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 628, Tue Apr 25 21:03:34 2000 UTC revision 1395, Fri Sep 26 21:03:26 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: 2003/09/26 16:05:00 CDT
18    Tag: blume-20030926-wrappriv
19    Description:
20    
21    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
22    
23    ----------------------------------------------------------------------
24    Name: Matthias Blume (blume (at) tti - c (dot) org)
25    Date: 2003/09/26 15:00:00 CDT
26    Tag: blume-20030926-110_43_3
27    Description:
28    
29    - additional cleanup
30    - version number bump, NEW BOOTFILES
31    
32    ----------------------------------------------------------------------
33    Name: Matthias Blume (blume (at) tti - c (dot) org)
34    Date: 2003/09/26 12:00:00 CDT
35    Tag: blume-20030926-ppautoload
36    Description:
37    
38    I modified the read-eval-print loop so that the autoloader gets
39    invoked whenever the prettyprinter tries to look up a symbol that
40    is not currently defined in the toplevel environment but which
41    appears in CM's autoload registry.  As a result, we see far fewer of
42    those ?.Foo.Bar.xxx names in the prettyprinter's output.
43    
44    In addition to this I tried to clean up some pieces of the Basis
45    implementation (e.g., Socket, Word8Array) in order to prevent other
46    instances of these ?.Foo.Bar.xxx names from being printed.
47    
48    The mechanism that picks names for types still needs some work, though.
49    (Right now it seems that if there is a type A.t which is defined to
50    be B.u, but B is unavailable at toplevel, then A.t gets printed as
51    "?.B.u" although the perhaps more sensible solution would be to use
52    "A.t" in this case.  In other words, the prettyprinter should follow
53    a chain of DEFtycs not farther than there are corresponding toplevel
54    names in the current environment.)
55    
56    ----------------------------------------------------------------------
57    Name: Matthias Blume (blume (at) tti - c (dot) org)
58    Date: 2003/09/24 16:31:00 CDT
59    Tag: blume-20030924-installer
60    Description:
61    
62    Another installer tweak:  All the ML code for the installer is now
63    compiled during CMB.make and put into a little library called
64    $smlnj/installer.cm.  The installation then simply invokes
65    
66       sml -m $smlnj/installer.cm
67    
68    and everything happens automagically.
69    
70    Win32:  ML code senses value of environment variable SMLNJ_HOME.
71    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
72            and BINDIR.
73    
74    The new scheme guarantees that the ML code responsible for the installation
75    is in sync with the APIs of the main system.  Also, the installer is
76    somewhat faster because the installer script is precompiled.
77    
78    ----------------------------------------------------------------------
79    Name: Matthias Blume (blume (at) tti - c (dot) org)
80    Date: 2003/09/24 15:35:00 CDT
81    Tag: blume-20030924-synsock
82    Description:
83    
84    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
85    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
86    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
87    implementations of non-blocking operations from CML's Socket
88    structure.
89    
90    ----------------------------------------------------------------------
91    Name: Matthias Blume (blume (at) tti - c (dot) org)
92    Date: 2003/09/24 15:10:05 CDT
93    Tag: blume-20030924-sockets
94    Description:
95    
96    1. Fixed SOCKET API and implementation to match Basis spec.
97       This required changing the internal representation of sockets to one
98       that remembers (for each socket file descriptor) whether it is currently
99       blocking or non-blocking.  This state is maintained lazily (i.e., a system
100       call is made only if the state actually needs to change).
101    
102    2. OS-specific details of sockets were moved into separate files, thus
103       making it possible to unify the bulk of the socket implementations
104       between Unix and Win32.
105    
106    3. CML's socket API changed accordingly.
107       (Note that we need to remove non-blocking functions from this API
108       since they are redundant in the case of CML!)
109    
110    4. CML's socket implementation now makes use of non-blocking functions
111       provided by Basis, thus removing all OS-dependent code from this part
112       of CML.
113    
114    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
115    
116    ----------------------------------------------------------------------
117    Name: Matthias Blume (blume (at) tti - c (dot) org)
118    Date: 2003/09/22 12:10:00 CDT
119    Tag: blume-20030922-110_43_2
120    Description:
121    
122    Made a new interim version and bootfiles for developer's bootstrapping
123    convenience.
124    
125       110.43.2 -- NEW BOOTFILES
126    
127    ----------------------------------------------------------------------
128    Name: Matthias Blume (blume (at) tti - c (dot) org)
129    Date: 2003/09/19 15:55:00 CDT
130    Tag: blume-20030919-cmdir
131    Description:
132    
133    1. new-install.sh -> install.sh
134    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
135    3. tweaked installer so that another name instead of .cm can be chosen
136       at install time (by setting the CM_DIR_ARC environment variable
137       during installation); once installation is complete, the name is
138       fixed
139    
140    ----------------------------------------------------------------------
141    Name: Matthias Blume (blume (at) tti - c (dot) org)
142    Date: 2003/09/18 16:00:00 CDT
143    Tag: blume-20030918-110_43_1
144    Description:
145    
146    Made a new interim version and bootfiles for developer's bootstrapping
147    convenience.
148    
149       110.43.1 -- NEW BOOTFILES
150    
151    ----------------------------------------------------------------------
152    Name: Matthias Blume (blume (at) tti - c (dot) org)
153    Date: 2003/09/18 15:20:00 CDT
154    Tag: blume-20030918-misc
155    Description:
156    
157    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
158       this seems to be controversial at the moment) and used those in
159       Posix.ProcEnv.times.
160    
161    2. Added Time.{from,to}Nanoseconds to Time.
162    
163    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
164       For example, fromLargeInt never needs to look at more than 3 "big
165       digits" to get its 53 bits of precision.
166    
167    ----------------------------------------------------------------------
168    Name: Matthias Blume (blume (at) tti - c (dot) org)
169    Date: 2003/09/17 16:30:00 CDT
170    Tag: blume-20030917-real32-slices
171    Description:
172    
173    Added an entry to the primitive environment
174    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
175    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
176    
177    Removed some of the "magic" constants in real64.sml and replaced them
178    with code that generates these values from their corresponding
179    integer counterparts.
180    
181    Made all(?) the slice-related changes to the Basis and made everything
182    compile again...
183    
184    ----------------------------------------------------------------------
185    Name: Matthias Blume (blume (at) tti - c (dot) org)
186    Date: 2003/09/15 17:45:00 CDT
187    Tag: blume-20030915-rbase
188    Description:
189    
190    Fixed bug in Real.fromLargeInt.
191    
192    ----------------------------------------------------------------------
193    Name: Matthias Blume (blume (at) tti - c (dot) org)
194    Date: 2003/09/13 18:11:00 CDT
195    Tag: blume-20030913-libinstall
196    Description:
197    
198    Minor bugfix in config/libinstall (set anchor with path to
199    standalone tool after installing it, otherwise libraries that
200    need ml-lex or ml-yacc won't compile the first time the installer
201    runs).
202    
203    ----------------------------------------------------------------------
204    Name: Matthias Blume (blume (at) tti - c (dot) org)
205    Date: 2003/09/12 11:45:00 CDT
206    Tag: blume-20030912-various
207    Description:
208    
209     - fixed bug in Real.toLargeInt
210     - fixed bug in Posix.ProcEnv.times
211     - changed inputLine functions to return an option
212     - minor installer improvements / bugfixes
213     - changed default @SMLalloc parameter for x86/celeron to 64k
214    
215    ----------------------------------------------------------------------
216    Name: Matthias Blume (blume (at) tti - c (dot) org)
217    Date: 2003/09/09 22:00:00 CDT
218    Tag: Release_110_43
219    Description:
220    
221        New working release 110.43.  New bootfiles.
222    
223    ----------------------------------------------------------------------
224    Name: Matthias Blume (blume (at) tti - c (dot) org)
225    Date: 2003/09/09 19:20:00 CDT
226    Tag: blume-20030909-installer
227    Description:
228    
229    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
230    Modified config/install.bat to take advantage of it.  Also modified
231    config/install.sh (and called it config/new-install.sh) to take advantage
232    of it on Unix systems.  (The SML code is (supposed to be) platform-
233    independent.)
234    
235    The installer can now install everything under Win32
236    as well as under *nix as long as it compiles.
237    
238    Other changes:
239    
240      - made CML compile again under Win32
241      - made eXene compile under Win32 (by providing a fake structure UnixSock
242        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
243      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
244        same as type int;  under Win32 it isn't
245      - fixed some slice-related problems in the win32-specific parts of CML
246      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
247        (under Win32, the volume name is case-insensitive, and the
248        OS.Path code compares volume names for equality)
249    
250    ----------------------------------------------------------------------
251    Name: Matthias Blume (blume (at) tti - c (dot) org)
252    Date: 2003/09/08 11:55:00 CDT
253    Tag: blume-20030908-fullpath
254    Description:
255    
256    Made Win32 version of OS.FileSys.fullPath return current directory
257    when given an empty string.  This is what the spec says, and incidentally,
258    CM depends on it.  (CM otherwise goes into an infinite loop in certain
259    cases when presented with the name of a non-existing .cm file.)
260    
261    ----------------------------------------------------------------------
262    Name: Matthias Blume (blume (at) tti - c (dot) org)
263    Date: 2003/09/04 16:30:00 CDT
264    Tag: blume-20030905-slices-etc
265    Description:
266    
267      1. Changed interface to vectors and arrays in Basis to match
268         (draft) Basis spec.
269      2. Added signatures and implementations of slices according to
270         Basis spec.
271      3. Edited source code throughout the system to make it compile again
272         under 1. and 2.  (In some cases code had to be added to have it
273         match the new signatures.)
274      4. MLRISC should be backward-compatible: the copies of the originals
275         of files that needed to change under 3. were retained, the .cm files
276         check the compiler version number and use old versions when
277         appropriate.
278      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
279         dirstream -> string option (in accordance with Basis spec).
280      6. When generating code that counts lines, ml-lex used function
281         CharVector.foldli, taking advantage of its old interface.
282         This has been replaced with the corresponding code from
283         CharVectorSlice.  (html-lex must be re-lexed!)
284      7. BitArray in smlnj-lib/Util has been extended/modified to match the
285         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
286      8. Removed temporary additions (fromInternal, toInternal) from the
287         (now obsolete) IntInf in smlnj-lib/Util.
288      9. Cleaned up structure Byte.
289     10. Added localOffset, scan, and fromString to Date (according to spec).
290         Cleaned/corrected implementation of Date.
291         (Still need to check for correctness; implement better canonicalizeDate.)
292     11. Added "scan" to signature IEEE_REAL.
293     12. Some improvements to IntInf [in particular: efficiency-hack for
294         mod and rem when second operand is 2 (for parity checks).]
295     13. Changed representation of type Time.time, using a single IntInf.int
296         value counting microseconds.  This considerably simplified the
297         implementation of structure Time.  We now support negative time
298         values; scan and fromString handle signs.
299     14. Functor PrimIO now takes two additional arguments (VectorSlice and
300         ArraySlice).
301    
302    ----------------------------------------------------------------------
303    Name: Matthias Blume (blume (at) tti - c (dot) org)
304    Date: 2003/08/28 17:00:00 CDT
305    Tag: blume-20030828-intinf
306    Description:
307    
308    This is a major update which comes with a version number bump
309    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
310    and an implementation of IntInf in the Basis.
311    
312    There are a fairly large number of related changes and updates throughout
313    the system:
314    
315    Basis:
316            - Implemented IntInf.
317            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
318            - Added some missing Real64 operations, most notably Real.toLargeInt.
319            - Added FixedInt as a synonym for Int32.
320    
321    compiler:
322            * Added support for a built-in intinf type.
323              - literals
324              - pattern matching
325              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
326              - overloading on literals and operations
327    
328            This required adding a primitive type intinf, some additional
329            primops, and implementations for several non-trivial intinf
330            operations in Core.  (The intinf type is completely abstract
331            to the compiler; all operations get delegated back to the Core.)
332    
333            * Intinf equality is handled by polyequal.  However, the compiler
334            does not print its usual warning in this case (since polyequal
335            is the right thing to do there).
336    
337            * Improved the organization of structure InlineT.
338    
339            * A word about conversion primops:
340            If conversions involving intinf do not cancel out during
341            CPS contract, then the compiler must insert calls to Core functions.
342            Since all core access must be resolved already during the FLINT
343            translate phase, it would be too late a the time of CPS contract
344            to add new Core calls.  For this reason, conversion primops
345            for intinf carry two arguments:  1. the numeric argument that
346            they are supposed to convert, and 2. the Core function that
347            can help with this conversion if necessary.  If CPS contract
348            eliminates a primop, then the associated Core function becomes
349            dead and goes away.  Intinf conversion primops that do not get
350            eliminated by CPS contract get rewritten into calls of their
351            core functions by a separate, new phase.
352    
353    interactive system:
354            - Control.Print.intinfDepth controls max length of intinf constants
355            being printed. (Analogous to Control.Print.stringDepth.)
356            - Cleanup in printutil and pputil: got rid of unused stuff and
357            duplicates; replaced some of the code with code that makes better
358            use of library functionality.
359    
360    CM:
361            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
362            are no longer silent.
363    
364    CKIT:
365            Fixed mismatched uses of Int32 and LargeInt.  I always decided
366            in favor of LargeInt -- which is now the same as IntInf.
367            CKIT-knowledgable people should check whether this is what's
368            intended and otherwise change things back to using Int32 or
369            FixedInt.
370    
371    Throughout the code:
372            Started using IntInf.int literals and built-in operations
373            (e.g., comparison with 0) where this seems appropriate.
374    
375    
376    ----------------------------------------------------------------------
377    Name: Dave MacQueen (dbm@cs.uchicago.edu)
378    Date: 2003/08/13 11:36:00 CDT
379    Tag: dbm-20030813-mcz-merge1
380    Description:
381    
382    Merging changes from the mcz-branch development branch into trunk.
383    These changes involve replacement of the emulated old prettyprinter
384    interface with direct use of the SML/NJ Lib PP library, and fixing
385    of a couple of bugs (895, 1186) relating to error messages.  A new
386    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
387    has been added.
388    
389    ----------------------------------------------------------------------
390    Name: Matthias Blume (blume (at) tti - c (dot) org)
391    Date: 2003/08/11 15:45:00 CDT
392    Tag: blume-20030811-windows
393    Description:
394    
395    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
396    
397        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
398    
399    This patch restores SML/NJ's ability to run under win32.  There are a
400    number of changes, including fixes for several bugs that had gone
401    unnoticed until now:
402    
403      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
404        trying to run the win32 version from within cygwin.)
