccall" and "stdcall".
3337    6. The UNIMP instruction has been added to the supported Sparc instruction
3338       set. (This is needed for implementing the official C calling convention
3339       on this architecture.)
3340    7. I fixed some of the .cm files under src/MLRISC/Tools to make them
3341       work with the latest CM.
3342    
3343    ----------------------------------------------------------------------
3344    Name: Matthias Blume
3345    Date: 2001/06/05 15:10:00 EDT
3346    Tag: blume-20010605-cm-index
3347    Description:
3348    
3349    0. The "lambdasplit" parameter for class "sml" in CM has been documented.
3350    
3351    1. CM can now generate "index files".  These are human-readable files
3352       that list on a per-.cm-file basis each toplevel symbol defined or
3353       imported.  The location of the index file for
3354       <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
3355       To enable index-file generation, set CM.Control.generate_index to true
3356       or export an environment-symbol: export CM_GENERATE_INDEX=true.
3357    
3358       The CM manual has been updated accordingly.
3359    
3360    2. I made some slight modifications to the c-calls API in MLRISC.
3361    
3362         a) There is now a callback to support saving/restoring of
3363            dedicated but caller-save registers around the actual call
3364            instruction.
3365         b) One can optionally specify a comment-annotation for the
3366            call instruction.
3367    
3368    3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
3369       (For example, the comment annotation shows the C prototype of
3370        the function being called.)
3371    
3372    ----------------------------------------------------------------------
3373    Name: Matthias Blume
3374    Date: 2001/06/01 13:30:00 EDT
3375    Tag: blume-20010601-nlffi-cleanup
3376    Description:
3377    
3378    This is mostly a cleanup of MLFFI stuff:
3379    
3380       - some signature files have been put into a more exposed place
3381       - the ugly 'f type parameter is gone (simplifies types tremendously!)
3382       - ml-nlffigen changed accordingly
3383       - tutorial updated
3384    
3385    Other changes:
3386    
3387       - author's affiliation in CM manual(s) updated
3388       - some more recognized keywords added to Allen's sml.sty
3389    
3390    ----------------------------------------------------------------------
3391    Name: Matthias Blume
3392    Date: 2001/05/25 15:30:00 EDT
3393    Tag: blume-20010525-iptr
3394    Description:
3395    
3396      - put the official 110.33-README (as it appears on the ftp server) under
3397        CVS
3398      - fixed a small bug related to incomplete pointer types in
3399        ml-nlffigen
3400      - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
3401        (it does not need the 'f type parameter)
3402    
3403    ----------------------------------------------------------------------
3404    Name: Matthias Blume
3405    Date: 2001/05/23 14:30:00 EDT
3406    Tag: Release_110_33
3407    Description:
3408    
3409    New version number (110.33).  New bootfiles.
3410    
3411    ----------------------------------------------------------------------
3412    Name: Matthias Blume
3413    Date: 2001/05/22 18:06:00 EDT
3414    Tag: blume-20010522-targets
3415    Description:
3416    
3417    Made install.sh use file config/targets.customized if it exists, falling
3418    back to config/targets if it doesn't.  This way one can have a customized
3419    version of the targets file without touching the "real thing", thus
3420    eliminating the constant fear of accidentally checking something bogus
3421    back into the CVS repository...  (File config/targets.customized must
3422    not be added to the repository!)
3423    
3424    ----------------------------------------------------------------------
3425    Name: Matthias Blume
3426    Date: 2001/05/22 16:30:00 EDT
3427    Tag: blume-20010522-minitut
3428    Description:
3429    
3430    1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
3431       struct returns.
3432    2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
3433       incomplete, preliminary documentation for NLFFI.
3434    
3435    ----------------------------------------------------------------------
3436    Name: Matthias Blume
3437    Date: 2001/05/14 11:30:00 EDT
3438    Tag: blume-20010514-script
3439    Description:
3440    
3441    Some bugs in install script fixed.
3442    
3443    In addition to that I also made a slight change to the NLFFI API:
3444    Functors generated by ml-nlffigen now take the dynamic library as a
3445    straight functor argument, not as a suspended one.  (The original
3446    functor code used to force the suspension right away anyway, so there
3447    was nothing gained by this complication of the interface.)
3448    
3449    ----------------------------------------------------------------------
3450    Name: Matthias Blume
3451    Date: 2001/05/11 14:35:00 EDT
3452    Tag: blume-20010511-ml-nlffi
3453    Description:
3454    
3455    I finally took the plunge and added my new FFI code to the main
3456    repository.  For x86-linux it is now ready for prime-time.
3457    
3458    There are two new subdirectories of "src":
3459    
3460      - ml-nlffi-lib:
3461           The utility library for programs using the FFI interface.
3462           Here is the implementation of $/c.cm and its associated low-level
3463           partners $/c-int.cm and $/memory.cm.
3464      - ml-nlffigen:
3465           A stand-alone program for generating ML glue code from C source
3466           code.
3467    
3468    Building ml-nlffigen requires $/ckit-lib.cm.
3469    
3470    The config/install.sh script has been updates to do the Right Thing
3471    (hopefully).
3472    
3473    Notice that the source tree for the C-Kit will not be put under "src"
3474    but directly under the installation root directory.  (This is the
3475    structure that currently exists on the CVS server when you check out
3476    module "sml".)  Fortunately, config/install.sh knows about this oddity.