405    
406      - There are a number of simple .bat scripts that substitute for their
407        corresponding Unix shell-scripts.  (See below.)
408    
409      - The internals of ml-build have been modified slightly.  The main
410        difference is that instead of calling ".link-sml" (or link-sml.bat)
411        using OS.Process.system, the ML process delegates this task back
412        to the script.  Otherwise problems arise in mixed environments such
413        as Cygwin where scripts look and work like Unix scripts, but
414        where OS.Process.system cannot run them.
415    
416      - In CM, the srcpath pickler used native pathname syntax -- which
417        is incorrect in the case of cross-compilation.  The new pickle format
418        is independent of platform-specific naming conventions.
419    
420      - Path configuration files (such as lib/pathconfig) can now choose
421        between native and standard syntax.  Placing a line of the form
422    
423            standard!
424    
425        into the file causes all subsequent paths to be interpreted using
426        CM standard pathname syntax (= Unix conventions); a line
427    
428            native!
429    
430        switches back to native style.  This was needed so that
431        path config files can be written portably, see src/system/pathconfig.
432    
433      - Runtime system:
434    
435          - win32-filesys.c:  get_file_time and set_file_time now
436            access modification time, not creation time.
437    
438          - I/O code made aware of new array representation.
439    
440          - Bug fixes in X86.prim.masm.
441    
442       - src/system/makeml made aware of win32. (For use under cygwin
443            and other Unix-environments for windows.)
444    
445       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
446         which caused BinIO.inputAll to fail consistently.
447    
448    .bat scripts:
449    
450       Windows .bat scripts assume that SMLNJ_HOME is defined.
451    
452       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
453         applications (sml, ml-yacc, ml-lex).
454       - ml-build.bat: analogous to ml-build.
455       - config\install.bat:  Analogous to config/install.sh.  This requires
456         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
457         (nmake etc. must be on the path, and vcvars32 must have been run.)
458         Moreover, sources for ml-lex and ml-yacc need to exist under src,
459         and the bootfile hierarchy must have been unpacked under
460         sml.boot.x86-win32.
461         The script is very primitive and does a poor job at error checking.
462         It only installs the base system, ml-lex, and ml-yacc.  No other
463         libraries are being installed (i.e., you get only those that
464         are part of the compiler.)
465       - link-sml.bat:  analogous to .link-sml, but not currently used
466    
467    Unrelated bug fixes:
468    
469      - ml-nlffigen now exports structures ST_* corresponding to incomplete
470        types.
471      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
472        otherwise.)
473    
474    ----------------------------------------------------------------------
475    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
476    Date: 2003/06/17
477    Tag: macqueen-20030617-bug895
478    Description:
479    
480    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
481    Tag will be used for new development branch (mcz-branch) for
482    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
483    
484    ----------------------------------------------------------------------
485    Name: Matthias Blume (blume (at) tti - c (dot) org)
486    Date: 2003/05/27 16:55:00 CDT
487    Tag: blume-20030527-polyeq
488    Description:
489    
490    Tried to eliminated most cases of polymorphic equality.
491    
492    ----------------------------------------------------------------------
493    Name: Matthias Blume (blume (at) tti - c (dot) org)
494    Date: 2003/05/21 17:45:00 CDT
495    Tag: blume-20030517-complete
496    Description:
497    
498    Two changes:
499    
500      1. Added a flag for controlling whether non-exhaustive bindings will
501         be treated as errors (default is false).
502      2. Cleaned up the *entire* source tree so that CMB.make goes through
503         without a single non-exhaustive match- or bind warning.
504    
505    ----------------------------------------------------------------------
506    Name: Matthias Blume (blume (at) tti - c (dot) org)
507    Date: 2003/05/17 10:20:00 CDT
508    Tag: blume-20030517-absyn
509    Description:
510    
511    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
512    
513       This mainly affects the quality of error messages.  However, some
514       of the code is now more straightforward than before. (Treatment of
515       the above four constructs in translate.sml is much simpler than
516       the "macro-expansion" that was going on before.  Plus, the mach-
517       compiler no longer gets invoked just to be able to compile an
518       if-expression.)
519    
520    2. The ErrorMsg.Error exception is now caught and absorbed by the
521       interactive loop.
522    
523    ----------------------------------------------------------------------
524    Name: Allen Leung
525    Date: 2003/05/16 13:05:00 CDT
526    Tag: leunga-20030516-cygwin-runtime
527    Description:
528    
529        Ported the runtime system to cygwin, which uses the unix
530    x86-unix bin files.  Missing/buggy features:
531    
532       o getnetbyname, getnetbyaddr: these functions seem to be missing in
533         the Cygwin library.
534       o Ctrl-C handling may be flaky.
535       o Windows system calls and Windows I/O are not supported.
536    
537       A new set of binfiles is located at:
538    
539         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
540    
541       This is only needed for bootstrapping the cygwin version of smlnj.
542       Other x86 versions can use the existing binfiles.
543    
544    ----------------------------------------------------------------------
545    Name: Matthias Blume
546    Date: 2003/04/08 15:42:00 CDT
547    Tag: blume-20030408-listpair
548    Description:
549    
550    1. Added a target 'mlrisc' to installer.
551    
552    2. Added missing elements to structure ListPair.
553    
554    ----------------------------------------------------------------------
555    Name: Allen Leung
556    Date: 2003/01/07 10:40:00 EST
557    Tag: leunga-20030107-int-rem
558    Description:
559    
560    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
561    The arguments to the SUBL instruction were swapped.
562    
563    ----------------------------------------------------------------------
564    Name: Matthias Blume
565    Date: 2002/12/12 16:25:00 EST
566    Tag: blume-20021212-risc-ra
567    Description:
568    
569    Fixed a serious bug in the rewrite code for FP spilling/reloading that
570    sent the RA into an infinite loop when floating point registers get
571    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
572    110.37 and 110.38.)
573    There was another set of potential problems related to the handling of
574    MLRISC annotations (but those did not yet cause real problems, apparently).
575    
576    ----------------------------------------------------------------------
577    Name: Matthias Blume
578    Date: 2002/12/06 22:40:00 EST
579    Tag: blume-20021206-cm-fileid
580    Description:
581    
582    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
583    This fixes the problem of CM getting confused by files that suddenly
584    change their identity (e.g., by getting unlinked and recreated by some
585    text editor such as vi).  There might be a better/cheaper/cleaner way
586    of doing this, but for now this will have to do.
587    
588    ----------------------------------------------------------------------
589    Name: Matthias Blume
590    Date: 2002/10/28 09:50:00 EST
591    Tag: blume-20021028-typecheck
592    Description:
593    
594    Exported structure Typecheck from $smlnj/viscomp/core.cm.
595    
596    ----------------------------------------------------------------------
597    Name: Matthias Blume
598    Date: 2002/10/17 09:10:00 EDT
599    Tag: Release_110_42
600    Description:
601    
602    In good old tradition, there has been a slight hiccup so that we have
603    to patch 110.42 after the fact.   The old release tag has been replaced
604    (see below).
605    
606    The change solves a problem with two competing approaches the
607    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
608    each other's way.
609    
610    This change only affects the runtime system code and the installer script.
611    (No new bootfiles.)
612    
613    ----------------------------------------------------------------------
614    Name: Matthias Blume
615    Date: 2002/10/16 12:00:00 EDT
616    Tag: Release_110_42_removed
617    Description:
618    
619    New working release.  New bootfiles.
620    
621    ----------------------------------------------------------------------
622    Name: Matthias Blume
623    Date: 2002/10/10 13:10:00 EDT
624    Tag: blume-20021010-ppc-divs
625    Description:
626    
627    The mltree operator DIVS must be implemented with an overflow check on
628    the PPC because the hardware indicates divide-by-zero using "overflow" as
629    well.
630    
631    ----------------------------------------------------------------------
632    Name: Matthias Blume
633    Date: 2002/07/23 11:45:00
634    Tag: blume-20020723-smlnj-home
635    Description:
636    
637    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
638    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
639    CM_PATHCONFIG is also set), the path configuration file is assumed
640    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
641    the entire tree to some other place and everything will "just work".
642    
643    (Companion commands such as ml-build and ml-makedepend also sense this
644    variable.)
645    
646    ----------------------------------------------------------------------
647    Name: Matthias Blume
648    Date: 2002/07/12 21:19:00 EDT
649    Tag: blume-20020712-liveness
650    Description:
651    
652    Exported two useful "step" functions from liveness module (MLRISC).
653    
654    ----------------------------------------------------------------------
655    Name: Matthias Blume
656    Date: 2002/07/05 16:00 EDT
657    Tag: Release_110_41
658    Description:
659    
660    New working release.  New bootfiles.
661    
662    ----------------------------------------------------------------------
663    Name: Matthias Blume
664    Date: 2002/07/05 10:25:00 EDT
665    Tag: blume-20020705-btimp
666    Description:
667    
668    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
669    other clients can set up backtracing support.
670    
671    ----------------------------------------------------------------------
672    Name: Matthias Blume
673    Date: 2002/06/25 17:23:00 EDT
674    Tag: blume-20020625-fpmax
675    Description:
676    
677    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
678    case.  (The sense of the isNaN test was reversed -- which made min and
679    max always return their first argument.)
680    
681    ----------------------------------------------------------------------
682    Name: Matthias Blume
683    Date: 2002/06/11
684    Tag: blume-20020611-unixpath
685    Description:
686    
687    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
688    
689    ----------------------------------------------------------------------
690    Name: Matthias Blume
691    Date: 2002/06/10 16:35:00 EDT
692    Tag: blume-20020610-ieeereal
693    Description:
694    
695    I back-ported my implementation of IEEEReal.fromString from the
696    idlbasis-devel branch so that we can test it.
697    
698    Another small change is that ppDec tries to give more information
699    than just "<sig>" in the case of functors.  However, this code is
700    broken in some mysterious way if the functor's body's signature
701    has not been declared by ascription but gets inferred from the
702    implementation.  This needs fixing...
703    
704    ----------------------------------------------------------------------
705    Name: Matthias Blume
706    Date: 2002/05/31
707    Tag: blume-20020531-btrace-mode
708    Description:
709    
710    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
711    the wayside when I switched over to using Controls everywhere.)
712    
713    ----------------------------------------------------------------------
714    Name: Lal George
715    Date: 2002/05/23 12:21:40 EDT
716    Tag: george-20020523-visual-labels
717    Description:
718    
719    Labels are now displayed in the graphical output to make
720    the fall-through and target blocks obvious.
721    
722    ----------------------------------------------------------------------
723    Name: Matthias Blume
724    Date: 2002/05/22 11:03:00 EDT
725    Tag: blume-20020522-shrink
726    Description:
727    
728    John tweaked yesterday's fix for 1131 to handle an out-of-memory
729    situation that comes up when allocating huge arrays.
730    
731    ----------------------------------------------------------------------
732    Name: Matthias Blume
733    Date: 2002/05/21 16:00:00 EDT
734    Tag: Release_110_40
735    Description:
736    
737    New working release (110.40).  New bootfiles.
738    
739    [Also: John Reppy fixed GC bug 1131.]
740    
741    ----------------------------------------------------------------------
742    Name: Matthias Blume
743    Date: 2002/05/21 12:35:00 EDT
744    Tag: blume-20020521-cmdoc
745    Description:
746    
747    CM documentation update.
748    
749    ----------------------------------------------------------------------
750    Name: Matthias Blume
751    Date: 2002/05/21 10:55:00 EDT
752    Tag: blume-20020521-misc
753    Description:
754    
755    - John tweaked runtime to be silent on heap export (except when
756      GC messages are on).
757    - I added a few more things (cross-compiling versions of CMB) to
758      config/preloads (as suggestions).
759    
760    ----------------------------------------------------------------------
761    Name: Matthias Blume
762    Date: 2002/05/20 22:25:00 EDT
763    Tag: blume-20020520-controls
764    Description:
765    
766    - Added ControlUtil structure to control-lib.cm.
767    - Use it throughout.
768    - Used Controls facility to define MLRISC controls (as opposed to
769      registering MLRISC control ref cells with Controls after the fact)
770    - Fixed messed-up controls priorities.
771    
772    * Removed again all the stuff from config/preloads that one wouldn't
773      be able to preload at the time the initial heap image is built.
774      (Many libraries, e.g., CML, do not exist yet at this time.  The
775       only libraries that can be preloaded via config/preloads are those
776       that come bundled with the bootfiles.)
777    
778    ----------------------------------------------------------------------
779    Name: Matthias Blume
780    Date: 2002/05/20 10:59:00 EDT
781    Tag: blume-20020520-preloads
782    Description:
783    
784    Added a lot of commented-out suggestions for things to be included
785    in config/preloads.
786    
787    ----------------------------------------------------------------------
788    Name: Allen Leung
789    Date: 2002/05/18 14:20:00 EDT
790    Tag: leunga-20020518-mdl
791    Description:
792    
793    o Made the mdl tool stuff compile and run again.
794    o I've disabled all the stuff that depends on RTL specifications; they
795      are all badly broken anyway.
796    
797    ----------------------------------------------------------------------
798    Name: Matthias Blume
799    Date: 2002/05/17 16:49:00 EDT
800    Tag: blume-20020517-controls
801    Description:
802    
803    0. John Reppy made several modifications to the SML/NJ library.
804       In particular, there is a shiny new controls-lib.cm.
805    
806    1. Pushed new controls interface through compiler so that everything
807       compiles again.
808    
809    2. Added FormatComb and FORMAT_COMB to the CML version of the
810       SML/NJ library (so that CML compiles again).
811    
812    3. Modified init scripts because XXX_DEFAULT environment variables
813       are no longer with us.  (Boot-time initialization is now done
814       using the same environment variables that are also used for
815       startup-time initialization of controls.)
816    
817    ----------------------------------------------------------------------
818    Name: Lal George
819    Date: 2002/05/15 09:20:10 EDT
820    Tag: george-20020515-pseudo-op-decls
821    Description:
822    
823            All pseudo-ops emitted before the first segment declaration
824    such as TEXT, DATA, and BSS directives are assumed to be global
825    declarations and are emitted first in the assembly file. This is
826    useful in a number of situations where one has pseudo-ops that are not
827    specific to any segment, and also works around the constraint that one
828    cannot have client pseudo-ops in the TEXT segment.
829    
830            Because no segment is associated with these declarations it is
831    an error to allocate any space or objects before the first segment
832    directive and an exception will be raised. However, we cannot make
833    this check for client pseudo-ops.
834    
835            These top level declarations are a field in the CFG graph_info.
836    In theory you can continue to add to this field after the CFG has been
837    built -- provided you know what you are doing;-)
838    
839    ----------------------------------------------------------------------
840    Name: Matthias Blume
841    Date: 2002/05/13 16:40:00 EDT
842    Tag: blume-20020513-pp-etc
843    Description:
844    
845    A few minor bugfixes:
846    
847      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
848        (Bogus printouts for redefined bindings still occur.  Compiler
849        bug should no longer occur now.  We need to redo the prettyprinter
850        from scratch.)
851    
852      - CM pathname printer now also adds escape sequences for ( and )
853    
854      - commend and docu fixes for ml-nlffi
855    
856    ----------------------------------------------------------------------
857    Name: Matthias Blume
858    Date: 2002/05/10 16:40:00 EDT
859    Tag: blume-20020510-erg-textio
860    Description:
861    
862    Applied the following bugfix provided by Emden Gansner:
863    
864        Output is corrupted when outputSubstr is used rather than output.
865    
866        The problem occurs when a substring
867    
868            ss = (s, dataStart, dataLen)
869    
870        where dataStart > 0, fills a stream buffer with avail bytes left.
871        avail bytes of s, starting at index dataStart, are copied into the
872        buffer, the buffer is flushed, and then the remaining dataLen-avail
873        bytes of ss are copied into the beginning of the buffer. Instead of
874        starting this copy at index dataStart+avail in s, the current code
875        starts the copy at index avail.
876    
877        Fix:
878        In text-io-fn.sml, change line 695 from
879             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
880        to
881             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
882    
883    ----------------------------------------------------------------------
884    Name: Matthias Blume
885    Date: 2002/04/12 13:55:00 EDT
886    Tag: blume-20020412-assyntax
887    Description:
888    
889    1. Grabbed newer assyntax.h from the XFree86 project.
890    2. Fiddled with how to compile X86.prim.asm without warnings.
891    3. (Very) Minor cleanup in CM.
892    
893    ----------------------------------------------------------------------
894    Name: Matthias Blume
895    Date: 2002/04/01 (no joke!) 17:07:00 EST
896    Tag: blume-20020401-x86div
897    Description:
898    
899    Added full support for div/mod/rem/quot on the x86, using the machine
900    instruction's two results (without clumsily recomputing the remainder)
901    directly where appropriate.
902    
903    Some more extensive power-of-two support was added to the x86 instruction
904    selector (avoiding expensive divs, mods, and muls where they can be
905    replaced with cheaper shifts and masks).  However, this sort of thing
906    ought to be done earlier, e.g., within the CPS optimizer so that
907    all architectures benefit from it.
908    
909    The compiler compiles to a fixed point, but changes might be somewhat
910    fragile nevertheless.  Please, report any strange things that you might
911    see wrt. div/mod/quot/rem...
912    
913    ----------------------------------------------------------------------
914    Name: Matthias Blume
915    Date: 2002/03/29 17:22:00
916    Tag: blume-20020329-div
917    Description:
918    
919    Fixed my broken div/mod logic.  Unfortunately, this means that the
920    inline code for div/mod now has one more comparison than before.
921    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
922    The problem was with quotient = 0, because that alone does not tell
923    us which way the rounding went.  One then has to look at whether
924    remainder and divisor have the same sign...  :(
925    
926    Anyway, I replaced the bootfiles with fresh ones...
927    
928    ----------------------------------------------------------------------
929    Name: Matthias Blume
930    Date: 2002/03/29 14:10:00 EST
931    Tag: blume-20020329-inlprims
932    Description:
933    
934    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
935    
936    Primops have changed. This means that the bin/boot-file formats have
937    changed as well.
938    
939    To make sure that there is no confusion, I made a new version.
940    
941    
942    CHANGES:
943    
944    * removed REMT from mltree (remainder should never overflow).
945    
946    * added primops to deal with divisions of all flavors to the frontend
947    
948    * handled these primops all the way through so they map to their respective
949      MLRISC support
950    
951    * used these primops in the implementation of Int, Int32, Word, Word32
952    
953    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
954    
955    * parameterized INLMIN, INLMAX, and INLABS by a numkind
956    
957    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
958      floating point
959    
960    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
961      and Real (but Real.abs maps to a separate floating-point-only primop)
962    
963    
964    TODO items:
965    
966    * Hacked Alpha32 instruction selection, disabling the selection of REMx
967      instructions because the machine instruction encoder cannot handle
968      them.  (Hppa, PPC, and Sparc instruction selection did not handle
969      REM in the first place, and REM is supported by the x86 machine coder.)
970    
971    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
972      selection phase.  (The two can be streamlined because the hardware
973      delivers both quotient and remainder at the same time anyway.)
974    
975    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
976      (Currently the behavior is inconsistent both across architectures and
977      wrt. the draft Basis spec.)
978    
979    * Word8 should eventually be handled natively, too.
980    
981    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
982      as if there currently is no execution path that could trigger it in
983      SML/NJ.  (The assumptions underlying functions arith and promotable do not
984      hold for things like multiplication and division.)
985    
986    ----------------------------------------------------------------------
987    Name: Matthias Blume
988    Date: 2002/03/27 16:27:00 EST
989    Tag: blume-20020327-mlrisc-divisions
990    Description:
991    
992    Added support for all four division operations (ML's div, mod, quot,
993    and rem) to MLRISC.  In the course of doing so, I also rationalized
994    the naming (no more annoying switch-around of DIV and QUOT), by
995    parameterizing the operation by div_rounding_mode (which can be either
996    DIV_TO_ZERO or DIV_TO_NEGINF).
997    
998    The generic MLTreeGen functor takes care of compiling all four
999    operations down to only round-to-zero div.
1000    
1001    Missing pieces:
1002    
1003      * Doing something smarter than relying on MLTreeGen on architectures
1004        like, e.g., the x86 where hardware division delivers both quotient and
1005        remainder at the same time.  With this, the implementation of the
1006        round-to-neginf operations could be further streamlined.
1007    
1008      * Remove inlining support for div/mod/rem from the frontend and replace it
1009        with primops that get carried through to the backend.  Do this for all
1010        int and word types.
1011    
1012    ----------------------------------------------------------------------
1013    Name: Matthias Blume
1014    Date: 2002/03/25 17:25:00 EST
1015    Tag: blume-20020325-divmod
1016    Description:
1017    
1018    I improved (hopefully without breaking them) the implementation of Int.div,
1019    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1020    advantage of the following observations:
1021    
1022      Let  q = x quot y      r = x rem y
1023           d = x div  y      m = x mod y
1024    
1025    where "quot" is the round-to-zero version of integer division that
1026    hardware usually provides.  Then we have:
1027    
1028         r = x - q * y        where neither the * nor the - will overflow
1029         d = if q >= 0 orelse x = q * y then q else q - 1
1030                              where neither the * nor the - will overflow
1031         m = if q >= 0 orelse r = 0 then r else r + y
1032                              where the + will not overflow
1033    
1034    This results in substantial simplification of the generated code.
1035    The following table shows the number of CFG nodes and edges generated
1036    for
1037            fun f (x, y) = x OPER y
1038            (* with OPER \in div, mod, quot, rem *)
1039    
1040    
1041        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1042        --------------------------------------------------------
1043         div |         24 |         39 |         12 |         16
1044         mod |         41 |         71 |         12 |         16
1045        quot |          8 |         10 |          8 |         10
1046         rem |         10 |         14 |          8 |         10
1047    
1048    
1049    ----------------------------------------------------------------------
1050    Name: Matthias Blume
1051    Date: 2002/03/25 22:06:00 EST
1052    Tag: blume-20020325-cprotobug
1053    Description:
1054    
1055    Fixed a bug in cproto (c prototype decoder).
1056    
1057    ----------------------------------------------------------------------
1058    Name: Matthias Blume
1059    Date: 2002/03/25 16:00:00 EST
1060    Tag: blume-20020325-raw-primops
1061    Description:
1062    
1063    I did some cleanup to Allen's new primop code and
1064    replaced yesterday's bootfiles with new ones.
1065    (But they are stored in the same place.)
1066    
1067    ----------------------------------------------------------------------
1068    Name: Matthias Blume
1069    Date: 2002/03/24 22:40:00 EST
1070    Tag: blume-20020324-bootfiles
1071    Description:
1072    
1073    Made the bootfiles that Allen asked for.
1074    
1075    ----------------------------------------------------------------------
1076    Name: Allen Leung
1077    Date: 2002/03/23 15:50:00 EST
1078    Tag: leunga-20020323-flint-cps-rcc-primops
1079    Description:
1080    
1081      1. Changes to FLINT primops:
1082    
1083        (* make a call to a C-function;
1084         * The primop carries C function prototype information and specifies
1085         * which of its (ML-) arguments are floating point. C prototype
1086         * information is for use by the backend, ML information is for
1087         * use by the CPS converter. *)
1088      | RAW_CCALL of { c_proto: CTypes.c_proto,
1089                       ml_args: ccall_type list,
1090                       ml_res_opt: ccall_type option,
1091                       reentrant : bool
1092                     } option
1093       (* Allocate uninitialized storage on the heap.
1094        * The record is meant to hold short-lived C objects, i.e., they
1095        * are not ML pointers.  With the tag, the representation is
1096        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1097        *)
1098      | RAW_RECORD of {tag:bool,sz:int}
1099      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1100    
1101      2.  These CPS primops are now overloaded:
1102    
1103           rawload of {kind:numkind}
1104           rawstore of {kind:numkind}
1105    
1106          The one argument form is:
1107    
1108             rawload {kind} address
1109    
1110          The two argument form is:
1111    
1112             rawload {kind} [ml object, byte-offset]
1113    
1114      3. RAW_CCALL/RCC now takes two extra arguments:
1115    
1116         a. The first is whether the C call is reentrant, i.e., whether
1117            ML state should be saved and restored.
1118         b. The second argument is a string argument specifying the name of
1119            library and the C function.
1120    
1121         These things are currently not handled in the code generator, yet.
1122    
1123      4. In CProto,
1124    
1125         An encoding type of "bool" means "ml object" and is mapped into
1126         C prototype of PTR.  Note that "bool" is different than "string",
1127         even though "string" is also mapped into PTR, because "bool"
1128         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1129    
1130      5. Pickler/unpicker
1131    
1132         Changed to handle RAW_RECORD and newest RAW_CCALL
1133    
1134      6. MLRiscGen,
1135    
1136         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1137         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1138            in the file CodeGen/cpscompile/cps-c-calls.sml
1139    
1140      7. Added the conditional move operator
1141    
1142             condmove of branch
1143    
1144         to cps.  Generation of this is still buggy so it is currently
1145         disabled.
1146    
1147    ----------------------------------------------------------------------
1148    Name: Lal George
1149    Date: 2002/03/22 14:18:25 EST
1150    Tag: george-20020322-cps-branch-prob
1151    Description:
1152    
1153    Implemented the Ball-Larus branch prediction-heuristics, and
1154    incorporated graphical viewers for control flow graphs.
1155    
1156    Ball-Larus Heuristics:
1157    ---------------------
1158    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1159    
1160    By design it uses the Dempster-Shafer theory for combining
1161    probabilities.  For example, in the function:
1162    
1163        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1164    
1165    the ball-larus heuristics predicts that the n=0 is unlikely
1166    (OH-heuristic), and the 'then' branch is unlikely because of the
1167    RH-heuristic -- giving the 'then' branch an even lower combined
1168    probability using the Dempster-Shafer theory.
1169    
1170    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1171    probability of the 'then' branch because of the loop in the else
1172    branch.
1173    
1174    
1175    Graphical Viewing:
1176    ------------------
1177    I merely plugged in Allen's graphical viewers into the compiler. The
1178    additional code is not much. At the top level, saying:
1179    
1180            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1181    
1182    will display the graphical view of the control flow graph just before
1183    back-patching.  daVinci must be in your path for this to work. If
1184    daVinci is not available, then the default viewer can be changed
1185    using:
1186    
1187            Control.MLRISC.getString "viewer"
1188    
1189    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1190    course, these viewers must be in your path.
1191    
1192    The above will display the compilation unit at the level of clusters,
1193    many of which are small, boring, and un-interesting. Also setting:
1194    
1195            Control.MLRISC.getInt "cfg-graphical-view_size"
1196    
1197    will display clusters that are larger than the value set by the above.
1198    
1199    
1200    ----------------------------------------------------------------------
1201    Name: Matthias Blume
1202    Date: 2002/03/21 22:20:00 EST
1203    Tag: blume-20020321-kmp-bugfix
1204    Description:
1205    
1206    Changed the interface to the KMP routine in PreString and fixed
1207    a minor bug in one place where it was used.
1208    
1209    ----------------------------------------------------------------------
1210    Name: Allen Leung
1211    Date: 2002/03/21 20:30:00 EST
1212    Tag: leunga-20020321-cfg
1213    Description:
1214    
1215      Fixed a potential problem in cfg edge splitting.
1216    
1217    ----------------------------------------------------------------------
1218    Name: Allen Leung
1219    Date: 2002/03/21 17:15:00 EST
1220    Tag: leunga-20020321-x86-fp-cfg
1221    Description:
1222    
1223      1. Recoded the buggy parts of x86-fp.
1224    
1225         a. All the block reordering code has been removed.
1226            We now depend on the block placement phases to do this work.
1227    
1228         b. Critical edge splitting code has been simplified and moved into the
1229            CFG modules, as where they belong.
1230    
1231         Both of these were quite buggy and complex.  The code is now much, much
1232         simpler.
1233    
1234      2. X86 backend.
1235    
1236         a. Added instructions for 64-bit support.  Instruction selection for
1237            64-bit has not been committed, however, since that
1238            requires changes to MLTREE which haven't been approved by
1239            Lal and John.
1240    
1241         b. Added support for FUCOMI and FUCOMIP when generating code for
1242            PentiumPro and above.  We only generate these instructions in
1243            the fast-fp mode.