3477    
3478    Bugs: No documentation yet.
3479    
3480    ----------------------------------------------------------------------
3481    Name: Matthias Blume
3482    Date: 2001/05/09 16:35:00 EDT
3483    Tag: blume-20010509-cpscontract
3484    Description:
3485    
3486    Fixed a bug in the accounting code in cpsopt/contract.sml.  (The
3487    wrapper/unwrapper elimination did not decrement usage counts and some
3488    dead variables got overlooked by the dead-up logic.)
3489    
3490    ----------------------------------------------------------------------
3491    Name: Lal George
3492    Date: 2001/05/08  17:26:09 EDT
3493    Tag: george-20010508-omit-frameptr
3494    Description:
3495    
3496    Changes to implement the omit-frame-pointer optimization to support
3497    raw C calls. For now, there is only support on the Intel x86, but
3498    other architectures will follow as more experience is gained with this.
3499    
3500    
3501    ----------------------------------------------------------------------
3502    Name: Matthias Blume
3503    Date: 2001/05/07 14:40:00 EDT
3504    Tag: blume-20010507-proxies
3505    Description:
3506    
3507    I made into "proxy libraries" all libraries that qualify for such a
3508    change.  (A qualifying library is a library that has another library or
3509    groups as its sole member and repeats that member's export list
3510    verbatim.  A proxy library avoids this repetition by omitting its export
3511    list, effectively inheriting the list that its (only) member exports.
3512    See the CM manual for more explanation.)
3513    The main effect is that explicit export lists for these libraries
3514    do not have to be kepts in sync, making maintenance a bit easier.
3515    
3516    I also added copyright notices to many .cm-files.
3517    
3518    Last but not least, I made a new set of bootfiles.
3519    
3520    ----------------------------------------------------------------------
3521    Name: Matthias Blume
3522    Date: 2001/05/04 17:00:00 EDT
3523    Tag: blume-20010504-cm-lsplit
3524    Description:
3525    
3526    0. John merged pending changes to $/smlnj-lib.cm
3527    
3528    1. Allen's previous change accidentally backed out of one of Lal's
3529       earlier changes.  I undid this mistake (re-introducing Lal's change).
3530    
3531    2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
3532       within the compiler where applicable.  There is some code simplification
3533       because of that.
3534    
3535    3. The "split" phase (in FLINT) is now part of the default list of phases.
3536       Compiler.Control.LambdaSplitting.* can be used to globally control the
3537       lambda-splitting (cross-module-inlining) engine.  In addition to that,
3538       it can now also be controlled on a per-source basis: CM has been taught
3539       a new tool parameter applicable to ML source files.
3540    
3541       - To turn lambda-splitting off completely:
3542            local open Compiler.Control.LambdaSplitting in
3543                val _ = set Off
3544            end
3545       - To make "no lambda-splitting" the global default (but allow per-source
3546         overriding); this is the initial setting:
3547            local open Compiler.Control.LambdaSplitting in
3548                val _ = set (Default NONE)
3549            end
3550       - To make "lambda-splitting with aggressiveness a" the global default
3551         (and allow per-source overriding):
3552            local open Compiler.Control.LambdaSplitting in
3553                val _ = set (Default (SOME a))
3554            end
3555    
3556       - To turn lambda-splitting off for a given ML souce file (say: a.sml)
3557         write (in the respective .cm-file):
3558            a.sml (lambdasplitting:off)
3559       - To turn lambda-splitting for a.sml on with minimal aggressiveness:
3560            a.sml (lambdasplitting:on)
3561       - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
3562         <a> is a decimal non-negative integer):
3563            a.sml (lambdasplitting:<a>)
3564       - To turn lambda-splitting for a.sml on with maximal aggressiveness:
3565            a.sml (lambdasplitting:infinity)
3566       - To use the global default for a.sml:
3567            a.sml (lambdasplitting:default)
3568         or simply
3569            a.sml
3570    
3571    ----------------------------------------------------------------------
3572    Name: Allen Leung
3573    Date: 2001/05/04 01:57:00 EDT
3574    Tag: leunga-20010504-sync
3575    Description:
3576    
3577      MLRISC features.
3578    
3579      1. Fix to CMPXCHG instructions.
3580      2. Changed RA interface to allow annotations in callbacks.
3581      3. Added a new method to the stream interface to allow annotations updates.
3582    
3583    ----------------------------------------------------------------------
3584    Name: Matthias Blume
3585    Date: 2001/05/01 11:45:00 EDT
3586    Tag: blume-20010501-pcedittmp
3587    Description:
3588    
3589    Changed install.sh to use the current working directory instead of
3590    /usr/tmp for a temporary file (pcedittmp).  The previous choice
3591    of /usr/tmp caused trouble with MacOS X because of file premission
3592    problems.
3593    
3594    ----------------------------------------------------------------------
3595    Name: Matthias Blume
3596    Date: 2001/04/20 11:10:00 EDT
3597    Tag: blume-20010420-inMLflag
3598    Description:
3599    
3600     - added vp_limitPtrMask to vproc-state.h
3601       (for use by the raw-C-calls mechanism to implement proper interrupt
3602        handling)
3603     - made the ML compiler aware of various data-structure offsets so it
3604       can generate code for accessing the vp_inML flag and vp_limitPtrMask
3605     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
3606       raw C-calls
3607    
3608    ----------------------------------------------------------------------
360