1244    
1245         c. Added cases for JP and JNP in X86FreqProps.
1246    
1247      3. CFG
1248    
1249         CFG now has a bunch of methods for edge splitting and merging.
1250    
1251      4. Machine description.
1252    
1253         John's simplification of MLTREE_BASIS.fcond broke a few machine
1254         description things:
1255    
1256         rtl-build.{sig,sml} and hppa.mdl fixed.
1257    
1258         NOTE: the machine description stuff in the repository is still broken.
1259               Again, I can't put my fixes in because that involves
1260               changes to MLTREE.
1261    
1262    ----------------------------------------------------------------------
1263    Name: Matthias Blume
1264    Date: 2002/03/20 15:55:00 EST
1265    Tag: blume-20020320-kmp
1266    Description:
1267    
1268    Implemented Knuth-Morris-Pratt string matching in PreString and used
1269    it for String.isSubstring, Substring.isSubstring, and
1270    Substring.position.
1271    
1272    (Might need some stress-testing.  Simple examples worked fine.)
1273    
1274    ----------------------------------------------------------------------
1275    Name: Matthias Blume
1276    Date: 2002/03/19 16:37:00 EST
1277    Tag: blume-20020319-witnesses
1278    Description:
1279    
1280    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1281    
1282    This implements a generic mechanism for changing constness qualifiers
1283    anywhere within big C types without resorting to outright "casts".
1284    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1285    modify the constness at the outermost level.)
1286    The implementation of "convert" is based on the idea of "witness"
1287    values -- values that are not used by the operation but whose types
1288    "testify" to their applicability.  On the implementation side, "convert"
1289    is simply a projection (returning its second curried argument).  With
1290    cross-module inlining, it should not result in any machine code being
1291    generated.
1292    
1293    ----------------------------------------------------------------------
1294    Name: Matthias Blume
1295    Date: 2002/03/15 16:40:00 EST
1296    Tag: blume-20020315-basis
1297    Description:
1298    
1299    Provided (preliminary?) implementations for
1300    
1301      {String,Substring}.{concatWith,isSuffix,isSubstring}
1302    
1303    and
1304    
1305      Substring.full
1306    
1307    Those are in the Basis spec but they were missing in SML/NJ.
1308    
1309    ----------------------------------------------------------------------
1310    Name: Matthias Blume
1311    Date: 2002/03/14 21:30:00 EST
1312    Tag: blume-20020314-controls
1313    Description:
1314    
1315    Controls:
1316    ---------
1317    
1318    1. Factored out the recently-added Controls : CONTROLS stuff and put
1319       it into its own library $/controls-lib.cm.  The source tree for
1320       this is under src/smlnj-lib/Controls.
1321    
1322    2. Changed the names of types and functions in this interface, so they
1323       make a bit more "sense":
1324    
1325          module -> registry
1326          'a registry -> 'a group
1327    
1328    3. The interface now deals in ref cells only.  The getter/setter interface
1329       is (mostly) gone.
1330    
1331    4. Added a function that lets one register an already-existing ref cell.
1332    
1333    5. Made the corresponding modifications to the rest of the code so that
1334       everything compiles again.
1335    
1336    6. Changed the implementation of Controls.MLRISC back to something closer
1337       to the original.  In particular, this module (and therefore MLRISC)
1338       does not depend on Controls.  There now is some link-time code in
1339       int-sys.sml that registers the MLRISC controls with the Controls
1340       module.
1341    
1342    CM:
1343    ---
1344    
1345      * One can now specify the lambda-split aggressiveness in init.cmi.
1346    
1347    ----------------------------------------------------------------------
1348    Name: Allen Leung
1349    Date: 2002/03/13 17:30:00 EST
1350    Tag: leunga-20020313-x86-fp-unary
1351    Description:
1352    
1353    Bug fix for:
1354    
1355    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1356    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1357    > - fun f(x,(y,z)) = Real.~ y;
1358    > [autoloading]
1359    > [autoloading done]
1360    >       fchsl   (%eax), 184(%esp)
1361    > Error: MLRisc bug: X86MCEmitter.emitInstr
1362    >
1363    > uncaught exception Error
1364    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1365    
1366    The problem was that the code generator did not generate any fp registers
1367    in this case, and the ra didn't know that it needed to run the X86FP phase to
1368    translate the pseudo fp instruction.   This only happened with unary fp
1369    operators in certain situations.
1370    
1371    ----------------------------------------------------------------------
1372    Name: Matthias Blume
1373    Date: 2002/03/13 14:00:00 EST
1374    Tag: blume-20020313-overload-etc
1375    Description:
1376    
1377    1. Added _overload as a synonym for overload for backward compatibility.
1378       (Control.overloadKW must be true for either version to be accepted.)
1379    
1380    2. Fixed bug in install script that caused more things to be installed
1381       than what was requested in config/targets.
1382    
1383    3. Made CM aware of the (_)overload construct so that autoloading
1384       works.
1385    
1386    ----------------------------------------------------------------------
1387    Name: Matthias Blume
1388    Date: 2002/03/12 22:03:00 EST
1389    Tag: blume-20020312-url
1390    Description:
1391    
1392    Forgot to update BOOT and srcarchiveurl.
1393    
1394    ----------------------------------------------------------------------
1395    Name: Matthias Blume
1396    Date: 2002/03/12 17:30:00 EST
1397    Tag: blume-20020312-version110392
1398    Description:
1399    
1400    Yet another version number bump (because of small changes to the
1401    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1402    
1403    Changes:
1404    
1405      The new pid generation scheme described a few weeks ago was overly
1406      complicated.  I implemented a new mechanism that is simpler and
1407      provides a bit more "stability":  Once CM has seen a compilation
1408      unit, it keeps its identity constant (as long as you do not delete
1409      those crucial CM/GUID/* files).  This means that when you change
1410      an interface, compile, then go back to the old interface, and
1411      compile again, you arrive at the original pid.
1412    
1413      There now also is a mechanism that instructs CM to use the plain
1414      environment hash as a module's pid (effectively making its GUID
1415      the empty string).  For this, "noguid" must be specified as an
1416      option to the .sml file in question within its .cm file.
1417      This is most useful for code that is being generated by tools such
1418      as ml-nlffigen (because during development programmers tend to
1419      erase the tool's entire output directory tree including CM's cached
1420      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1421      revert to the old, broken behavior of SML/NJ, but in specific cases
1422      where there is no danger of interface confusion, its use is ok
1423      (I think).
1424    
1425      ml-nlffigen by default generates "noguid" annotations.  They can be
1426      turned off by specifying -guid in its command line.
1427    
1428    ----------------------------------------------------------------------
1429    Name: Lal George
1430    Date: 2002/03/12 12 14:42:36 EST
1431    Tag: george-20020312-frequency-computation
1432    Description:
1433    
1434    Integrated jump chaining and static block frequency into the
1435    compiler. More details and numbers later.
1436    
1437    ----------------------------------------------------------------------
1438    Name: Lal George
1439    Date: 2002/03/11 11 22:38:53 EST
1440    Tag: george-20020311-jump-chain-elim
1441    Description:
1442    
1443    Tested the jump chain elimination on all architectures (except the
1444    hppa).  This is on by default right now and is profitable for the
1445    alpha and x86, however, it may not be profitable for the sparc and ppc
1446    when compiling the compiler.
1447    
1448    The gc test will typically jump to a label at the end of the cluster,
1449    where there is another jump to an external cluster containing the actual
1450    code to invoke gc. This is to allow factoring of common gc invocation
1451    sequences. That is to say, we generate:
1452    
1453            f:
1454               testgc
1455               ja   L1      % jump if above to L1
1456    
1457            L1:
1458               jmp L2
1459    
1460    
1461    After jump chain elimination the 'ja L1' instructions is converted to
1462    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1463    up being implemented in their long form (if L2 is far away) using:
1464    
1465            jbe     L3      % jump if below or equal to L3
1466            jmp     L2
1467         L3:
1468            ...
1469    
1470    
1471    For large compilation units L2  may be far away.
1472    
1473    
1474    ----------------------------------------------------------------------
1475    Name: Matthias Blume
1476    Date: 2002/03/11 13:30:00 EST
1477    Tag: blume-20020311-mltreeeval
1478    Description:
1479    
1480    A functor parameter was missing.
1481    
1482    ----------------------------------------------------------------------
1483    Name: Allen Leung
1484    Date: 2002/03/11 10:30:00 EST
1485    Tag: leunga-20020311-runtime-string0
1486    Description:
1487    
1488       The representation of the empty string now points to a
1489    legal null terminated C string instead of unit.  It is now possible
1490    to convert an ML string into C string with InlineT.CharVector.getData.
1491    This compiles into one single machine instruction.
1492    
1493    ----------------------------------------------------------------------
1494    Name: Allen Leung
1495    Date: 2002/03/10 23:55:00 EST
1496    Tag: leunga-20020310-x86-call
1497    Description:
1498    
1499       Added machine generation for CALL instruction (relative displacement mode)
1500    
1501    ----------------------------------------------------------------------
1502    Name: Matthias Blume
1503    Date: 2002/03/08 16:05:00
1504    Tag: blume-20020308-entrypoints
1505    Description:
1506    
1507    Version number bumped to 110.39.1.  NEW BOOTFILES!
1508    
1509    Entrypoints: non-zero offset into a code object where execution should begin.
1510    
1511    - Added the notion of an entrypoint to CodeObj.
1512    - Added reading/writing of entrypoint info to Binfile.
1513    - Made runtime system bootloader aware of entrypoints.
1514    - Use the address of the label of the first function given to mlriscGen
1515      as the entrypoint.  This address is currently always 0, but it will
1516      not be 0 once we turn on block placement.
1517    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
1518      with entry points) from mlriscGen.
1519    
1520    ----------------------------------------------------------------------
1521    Name: Allen Leung
1522    Date: 2002/03/07 20:45:00 EST
1523    Tag: leunga-20020307-x86-cmov
1524    Description:
1525    
1526       Bug fixes for CMOVcc on x86.
1527    
1528       1. Added machine code generation for CMOVcc
1529       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1530       3. CMOVcc cannot have an immediate operand as argument.
1531    
1532    ----------------------------------------------------------------------
1533    Name: Matthias Blume
1534    Date: 2002/03/07 16:15:00 EST
1535    Tag: blume-20020307-controls
1536    Description:
1537    
1538    This is a very large but mostly boring patch which makes (almost)
1539    every tuneable compiler knob (i.e., pretty much everything under
1540    Control.* plus a few other things) configurable via both the command
1541    line and environment variables in the style CM did its configuration
1542    until now.
1543    
1544    Try starting sml with '-h' (or, if you are brave, '-H')
1545    
1546    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1547    implements the underlying generic mechanism.
1548    
1549    The interface to some of the existing such facilities has changed somewhat.
1550    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1551    (The getFoo interface is still there for backward-compatibility, but its
1552    use is deprecated.)
1553    
1554    The ml-build script passes -Cxxx=yyy command-line arguments through so
1555    that one can now twiddle the compiler settings when using this "batch"
1556    compiler.
1557    
1558    TODO items:
1559    
1560    We should go through and throw out all controls that are no longer
1561    connected to anything.  Moreover, we should go through and provide
1562    meaningful (and correct!) documentation strings for those controls
1563    that still are connected.
1564    
1565    Currently, multiple calls to Controls.new are accepted (only the first
1566    has any effect).  Eventually we should make sure that every control
1567    is being made (via Controls.new) exactly once.  Future access can then
1568    be done using Controls.acc.
1569    
1570    Finally, it would probably be a good idea to use the getter-setter
1571    interface to controls rather than ref cells.  For the time being, both
1572    styles are provided by the Controls module, but getter-setter pairs are
1573    better if thread-safety is of any concern because they can be wrapped.
1574    
1575    *****************************************
1576    
1577    One bug fix: The function blockPlacement in three of the MLRISC
1578    backpatch files used to be hard-wired to one of two possibilities at
1579    link time (according to the value of the placementFlag).  But (I
1580    think) it should rather sense the flag every time.
1581    
1582    *****************************************
1583    
1584    Other assorted changes (by other people who did not supply a HISTORY entry):
1585    
1586    1. the cross-module inliner now works much better (Monnier)
1587    2. representation of weights, frequencies, and probabilities in MLRISC
1588       changed in preparation of using those for weighted block placement
1589       (Reppy, George)
1590    
1591    ----------------------------------------------------------------------
1592    Name: Lal George
1593    Date: 2002/03/07 14:44:24 EST 2002
1594    Tag: george-20020307-weighted-block-placement
1595    
1596    Tested the weighted block placement optimization on all architectures
1597    (except the hppa) using AMPL to generate the block and edge frequencies.
1598    Changes were required in the machine properties to correctly
1599    categorize trap instructions. There is an MLRISC flag
1600    "weighted-block-placement" that can be used to enable weighted block
1601    placement, but this will be ineffective without block/edge
1602    frequencies (coming soon).
1603    
1604    
1605    ----------------------------------------------------------------------
1606    Name: Lal George
1607    Date: 2002/03/05 17:24:48 EST
1608    Tag: george-20020305-linkage-cluster
1609    
1610    In order to support the block placement optimization, a new cluster
1611    is generated as the very first cluster (called the linkage cluster).
1612    It contains a single jump to the 'real' entry point for the compilation
1613    unit. Block placement has no effect on the linkage cluster itself, but
1614    all the other clusters  have full freedom in the manner in which they
1615    reorder blocks or functions.
1616    
1617    On the x86 the typical linkage code that is generated is:
1618       ----------------------
1619            .align 2
1620       L0:
1621            addl    $L1-L0, 72(%esp)
1622            jmp     L1
1623    
1624    
1625            .align  2
1626       L1:
1627       ----------------------
1628    
1629    72(%esp) is the memory location for the stdlink register. This
1630    must contain the address of the CPS function being called. In the
1631    above example, it contains the address of  L0; before
1632    calling L1 (the real entry point for the compilation unit), it
1633    must contain the address for L1, and hence
1634    
1635            addl $L1-L0, 72(%esp)
1636    
1637    I have tested this on all architectures except the hppa.The increase
1638    in code size is of course negligible
1639    
1640    ----------------------------------------------------------------------
1641    Name: Allen Leung
1642    Date: 2002/03/03 13:20:00 EST
1643    Tag: leunga-20020303-mlrisc-tools
1644    
1645      Added #[ ... ] expressions to mlrisc tools
1646    
1647    ----------------------------------------------------------------------
1648    Name: Matthias Blume
1649    Date: 2002/02/27 12:29:00 EST
1650    Tag: blume-20020227-cdebug
1651    Description:
1652    
1653    - made types in structure C and C_Debug to be equal
1654    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1655    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1656    
1657    ----------------------------------------------------------------------
1658    Name: Matthias Blume
1659    Date: 2002/02/26 12:00:00 EST
1660    Tag: blume-20020226-ffi
1661    Description:
1662    
1663    1. Fixed a minor bug in CM's "noweb" tool:
1664       If numbering is turned off, then truly don't number (i.e., do not
1665       supply the -L option to noweb).  The previous behavior was to supply
1666       -L'' -- which caused noweb to use the "default" line numbering scheme.
1667       Thanks to Chris Richards for pointing this out (and supplying the fix).
1668    
1669    2. Once again, I reworked some aspects of the FFI:
1670    
1671       A. The incomplete/complete type business:
1672    
1673       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
1674         gone!
1675       - ML types representing an incomplete type are now *equal* to
1676         ML types representing their corresponding complete types (just like
1677         in C).  This is still safe because ml-nlffigen will not generate
1678         RTTI for incomplete types, nor will it generate functions that
1679         require access to such RTTI.   But when ML code generated from both
1680         incomplete and complete versions of the C type meet, the ML types
1681         are trivially interoperable.
1682    
1683         NOTE:  These changes restore the full generality of the translation
1684         (which was previously lost when I eliminated functorization)!
1685    
1686       B. Enum types:
1687    
1688       - Structure C now has a type constructor "enum" that is similar to
1689         how the "su" constructor works.  However, "enum" is not a phantom
1690         type because each "T enum" has values (and is isomorphic to
1691         MLRep.Signed.int).
1692       - There are generic access operations for enum objects (using
1693         MLRep.Signed.int).
1694       - ml-nlffigen will generate a structure E_foo for each "enum foo".
1695         * The structure contains the definition of type "mlrep" (the ML-side
1696         representation type of the enum).  Normally, mlrep is the same
1697         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
1698         then mlrep will be defined as a datatype -- thus facilitating
1699         pattern matching on mlrep values.
1700         ("-ec" will be suppressed if there are duplicate values in an
1701          enumeration.)
1702         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
1703         will be generated for each C enum constant xxx.
1704         * Conversion functions m2i and i2m convert between mlrep and
1705         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
1706         * Coversion functions c and ml convert between mlrep and "tag enum".
1707         * Access functions (get/set) fetch and store mlrep values.
1708       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
1709         enumerations are merged into one single enumeration represented by
1710         structure E_'.
1711    
1712    ----------------------------------------------------------------------
1713    Name: Allen Leung
1714    Date: 2002/02/25 04:45:00 EST
1715    Tag: leunga-20020225-cps-spill
1716    
1717    This is a new implementation of the CPS spill phase.
1718    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
1719    In case of problems, replace it with the old file spill.sml
1720    
1721    The current compiler runs into some serious performance problems when
1722    constructing a large record.  This can happen when we try to compile a
1723    structure with many items.  Even a very simple structure like the following
1724    makes the compiler slow down.
1725    
1726        structure Foo = struct
1727           val x_1 = 0w1 : Word32.int
1728           val x_2 = 0w2 : Word32.int
1729           val x_3 = 0w3 : Word32.int
1730           ...
1731           val x_N = 0wN : Word32.int
1732        end
1733    
1734    The following table shows the compile time, from N=1000 to N=4000,
1735    with the old compiler:
1736    
1737    N
1738    1000   CPS 100 spill                           0.04u  0.00s  0.00g
1739           MLRISC ra                               0.06u  0.00s  0.05g
1740              (spills = 0 reloads = 0)
1741           TOTAL                                   0.63u  0.07s  0.21g
1742    
1743    1100   CPS 100 spill                           8.25u  0.32s  0.64g
1744           MLRISC ra                               5.68u  0.59s  3.93g
1745              (spills = 0 reloads = 0)
1746           TOTAL                                   14.71u  0.99s  4.81g
1747    
1748    1500   CPS 100 spill                           58.55u  2.34s  1.74g
1749           MLRISC ra                               5.54u  0.65s  3.91g
1750              (spills = 543 reloads = 1082)
1751           TOTAL                                   65.40u  3.13s  6.00g
1752    
1753    2000   CPS 100 spill                           126.69u  4.84s  3.08g
1754           MLRISC ra                               0.80u  0.10s  0.55g
1755              (spills = 42 reloads = 84)
1756           TOTAL                                   129.42u  5.10s  4.13g
1757    
1758    3000   CPS 100 spill                           675.59u  19.03s  11.64g
1759           MLRISC ra                               2.69u  0.27s  1.38g
1760              (spills = 62 reloads = 124)
1761           TOTAL                                   682.48u  19.61s  13.99g
1762    
1763    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
1764           MLRISC ra                               4.96u  0.27s  2.72g
1765              (spills = 85 reloads = 170)
1766           TOTAL                                   2375.26u  57.21s  48.00g
1767    
1768    As you can see the old cps spill module suffers from some serious
1769    performance problem.  But since I cannot decipher the old code fully,
1770    instead of patching the problems up, I'm reimplementing it
1771    with a different algorithm.  The new code is more modular,
1772    smaller when compiled, and substantially faster
1773    (O(n log n) time and O(n) space).  Timing of the new spill module:
1774    
1775    4000  CPS 100 spill                           0.02u  0.00s  0.00g
1776          MLRISC ra                               0.25u  0.02s  0.15g
1777             (spills=1 reloads=3)
1778          TOTAL                                   7.74u  0.34s  1.62g
1779    
1780    Implementation details:
1781    
1782    As far as I can tell, the purpose of the CPS spill module is to make sure the
1783    number of live variables at any program point (the bandwidth)
1784    does not exceed a certain limit, which is determined by the
1785    size of the spill area.
1786    
1787    When the bandwidth is too large, we decrease the register pressure by
1788    packing live variables into spill records.  How we achieve this is
1789    completely different than what we did in the old code.
1790    
1791    First, there is something about the MLRiscGen code generator
1792    that we should be aware of:
1793    
1794    o MLRiscGen performs code motion!
1795    
1796       In particular, it will move floating point computations and
1797       address computations involving only the heap pointer to
1798       their use sites (if there is only a single use).
1799       What this means is that if we have a CPS record construction
1800       statement
1801    
1802           RECORD(k,vl,w,e)
1803    
1804       we should never count the new record address w as live if w
1805       has only one use (which is often the case).
1806    
1807       We should do something similar to floating point, but the transformation
1808       there is much more complex, so I won't deal with that.
1809    
1810    Secondly, there are now two new cps primops at our disposal:
1811    
1812     1. rawrecord of record_kind option
1813        This pure operator allocates some uninitialized storage from the heap.
1814        There are two forms:
1815    
1816         rawrecord NONE [INT n]  allocates a tagless record of length n
1817         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
1818                                     and initializes the tag.
1819    
1820     2. rawupdate of cty
1821          rawupdate cty (v,i,x)
1822          Assigns to x to the ith component of record v.
1823          The storelist is not updated.
1824    
1825    We use these new primops for both spilling and increment record construction.
1826    
1827     1. Spilling.
1828    
1829        This is implemented with a linear scan algorithm (but generalized
1830        to trees).  The algorithm will create a single spill record at the
1831        beginning of the cps function and use rawupdate to spill to it,
1832        and SELECT or SELp to reload from it.  So both spills and reloads
1833        are fine-grain operations.  In contrast, in the old algorithm
1834        "spills" have to be bundled together in records.
1835    
1836        Ideally, we should sink the spill record construction to where
1837        it is needed.  We can even split the spill record into multiple ones
1838        at the places where they are needed.  But CPS is not a good
1839        representation for global code motion, so I'll keep it simple and
1840        am not attempting this.
1841    
1842     2. Incremental record construction (aka record splitting).
1843    
1844        Long records with many component values which are simulatenously live
1845        (recall that single use record addresses are not considered to
1846         be live) are constructed with rawrecord and rawupdate.
1847        We allocate space on the heap with rawrecord first, then gradually
1848        fill it in with rawupdate.  This is the technique suggested to me
1849        by Matthias.
1850    
1851        Some restrictions on when this is applicable:
1852        1. It is not a VECTOR record.  The code generator currently does not handle
1853           this case. VECTOR record uses double indirection like arrays.
1854        2. All the record component values are defined in the same "basic block"
1855           as the record constructor.  This is to prevent speculative
1856           record construction.
1857    
1858    ----------------------------------------------------------------------
1859    Name: Allen Leung
1860    Date: 2002/02/22 01:02:00 EST
1861    Tag: leunga-20020222-mlrisc-tools
1862    
1863    Minor bug fixes in the parser and rewriter
1864    
1865    ----------------------------------------------------------------------
1866    Name: Allen Leung
1867    Date: 2002/02/21 20:20:00 EST
1868    Tag: leunga-20020221-peephole
1869    
1870    Regenerated the peephole files.  Some contained typos in the specification
1871    and some didn't compile because of pretty printing bugs in the old version
1872    of 'nowhere'.
1873    
1874    ----------------------------------------------------------------------
1875    Name: Allen Leung
1876    Date: 2002/02/19 20:20:00 EST
1877    Tag: leunga-20020219-mlrisc-tools
1878    Description:
1879    
1880       Minor bug fixes to the mlrisc-tools library:
1881    
1882       1.  Fixed up parsing colon suffixed keywords
1883       2.  Added the ability to shut the error messages up
1884       3.  Reimplemented the pretty printer and fixed up/improved
1885           the pretty printing of handle and -> types.
1886       4.  Fixed up generation of literal symbols in the nowhere tool.
1887       5.  Added some SML keywords to to sml.sty
1888    
1889    ----------------------------------------------------------------------
1890    Name: Matthias Blume
1891    Date: 2002/02/19 16:20:00 EST
1892    Tag: blume-20020219-cmffi
1893    Description:
1894    
1895    A wild mix of changes, some minor, some major:
1896    
1897    * All C FFI-related libraries are now anchored under $c:
1898        $/c.cm      --> $c/c.cm
1899        $/c-int.cm  --> $c/internals/c-int.cm
1900        $/memory.cm --> $c/memory/memory.cm
1901    
1902    * "make" tool (in CM) now treats its argument pathname slightly
1903      differently:
1904        1. If the native expansion is an absolute name, then before invoking
1905           the "make" command on it, CM will apply OS.Path.mkRelative
1906           (with relativeTo = OS.FileSys.getDir()) to it.
1907        2. The argument will be passed through to subsequent phases of CM
1908           processing without "going native".  In particular, if the argument
1909           was an anchored path, then "make" will not lose track of that anchor.
1910    
1911    * Compiler backends now "know" their respective C calling conventions
1912      instead of having to be told about it by ml-nlffigen.  This relieves
1913      ml-nlffigen from one of its burdens.
1914    
1915    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
1916    
1917    * Export C_DEBUG and C_Debug from $c/c.cm.
1918    
1919    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
1920      subtyping relationship between incomplete pointers and their complete
1921      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
1922      with the convention of instantiating 'o with ('t, 'c) obj whenever
1923      the pointer target type is complete.  In the incomplete case, 'o
1924      will be instantiated with some "'c iobj" -- a type obtained by
1925      using one of the functors PointerToIncompleteType or PointerToCompleteType.
1926    
1927      Operations that work on both incomplete and complete pointer types are
1928      typed as taking an 'o ptr while operations that require the target to
1929      be known are typed as taking some ('t, 'c) obj ptr.
1930    
1931      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
1932      where void is an eqtype without any values.  This makes it possible
1933      to work on voidptr values using functions meant to operate on light
1934      incomplete pointers.
1935    
1936    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
1937      been vastly simplified.
1938    
1939    ----------------------------------------------------------------------
1940    Name: Matthias Blume
1941    Date: 2002/02/19 10:48:00 EST
1942    Tag: blume-20020219-pqfix
1943    Description:
1944    
1945    Applied Chris Okasaki's bug fix for priority queues.
1946    
1947    ----------------------------------------------------------------------
1948    Name: Matthias Blume
1949    Date: 2002/02/15 17:05:00
1950    Tag: Release_110_39
1951    Description:
1952    
1953    Last-minute retagging is becoming a tradition... :-(
1954    
1955    This is the working release 110.39.
1956    
1957    ----------------------------------------------------------------------
1958    Name: Matthias Blume
1959    Date: 2002/02/15 16:00:00 EST
1960    Tag: Release_110_39-orig
1961    Description:
1962    
1963    Working release 110.39.  New bootfiles.
1964    
1965    (Update: There was a small bug in the installer so it wouldn't work
1966    with all shells.  So I retagged. -Matthias)
1967    
1968    ----------------------------------------------------------------------
1969    Name: Matthias Blume
1970    Date: 2002/02/15 14:17:00 EST
1971    Tag: blume-20020215-showbindings
1972    Description:
1973    
1974    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
1975    the latter can be useful for exploring what bindings are available at
1976    the interactive prompt.  (The first function returns only the list
1977    of symbols that are really bound, the second prints those but also the
1978    ones that CM's autoloading mechanism knows about.)
1979    
1980    ----------------------------------------------------------------------
1981    Name: Matthias Blume
1982    Date: 2002/02/15 12:08:00 EST
1983    Tag: blume-20020215-iptrs
1984    Description:
1985    
1986    Two improvements to ml-nlffigen:
1987    
1988      1. Write files only if they do not exist or if their current contents
1989         do not coincide with what's being written.  (That is, avoid messing
1990         with the time stamps unless absolutely necessary.)
1991    
1992      2. Implement a "repository" mechanism for generated files related
1993         to "incomplete pointer types".   See the README file for details.
1994    
1995    ----------------------------------------------------------------------
1996    Name: Matthias Blume
1997    Date: 2002/02/14 11:50:00 EST
1998    Tag: blume-20020214-quote
1999    Description:
2000    
2001    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
2002    because of the new and improved tag generation scheme.  (Thanks to Allen
2003    Leung for pointing it out.)
2004    
2005    ----------------------------------------------------------------------
2006    Name: Lal George
2007    Date: 2002/02/14 09:55:27 EST 2002
2008    Tag: george-20020214-isabelle-bug
2009    Description:
2010    
2011    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2012    of Isabelle on the x86.
2013    
2014    From Allen:
2015    -----------
2016     I've found the problem:
2017    
2018         in ra-core.sml, I use the counter "blocked" to keep track of the
2019         true number of elements in the freeze queue.  When the counter goes
2020         to zero, I skip examining the queue.  But I've messed up the
2021         bookkeeping in combine():
2022    
2023             else ();
2024             case !ucol of
2025               PSEUDO => (if !cntv > 0 then
2026                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2027                                        ^^^^^^^^^^^^^^^^^^^^^^^
2028                      moveu := mergeMoveList(!movev, !moveu)
2029                     )
2030                  else ();
2031    
2032         combine() is called to coalesce two nodes u and v.
2033         I think I was thinking that if the move counts of u and v are both
2034         greater than zero then after they are coalesced then one node is
2035         removed from the freeze queue.  Apparently I was thinking that
2036         both u and v are of low degree, but that's clearly not necessarily true.
2037    
2038    
2039    02/12/2002:
2040        Here's the patch.  HOL now compiles.
2041    
2042        I don't know how this impact on performance (compile
2043        time or runtime).  This bug caused the RA (especially on the x86)
2044        to go thru the potential spill phase when there are still nodes on the
2045        freeze queue.
2046    
2047    
2048    
2049    
2050    ----------------------------------------------------------------------
2051    Name: Matthias Blume
2052    Date: 2002/02/13 22:40:00 EST
2053    Tag: blume-20020213-fptr-rtti
2054    Description:
2055    
2056    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2057    updates.
2058    
2059    ----------------------------------------------------------------------
2060    Name: Matthias Blume
2061    Date: 2002/02/13 16:41:00 EST
2062    Tag: blume-20020213-cmlpq
2063    Description:
2064    
2065    Added new priority queue export symbols (which have just been added to
2066    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2067    not compile and the installer would choke.)
2068    
2069    ----------------------------------------------------------------------
2070    Name: Matthias Blume
2071    Date: 2002/02/13 16:15:00 EST
2072    Tag: blume-20020213-various
2073    Description:
2074    
2075    1. More tweaks to ml-nlffigen:
2076    
2077       - better internal datastructures (resulting in slight speedup)
2078       - "-match" option requires exact match
2079       - "localized" gensym counters (untagged structs/unions nested within
2080         other structs/unions or within typedefs get a fresh counter; their
2081         tag will be prefixed by a concatenation of their parents' tags)
2082       - bug fixes (related to calculation of transitive closure of types
2083         to be included in the output)
2084    
2085    2. Minor Basis updates:
2086    
2087       - added implementations for List.collate and Option.app
2088    
2089    ----------------------------------------------------------------------
2090    Name: Matthias Blume
2091    Date: 2002/02/11 15:55:00 EST
2092    Tag: blume-20020211-gensym
2093    Description:
2094    
2095    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2096    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2097    names (ML structure names that correspond to unnamed C structs, unions,
2098    and enums), so that separate runs of ml-nlffigen do not clash.
2099    
2100    ----------------------------------------------------------------------
2101    Name: Matthias Blume
2102    Date: 2002/02/11 12:05:00 EST
2103    Tag: blume-20020211-gensml
2104    Description:
2105    
2106    A quick fix for a problem with GenSML (in the pgraph-util library):
2107    Make generation of toplevel "local" optional.  (Strictly speaking,
2108    signature definitions within "local" are not legal SML.)
2109    
2110    Other than that: updates to INSTALL and cm/TODO.
2111    
2112    ----------------------------------------------------------------------
2113    Name: Matthias Blume
2114    Date: 2002/02/08 15:00:00 EST
2115    Tag: blume-20020208-uniquepid
2116    Description:
2117    
2118    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2119    
2120    1. The installer (config/install.sh) has gotten smarter:
2121    
2122         - Configuration options are a bit easier to specify now
2123           (in config/targets).
2124         - Bug in recognizing .tar.bz2 files fixed.
2125         - Installer automatically resolves dependencies between
2126           configuration options (e.g., if you ask for eXene, you will
2127           also get cml -- regardless whether you asked for it or not).
2128         - Installer can run in "quieter mode" by setting the environment
2129           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2130           "completely silent", though.
2131         - Build HashCons library as part of smlnj-lib.
2132    
2133    2. A new scheme for assigning persistent identifiers to compilation
2134       units (and, by extension, to types etc.) has been put into place.
2135       This fixes a long-standing bug where types and even dynamic values
2136       can get internally confused, thereby compromising type safety
2137       (abstraction) and dynamic correctness.  See
2138    
2139         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2140    
2141       for an example of how things could go wrong until now.
2142    
2143       The downside of the new scheme is that pids are not quite as
2144       persistent as they used to be: CM will generate a fresh pid
2145       for every compilation unit that it thinks it sees for the first
2146       time.  That means that if you compile starting from a clean, fresh
2147       source tree at two different times, you end up with different
2148       binaries.
2149    
2150       Cutoff recompilation, however, has not been compromised because
2151       CM keeps pid information in special caches between runs.
2152    
2153    ----------------------------------------------------------------------
2154    Name: Lal George
2155    Date: 2002/02/07 15:34:13 EST 2002
2156    Tag: <none>
2157    Description:
2158    
2159    Compilers that generate assembly code may produce  global labels
2160    whose value is resolved at link time. The various peephole optimization
2161    modules did not take this in account.
2162    
2163    TODO. The Labels.addrOf function should really return an option
2164    type so that clients are forced to deal with this issue, rather
2165    than an exception being raised.
2166    
2167    ----------------------------------------------------------------------
2168    Name: Lal George
2169    Date: 2002/02/06 13:55:02 EST
2170    Tag: george-20020206-ra-breakup
2171    Description:
2172    
2173    1. A bug fix from Allen.
2174    
2175        A typo causes extra fstp %st(0)'s to be generated at compensation
2176        edges, which might cause stack underflow traps at runtime.  This
2177        occurs in fft where there are extraneous fstps right before the 'into'
2178        trap instruction (in this case they are harmless since none of the
2179        integers overflow.)
2180    
2181    2. Pulled out various utility modules that were embedded in the modules
2182       of the register allocator. I need these modules for other purposes, but
2183       they are not complete enough to put into a library (just yet).
2184    ----------------------------------------------------------------------
2185    Name: Matthias Blume
2186    Date: 2002/01/31 16:05:00 EST
2187    Tag: blume-20020131-sparc-ccalls
2188    Description:
2189    
2190    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2191       of extra stack space by mistake.  Fixed.
2192    
2193    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2194    
2195    ----------------------------------------------------------------------
2196    Name: Allen Leung
2197    Date: 2002/01/30
2198    Tag: leunga-20020130-nowhere-bug-fix
2199    Description:
2200    
2201       MLRISC bug fixes:
2202       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2203       2. MachineInt.fromString was negating its value.
2204    
2205    ----------------------------------------------------------------------
2206    Name: Matthias Blume
2207    Date: 2002/01/29
2208    Tag: blume-20020129-INSTALL
2209    Description:
2210    
2211    - Added somewhat detailed installation instructions (file INSTALL).
2212    - Fixed curl-detection bug in config/install.sh.
2213    - It is now possible to select the URL getter using the URLGETTER
2214      environment variable:
2215    
2216          not set / "unknown"      --> automatic detection (script tries wget,
2217                                       curl, and lynx)
2218          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2219                                       how to properly invoke them)
2220          other                    --> use $URLGETTER directly, it must take
2221                                       precisely two command-line arguments
2222                                       (source URL and destination file name)
2223    
2224    ----------------------------------------------------------------------
2225    Name: Matthias Blume
2226    Date: 2002/01/28
2227    Tag: blume-20020128-sparc-ccalls
2228    Description:
2229    
2230    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2231    - Make use of the allocParam argument in sparc-c-calls.
2232    
2233    ----------------------------------------------------------------------
2234    Name: Matthias Blume
2235    Date: 2002/01/28
2236    Tag: blume-20020128-allocParam
2237    Description:
2238    
2239    John Reppy:  Changes c-calls API to accept client-callback for
2240    allocating extra stack space.
2241    me: Corresponding changes to mlriscGen (using a dummy argument that
2242        does not change the current behavior).
2243    
2244    ----------------------------------------------------------------------
2245    Name: Matthias Blume
2246    Date: 2002/01/28 12:00:00
2247    Tag: Release_110_38
2248    Description:
2249    
2250    This time for real!!!
2251    
2252    ----------------------------------------------------------------------
2253    Name: Matthias Blume
2254    Date: 2002/01/28 10:56:00 EST
2255    Tag: blume-20020128-retraction
2256    Description:
2257    
2258    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
2259       with blume-Release_110_38-retracted.)
2260    
2261    1. Fixed a problem with incorrect rounding modes in real64.sml.
2262       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2263    
2264    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2265       and enums fixed.  The naming of corresponding ML identifiers should
2266       now be consistent again.
2267    
2268    ----------------------------------------------------------------------
2269    Name: Allen Leung
2270    Date: 2002/01/27
2271    Tag: leunga-20020127-nowhere
2272    Description:
2273    
2274       Added a target called nowhere in the configuration scripts.
2275       Enabling this will build the MLRISC 'nowhere' tool (for translating
2276       programs with where-clauses into legal SML code) during installation.
2277    
2278    ----------------------------------------------------------------------
2279    Name: Matthias Blume
2280    Date: 2002/01/25 21:27:00 EST
2281    Tag: blume-Release_110_38-retracted
2282    Description:
2283    
2284    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
2285    
2286    README will be added later.
2287    
2288    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
2289    (Release_110_38) removed.  Reason: Last-minute bug fixes.
2290    
2291    ----------------------------------------------------------------------
2292    Name: Matthias Blume
2293    Date: 2002/01/25
2294    Tag: blume-20020125-ffi
2295    Description:
2296    
2297    A large number of tweaks and improvements to ml-nlffi-lib and
2298    ml-nlffigen:
2299    
2300       - ML represenation types have been streamlined
2301       - getter and setter functions work with concrete values, not abstract
2302         ones where possible
2303       - ml-nlffigen command line more flexible (see README file there)
2304       - some bugs have been fixed (hopefully)
2305    
2306    ----------------------------------------------------------------------
2307    Name: Lal George
2308    Date: 2002/01/24
2309    Tag: george-20020124-risc-ra-interface
2310    Description:
2311    
2312       There is a dramatic simplification in the interface to the
2313       register allocator for RISC architectures as a result of making
2314       parallel copy instructions explicit.
2315    
2316    ----------------------------------------------------------------------
2317    Name: Matthias Blume
2318    Date: 2002/01/22
2319    Tag: blume-20020122-x86-ccalls
2320    Description:
2321    
2322    Bug fix for c-calls on x86 (having to do with how char- and
2323    short-arguments are being handled).
2324    
2325    ----------------------------------------------------------------------
2326    Name: Matthias Blume
2327    Date: 2002/01/21
2328    Tag: blume-20020121-ff
2329    Description:
2330    
2331    Another day of fiddling with the FFI...
2332    
2333    1. Bug fix/workaround:  CKIT does not complain about negative array
2334       dimensions, so ml-nlffigen has to guard itself against this possibility.
2335       (Otherwise a negative dimension would send it into an infinite loop.)
2336    
2337    2. Some of the abstract types (light objects, light pointers, most "base"
2338       types) in structure C are now eqtypes.
2339    
2340    3. Added constructors and test functions for NULL function pointers.
2341    
2342    ----------------------------------------------------------------------
2343    Name: Matthias Blume
2344    Date: 2002/01/18
2345    Tag: blume-20020118-ready-for-new-release
2346    Description:
2347    
2348    Made config/srcarchiveurl point to a new place.  (Will provide boot
2349    files shortly.)
2350    
2351    Maybe we christen this to be 110.38?
2352    
2353    ----------------------------------------------------------------------
2354    Name: Matthias Blume
2355    Date: 2002/01/18
2356    Tag: blume-20020118-more-ffifiddle
2357    Description:
2358    
2359    Today's FFI fiddling:
2360    
2361      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
2362        C.Set.* that use concrete (MLRep.*) arguments and results instead
2363        of abstract ones.
2364    
2365      - Provided word-style bit operations etc. for "int" representation
2366        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
2367        and Long.
2368    
2369    ----------------------------------------------------------------------
2370    Name: Matthias Blume
2371    Date: 2002/01/18
2372    Tag: blume-20020118-use-x86-fp
2373    Description:
2374    
2375    Now that x86-fast-fp seems to be working, I turned it back on again
2376    by default.  (Seems to work fine now, even with the FFI.)
2377    
2378    Other than that, I added some documentation about the FFI to
2379    src/ml-nlffigen/README and updated the FFI test examples in
2380    src/ml-nlffi-lib/Tests/*.
2381    
2382    ----------------------------------------------------------------------
2383    Name: Allen Leung
2384    Date: 2002/01/17
2385    Tag: leunga-20020117-x86-fast-fp-call
2386    Description:
2387    
2388       1. Fixed a problem with handling return fp values when x86's fast fp
2389          mode is turned on.
2390    
2391       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
2392          of %f32.
2393    
2394       3. Added a constructor INT32lit to the ast of MLRISC tools.
2395    
2396    ----------------------------------------------------------------------
2397    Name: Matthias Blume
2398    Date: 2002/01/16
2399    Tag: blume-20020116-ffifiddle
2400    Description:
2401    
2402    More fiddling with the FFI interface:
2403    
2404     - Make constness 'c instead of rw wherever possible.  This eliminates
2405       the need for certain explicit coercions.  (However, due to ML's
2406       value polymorphism, there will still be many cases where explicit
2407       coercions are necessary.  Phantom types are not the whole answer
2408       to modeling a subtyping relationship in ML.)
2409    
2410     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
2411    
2412     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
2413       clumsy workaround for varargs problem.)
2414    
2415    ----------------------------------------------------------------------
2416    Name: Lal George
2417    Date: 2002/01/15
2418    Tag: <none>
2419    Description:
2420    
2421    1. Since COPY instructions are no longer native to the architecture,
2422       a generic functor can be used to implement the expandCopies function.
2423    
2424    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
2425       TEXT segment.
2426    
2427    ----------------------------------------------------------------------
2428    Name: Matthias Blume
2429    Date: 2002/01/15
2430    Tag: blume-20020115-ffiupdates
2431    Description:
2432    
2433    1. Fix for bug resulting in single-precision float values being returned
2434       incorrectly from FFI calls.
2435    
2436    2. Small modifications to C FFI API:
2437    
2438        - memory-allocation routines return straight objects (no options)
2439          and raise an exception in out-of-memory situations
2440        - unsafe extensions to cast between function pointers and pointers
2441          from/to ints
2442        - added structure C_Debug as an alternative to structure C where
2443          pointer-dereferencing (|*| and |*!) always check for null-pointers
2444        - added open_lib' to DynLinkage;  open_lib' works like open_lib
2445          but also takes a (possibly empty) list of existing library handles
2446          that the current library depends on
2447    
2448    ----------------------------------------------------------------------
2449    Name: Matthias Blume
2450    Date: 2002/01/10
2451    Tag: blume-20020110-newffigen
2452    Description:
2453    
2454    1. Updates to portable graph code.
2455    
2456    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
2457       more scalable now so that even huge interfaces such as the one
2458       for GTK compile in finite time and space. :-)
2459       See src/ml-nlffigen/README for details on what's new.
2460    
2461    ----------------------------------------------------------------------
2462    Name: Lal George
2463    Date: 2001/01/09 14:31:35 EST 2002
2464    Tag: george-20011206-rm-native-copy
2465    Description:
2466    
2467            Removed the native COPY and FCOPY instructions
2468            from all the architectures and replaced it with the
2469            explicit COPY instruction from the previous commit.
2470    
2471            It is now possible to simplify many of the optimizations
2472            modules that manipulate copies. This has not been
2473            done in this change.
2474    
2475    ----------------------------------------------------------------------
2476    Name: Lal George
2477    Date: 2001/12/06 16:50:13 EST 2001
2478    Tag: george-20011206-mlrisc-instruction
2479    Description:
2480    
2481    Changed the representation of instructions from being fully abstract
2482    to being partially concrete. That is to say:
2483    
2484      from
2485            type instruction
2486    
2487      to
2488            type instr                              (* machine instruction *)
2489    
2490            datatype instruction =
2491                LIVE of {regs: C.cellset, spilled: C.cellset}
2492              | KILL of {regs: C.cellset, spilled: C.cellset}
2493              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
2494              | ANNOTATION of {i: instruction, a: Annotations.annotation}
2495              | INSTR of instr
2496    
2497    This makes the handling of certain special instructions that appear on
2498    all architectures easier and uniform.
2499    
2500    LIVE and KILL say that a list of registers are live or killed at the
2501    program point where they appear. No spill code is generated when an
2502    element of the 'regs' field is spilled, but the register is moved to
2503    the 'spilled' (which is present, more for debugging than anything else).
2504    
2505    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
2506    We used to generate:
2507    
2508            DEFFREG f1
2509            f1 := f2 + f3
2510            trapb
2511    
2512    but now generate:
2513    
2514            f1 := f2 + f3
2515            trapb
2516            LIVE {regs=[f1,f2,f3], spilled=[]}
2517    
2518    Furthermore, the DEFFREG (hack) required that all floating point instruction
2519    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
2520    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
2521    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
2522    this architecture).
2523    
2524    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
2525    all the architectures. This will result in further simplification of the
2526    register allocator that must be aware of them for coalescing purposes, and
2527    will also simplify certain aspects of the machine description that provides
2528    callbacks related to parallel copies.
2529    
2530    ANNOTATION should be obvious, and now INSTR represents the honest to God
2531    machine instruction set!
2532    
2533    The <arch>/instructions/<arch>Instr.sml files define certain utility
2534    functions for making porting easier -- essentially converting upper case
2535    to lower case. All machine instructions (of type instr) are in upper case,
2536    and the lower case form generates an MLRISC instruction. For example on
2537    the alpha we have:
2538    
2539      datatype instr =
2540         LDA of {r:cell, b:cell, d:operand}
2541       | ...
2542    
2543      val lda : {r:cell, b:cell, d:operand} -> instruction
2544        ...
2545    
2546    where lda is just (INSTR o LDA), etc.
2547    
2548    ----------------------------------------------------------------------
2549    Name: Matthias Blume
2550    Date: 2001/11/22 21:40:00 EST
2551    Tag: Release_110_37
2552    Description:
2553    
2554    Release 110.37.  This time for real.
2555    
2556    ----------------------------------------------------------------------
2557    Name: Matthias Blume
2558    Date: 2001/11/21 16:35:00 EST
2559    Tag: blume-20011121-foot-in-mouth
2560    Description:
2561    
2562    Removed the "Release_110_37" tag because of a serious bug.
2563    This will be re-tagged once the bug is fixed.
2564    
2565    ----------------------------------------------------------------------
2566    Name: Matthias Blume
2567    Date: 2001/11/21 16:14:00 EST
2568    Tag: blume-20011121-forgottenfile
2569    Description:
2570    
2571    Forgot to add a file.  (Just a .tex-file -- part of
2572    the CM manual source.)
2573    
2574    ----------------------------------------------------------------------
2575    Name: Matthias Blume
2576    Date: 2001/11/21 16:10:00 EST
2577    Tag: blume-20011121-invalid_110_37
2578    Description:
2579    
2580    Note: I removed the original tag "Release_110_37" from this commit
2581          because we found a serious bug in all non-x86 backends.
2582          - Matthias
2583    
2584    1. Modifications to the SML/NJ code generator and to the runtime system
2585       so that code object name strings are directly inserted into code
2586       objects at code generation time.  The only business the runtime system
2587       has with this is now to read the name strings on occasions.
2588       (The encoding of the name string has also changed somewhat.)
2589    
2590    2. CM now implements a simple "set calculus" for specifying export lists.
2591       In particular, it is now possible to refer to the export lists of
2592       other libraries/groups/sources and form unions as well as differences.
2593       See the latest CM manual for details.
2594    
2595    3. An separate notion of "proxy" libraries has again be eliminated from
2596       CM's model.  (Proxy libraries are now simply a special case of using
2597       the export list calculus.)
2598    
2599    4. Some of the existing libraries now take advantage of the new set
2600       calculus.
2601       (Notice that not all libraries have been converted because some
2602       of the existing .cm-files are supposed to be backward compatible
2603       with 110.0.x.)
2604    
2605    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
2606       "General.exnMessage"!  The former relies on a certain hook to be
2607       initialized, and that often does not happen in the stand-alone case.)
2608    
2609    ----------------------------------------------------------------------
2610    Name: Lal George
2611    Date: 2001/11/21  13:56:18 EST
2612    Tag: george-2001121-pseudo-ops
2613    Description:
2614    
2615      Implemented a complete redesign of MLRISC pseudo-ops. Now there
2616      ought to never be any question of incompatabilities with
2617      pseudo-op syntax expected by host assemblers.
2618    
2619      For now, only modules supporting GAS syntax are implemented
2620      but more should follow, such as MASM, and vendor assembler
2621      syntax, e.g. IBM as, Sun as, etc.
2622    
2623    ----------------------------------------------------------------------
2624    Name: Matthias Blume
2625    Date: 2001/11/14 11:52:00 EST
2626    Tag: blume-20011114-srcname
2627    Description:
2628    
2629    1. Routed the name of the current source file to mlriscgen where it
2630       should be directly emitted into the code object.  (This last part
2631       is yet to be done.)
2632    
2633    2. Some cleanup of the pgraph code to make it match the proposal that
2634       I put out the other day.  (The proposal notwithstanding, things are
2635       still in flux here.)
2636    
2637    ----------------------------------------------------------------------
2638    Name: Lal George
2639    Date: 2001/11/14 09:44:04 EST
2640    Tag:
2641    Description:
2642    
2643      Fix for a backpatching bug reported by Allen.
2644    
2645      Because the boundary between short and long span-dependent
2646      instructions is +/- 128, there are an astounding number of
2647      span-dependent instructions whose size is over estimated.
2648    
2649      Allen came up with the idea of letting the size of span
2650      dependent instructions be non-monotonic, for a maxIter
2651      number of times, after which the size must be monotonically
2652      increasing.
2653    
2654      This table shows the number of span-dependent instructions
2655      whose size was over-estimated as a function of maxIter, for the
2656      file Parse/parse/ml.grm.sml:
2657    
2658         maxIter            # of instructions:
2659            10                      687
2660            20                      438
2661            30                      198
2662            40                        0
2663    
2664      In compiling the compiler, there is no significant difference in
2665      compilation speed between maxIter=10 and maxIter=40. Actually,
2666      my measurements showed that maxIter=40 was a tad faster than
2667      maxIter=10! Also 96% of the  files in the compiler reach a fix
2668      point within 13 iterations, so fixing maxIter at 40, while high,
2669      is okay.
2670    
2671    ----------------------------------------------------------------------
2672    Name: Matthias Blume
2673    Date: 2001/10/31 15:25:00 EST
2674    Tag: blume-20011031-pgraph
2675    Description:
2676    
2677    CKIT:
2678    * Changed the "Function" constructor of type Ast.ctype to carry optional
2679      argument identifiers.
2680    * Changed the return type of TypeUtil.getFunction accordingly.
2681    * Type equality ignores the argument names.
2682    * TypeUtil.composite tries to preserve argument names but gives up quickly
2683      if there is a mismatch.
2684    
2685    installation script:
2686    * attempts to use "curl" if available (unless "wget" is available as well)
2687    
2688    CM:
2689    * has an experimental implementation of "portable graphs" which I will
2690      soon propose as an implementation-independent library format
2691    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
2692    
2693    NLFFI-LIB:
2694    * some cleanup (all cosmetic)
2695    
2696    NLFFIGEN:
2697    * temporarily disabled the mechanism that suppresses ML output for
2698      C definitions whose identifiers start with an underscore character
2699    * generate val bindings for enum constants
2700    * user can request that only one style (light or heavy) is being used;
2701      default is to use both (command-line arguments: -heavy and -light)
2702    * fixed bug in handling of function types involving incomplete pointers
2703    * generate ML entry points that take record arguments (i.e., using
2704      named arguments) for C functions that have a prototype with named
2705      arguments
2706      (see changes to CKIT)
2707    
2708    ----------------------------------------------------------------------
2709    Name: Allen Leung
2710    Date: 2001/10/27 20:34:00 EDT
2711    Tag: leunga-20011027-x86-fast-fp-call
2712    Description:
2713    
2714       Fixed the bug described in blume-20010920-slowfp.
2715    
2716       The fix involves
2717          1. generating FCOPYs in FSTP in ia32-svid
2718          2. marking a CALL with the appropriate annotation
2719    
2720    ----------------------------------------------------------------------
2721    Name: Matthias Blume
2722    Date: 2001/10/16 11:32:00 EDT
2723    Tag: blume-20011016-netbsd
2724    Description:
2725    
2726    Underscore patch from Chris Richards (fixing problem with compiling
2727    runtime system under recent NetBSD).
2728    
2729    ----------------------------------------------------------------------
2730    Name: Allen Leung
2731    Date: 2001/10/12 17:18:32 EDT 2001
2732    Tag: leung-20011012-x86-printflowgraph
2733    Description:
2734    
2735    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
2736    
2737    ----------------------------------------------------------------------
2738    Name: Lal George
2739    Date: 2001/10/11 23:51:34 EDT
2740    Tag: george-20011011-too-many-instrs
2741    Description:
2742    
2743    The representation of a program point never expected to see more
2744    than 65536 instructions in a basic block!
2745    
2746    ----------------------------------------------------------------------
2747    Name: Lal George
2748    Date: 2001/10/09 09:41:37 EDT
2749    Tag: george-20011008-mlrisc-labels
2750    Description:
2751    
2752    Changed the machine description files to support printing of
2753    local and global labels in assembly code, based on host assembler
2754    conventions.
2755    
2756    ----------------------------------------------------------------------
2757    Name: Matthias Blume
2758    Date: 2001/09/25 15:25:00 EDT
2759    Tag: blume-20010925-exninfo
2760    Description:
2761    
2762    I provided a non-hook implementation of exnName (at the toplevel) and
2763    made the "dummy" implementation of exnMessage (at the toplevel) more
2764    useful: if nothing gets "hooked in", then at least you are going to
2765    see the exception name and a message indicating why you don't see more.
2766    
2767    [For the time being, programs that need exnMessage and want to use
2768    ml-build should either use General.exnMessage (strongly recommended) or
2769    refer to structure General at some other point so that CM sees a
2770    static dependency.]
2771    
2772    [Similar remarks go for "print" and "use":  If you want to use their
2773    functionality in stand-alone programs generated by ml-build, then use
2774    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
2775    
2776    ----------------------------------------------------------------------
2777    Name: Matthias Blume
2778    Date: 2001/09/20 17:28:00 EDT
2779    Tag: blume-20010920-slowfp
2780    Description:
2781    
2782    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
2783    
2784    ----------------------------------------------------------------------
2785    Name: Matthias Blume
2786    Date: 2001/09/20 17:20:00 EDT
2787    Tag: blume-20010920-canonicalpaths
2788    Description:
2789    
2790    0. Updated the BOOT file (something that I forgot to do earlier).
2791    
2792    1. Small internal change to CM so that it avoids "/../" in filenames
2793       as much as possible (but only where it is safe).
2794    
2795    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
2796       that arguments that contain delimiters are passed through correctly.
2797       This change also means that all "special" arguments of the form
2798       @SMLxxx... must come first.
2799    
2800    3. Changed install script to put relative anchor names for tool commands
2801       into pathconfig.
2802    
2803    ----------------------------------------------------------------------
2804    Name: Matthias Blume
2805    Date: 2001/09/18 15:35:00 EDT
2806    Tag: blume-20010918-readme11036
2807    Description:
2808    
2809    Added README files.
2810    
2811    ----------------------------------------------------------------------
2812    Name: Matthias Blume
2813    Date: 2001/09/18 11:45:00 EDT
2814    Tag: Release_110_36 (retag)
2815    Description:
2816    
2817    Fixed mistake in config/preloads. Retagged as 110.36.
2818    
2819    ----------------------------------------------------------------------
2820    Name: Matthias Blume
2821    Date: 2001/09/18 09:40:00 EDT
2822    Tag: Release_110_36_orig (tag changed)
2823    Description:
2824    
2825    New version (110.36).  New bootfiles.
2826    
2827    ----------------------------------------------------------------------
2828    Name: Matthias Blume
2829    Date: 2001/09/14 16:15:00 EDT
2830    Tag: blume-20010914-x86fastfp
2831    Description:
2832    
2833    John committed some changes that Allen made, in particular a (hopefully)
2834    correctly working version of the x86-fp module.
2835    
2836    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
2837    flag to "true".  Everything seems to compile to a fixpoint ok, and
2838    "mandelbrot" speeds up by about 15%.
2839    
2840    ----------------------------------------------------------------------
2841    Name: Matthias Blume
2842    Date: 2001/09/13 11:20:00 EDT
2843    Tag: blume-20010913-minimal
2844    Description:
2845    
2846    1. Stefan Monnier's patch to fix a miscompilation problem that
2847       was brought to light by John Reppy's work on Moby.
2848    
2849    2. Implemented a minimal "structure Compiler" that contains just
2850       "version" and "architecture".  The minimal version will be
2851       available when the full version is not.  This is for backward-
2852       compatibility with code that wants to test Compiler.version.
2853    
2854    ----------------------------------------------------------------------
2855    Name: Matthias Blume
2856    Date: 2001/08/28 14:03:00 EDT
2857    Tag: blume-20010828-ml-lex
2858    Description:
2859    
2860    Fix for bug 1581, received from Neophytos Michael.
2861    
2862    ----------------------------------------------------------------------
2863    Name: Matthias Blume
2864    Date: 2001/08/27 11:20:00 EDT
2865    Tag: blume-20010827-readme11035
2866    Description:
2867    
2868    Fleshed out the README file for 110.35.
2869    
2870    ----------------------------------------------------------------------
2871    Name: Matthias Blume
2872    Date: 2001/08/24 17:10:00 EDT
2873    Tag: Release_110_35
2874    Description:
2875    
2876    New version number (110.35).  New bootfiles.
2877    
2878    ----------------------------------------------------------------------
2879    Name: Lal George
2880    Date: 2001/08/24 13:47:18 EDT 2001
2881    Tag: george-20010824-MLRISC-graphs
2882    Description:
2883    
2884     removed clusters from MLRISC completely and replaced with graphs.
2885    
2886    ----------------------------------------------------------------------
2887    Name: Matthias Blume
2888    Date: 2001/08/23 17:50:00 EDT
2889    Tag: blume-20010823-toplevel
2890    Description:
2891    
2892    - some reorganization of the code that implements various kinds of
2893      environments in the compiler (static, dynamic, symbolic, combined)
2894    - re-implemented the EnvRef module so that evalStream works properly
2895      (if the stream contains references to "use", "CM.make", etc.)
2896    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
2897    
2898    ----------------------------------------------------------------------
2899    Name: Matthias Blume
2900    Date: 2001/08/20 15:50 EDT
2901    Tag: blume20010820-slipup
2902    Description:
2903    
2904    I forgot to commit a few files.  Here they are...
2905    
2906    ----------------------------------------------------------------------
2907    Name: Matthias Blume
2908    Date: 2001/08/20 15:35:00 EDT
2909    Tag: blume-20010820-debugprof
2910    Description:
2911    
2912    !!!! NEW BOOTFILES !!!!
2913    
2914    This is another round of reorganizing the compiler sources.  This
2915    time the main goal was to factor out all the "instrumentation"
2916    passes (for profiling and backtracing) into their own library.
2917    The difficulty was to do it in such a way that it does not depend
2918    on elaborate.cm but only on elabdata.cm.
2919    
2920    Therefore there have been further changes to both elaborate.cm and
2921    elabdata.cm -- more "generic" things have been moved from the former
2922    to the latter.  As a result, I was forced to split the assignment
2923    of numbers indicating "primtyc"s into two portions: SML-generic and
2924    SML/NJ-specific.  Since it would have been awkward to maintain,
2925    I bit the bullet and actually _changed_ the mapping between these
2926    numbers and primtycs.  The bottom line of this is that you need
2927    a new set of bin- and bootfiles.
2928    
2929    I have built new bootfiles for all architectures, so doing a fresh
2930    checkout and config/install.sh should be all you need.
2931    
2932    The newly created library's name is
2933    
2934        $smlnj/viscomp/debugprof.cm
2935    
2936    and its sources live under
2937    
2938        src/compiler/DebugProf
2939    
2940    ----------------------------------------------------------------------
2941    Name: Matthias Blume
2942    Date: 2001/08/15 17:15:00 EDT
2943    Tag: blume-20010815-compreorg
2944    Description:
2945    
2946    This is a first cut at reorganizing the CM libraries that make up the
2947    core of the compiler.  The idea is to separate out pieces that could
2948    be used independently by tools, e.g., the parser, the typechecker, etc.
2949    
2950    The current status is a step in this direction, but it is not quite
2951    satisfactory yet.  Expect more changes in the future.
2952    
2953    Here is the current (new) organization...
2954    
2955        What used to be $smlnj/viscomp/core.cm is now divided into
2956        six CM libraries:
2957    
2958             $smlnj/viscomp/basics.cm
2959                           /parser.cm
2960                           /elabdata.cm
2961                           /elaborate.cm
2962                           /execute.cm
2963                           /core.cm
2964    
2965        The CM files for these libraries live under src/system/smlnj/viscomp.
2966        All these libraries are proxy libraries that contain precisely
2967        one CM library component.  Here are the locations of the components
2968        (all within the src/compiler tree):
2969    
2970             Basics/basics.cm
2971             Parse/parser.cm
2972             ElabData/elabdata.cm
2973             Elaborator/elaborate.cm
2974             Execution/execute.cm
2975             core.cm
2976    
2977         [This organization is the same that has been used already
2978         for a while for the architecture-specific parts of the visible
2979         compiler and for the old version of core.cm.]
2980    
2981         As you will notice, many source files have been moved from their
2982         respective original locations to a new home in one of the above
2983         subtrees.
2984    
2985         The division of labor between the new libraries is the following:
2986    
2987             basics.cm:
2988                - Simple, basic definitions that pertain to many (or all) of
2989                  the other libraries.
2990             parser.cm:
2991                - The SML parser, producing output of type Ast.dec.
2992                - The type family for Ast is also defined and exported here.
2993             elabdata.cm:
2994                - The datatypes that describe input and output of the elaborator.
2995                  This includes types, absyn, and static environments.
2996             elaborator.cm:
2997                - The SML/NJ type checker and elaborator.
2998                  This maps an Ast.dec (with a given static environment) to
2999                  an Absyn.dec (with a new static environment).
3000                - This libraries implements certain modules that used to be
3001                  structures as functors (to remove dependencies on FLINT).
3002             execute.cm:
3003                - Everything having to do with executing binary code objects.
3004                - Dynamic environments.
3005             core.cm:
3006                - SML/NJ-specific instantiations of the elaborator and MLRISC.
3007                - Top-level modules.
3008                - FLINT (this should eventually become its own library)
3009    
3010    Notes:
3011    
3012    I am not 100% happy with the way I separated the elaborator (and its
3013    data structures) from FLINT.  Two instances of the same problem:
3014    
3015        1. Data structures contain certain fields that carry FLINT-specific
3016           information.  I hacked around this using exn and the property list
3017           module from smlnj-lib.  But the fact that there are middle-end
3018           specific fields around at all is a bit annoying.
3019    
3020        2. The elaborator calculates certain FLINT-related information.  I tried
3021           to make this as abstract as I could using functorization, but, again,
3022           the fact that the elaborator has to perform calculations on behalf
3023           of the middle-end at all is not nice.
3024    
3025        3. Having to used exn and property lists is unfortunate because it
3026           weakens type checking.  The other alternative (parameterizing
3027           nearly *everything*) is not appealing, though.
3028    
3029    I removed the "rebinding =" warning hack because due to the new organization
3030    it was awkward to maintain it.  As a result, the compiler now issues some of
3031    these warnings when compiling init.cmi during bootstrap compilation. On
3032    the plus side, you also get a warning when you do, for example:
3033       val op = = Int32.+
3034    which was not the case up to now.
3035    
3036    I placed "assign" and "deref" into the _Core structure so that the
3037    code that deals with the "lazy" keyword can find them there.  This
3038    removes the need for having access to the primitive environment
3039    during elaboration.
3040    
3041    ----------------------------------------------------------------------
3042    Name: Matthias Blume
3043    Date: 2001/08/13
3044    Tag: blume-20010813-closures
3045    Description:
3046    
3047    This fix was sent to us by Zhong Shao.  It is supposed to improve the
3048    performance of certain loops by avoiding needless closure allocation.
3049    
3050    ----------------------------------------------------------------------
3051    Name: Lal George
3052    Date: 2001/07/31 10:03:23 EDT 2001
3053    Tag: george-20010731-x86-fmalloc
3054    Description: Fixed bug in x86 calls
3055    
3056        There was a bug where call instructions would mysteriously
3057        vanish. The call instruction had to be one that returned
3058        a floating point value.
3059    
3060    ----------------------------------------------------------------------
3061    Name: Lal George
3062    Date: 2001/07/19 16:36:29 EDT 2001
3063    Tag: george-20010719-simple-cells
3064    Description:
3065    
3066    I have dramatically simplified the interface for CELLS in MLRISC.
3067    
3068    In summary, the cells interface is broken up into three parts:
3069    
3070      1. CellsBasis : CELLS_BASIS
3071    
3072            CellsBasis is a top level structure and common for all
3073            architectures.  it contains the definitions of basic datatypes
3074            and utility  functions over these types.
3075    
3076      2. functor Cells() : CELLS
3077    
3078            Cells generates an interface for CELLS that incorporates the
3079            specific resources on the target architecture, such as the
3080            presence of special register classes, their number and size,
3081            and various useful substructures.
3082    
3083      3. <ARCH>CELLS
3084    
3085            e.g. SparcCells: SPARCCELLS
3086    
3087            <ARCH>CELLS usually contains additional bindings for special
3088            registers  on the architecture, such as:
3089    
3090                    val r0 : cell           (* register zero *)
3091                    val y : cell            (* Y register *)
3092                    val psr : cell          (* processor status register *)
3093                    ...
3094    
3095            The structure returned by applying the Cells functor is opened
3096            in this interface.
3097    
3098    The main implication of all this is that the datatypes for cells is
3099    split between CellsBasis and CELLS -- a fairly simple change for user
3100    code.
3101    
3102    In the old scheme the CELLS interface had a definitional binding of
3103    the form:
3104    
3105            signature CELLS = sig
3106    
3107               structure CellsBasis = CellsBasis
3108    
3109               ...
3110    
3111            end
3112    
3113    With all the sharing constraints that goes on in MLRISC, this old
3114    design  quickly leads to errors such as:
3115    
3116            "structure definition spec inside of sharing ... "
3117    
3118    
3119    and appears to require an unacceptable amount of sharing and where
3120    constraint hackery.
3121    
3122    I think this error message (the interaction of definitional specs and
3123    sharing) requires more explanation on our web page.
3124    
3125    ----------------------------------------------------------------------
3126    Name: Matthias Blume
3127    Date: 2001/07/19 15:00:00 EDT
3128    Tag: blume-20010719-libreorg
3129    Description:
3130    
3131    This update puts together a fairly extensive but straightforward change
3132    to the way the libraries that implement the interactive system are
3133    organized:
3134    
3135       The biggest change is the elimination of structure Compiler.  As a
3136       replacement for this structure, there is now a CM library
3137       (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
3138       that exports all the substructures of the original structure Compiler
3139       directly.  So instead of saying Compiler.Foo.bar one now simply
3140       says Foo.bar.  (The CM libraries actually export a collection of
3141       structures that is richer than the collection of substructures of
3142       structure Compiler.)
3143    
3144       To make the transition smooth, there is a separate library called
3145       $smlnj/compiler/compiler.cm which puts together and exports the
3146       original structure Compiler (or at least something very close to it).
3147    
3148       There are five members of the original structure Compiler
3149       that are not exported directly but which instead became members
3150       of a new structure Backend (described by signature BACKEND).  These are:
3151       structure Profile (: PROFILE), structure Compile (: COMPILE), structure
3152       Interact (: INTERACT), structure Machine (: MACHINE), and val
3153       architecture (: string).
3154    
3155       Structure Compiler.Version has become structure CompilerVersion.
3156    
3157       Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
3158       by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
3159       or x86, respectively.
3160       Each of these exports the same frontend structures that
3161       $smlnj/compiler.cm exports.  But they do not have a structure Backend
3162       and instead export some structure <Arch>Backend where <Arch> is Alpha32,
3163       Hppa, PPC, Sparc, or X86, respectively.
3164    
3165       Library $smlnj/compiler/all.cm exports the union of the exports of
3166       $smlnj/compiler/<arch>.cm
3167    
3168       There are no structures <Arch>Compiler anymore, use
3169       $smlnj/compiler/<arch>.cm instead.
3170    
3171       Library host-compiler-0.cm is gone.  Instead, the internal library
3172       that instantiates CM is now called cm0.cm.  Selection of the host
3173       compiler (backend) is no longer done here but. (Responsibility for it
3174       now lies with $smlnj/compiler/current.cm.  This seems to be more
3175       logical.)
3176    
3177       Many individual files have been moved or renamed.  Some files have
3178       been split into multiple files, and some "dead" files have been deleted.
3179    
3180    Aside from these changes to library organization, there are also changes
3181    to the way the code itself is organized:
3182    
3183       Structure Binfile has been re-implemented in such a way that it no
3184       longer needs any knowledge of the compiler.  It exclusively deals
3185       with the details of binfile layout.  It no longer invokes the
3186       compiler (for the purpose of creating new prospective binfile
3187       content), and it no longer has any knowledge of how to interpret
3188       pickles.
3189    
3190       Structure Compile (: COMPILE) has been stripped down to the bare
3191       essentials of compilation.  It no longer deals with linking/execution.
3192       The interface has been cleaned up considerably.
3193    
3194       Utility routines for dealing with linking and execution have been
3195       moved into their own substructures.
3196    
3197       (The ultimate goal of these changes is to provide a light-weight
3198       binfile loader/linker (at least for, e.g., stable libraries) that
3199       does not require CM or the compiler to be present.)
3200    
3201    CM documentation has been updated to reflect the changes to library
3202    organization.
3203    
3204    ----------------------------------------------------------------------
3205    Name: Matthias Blume
3206    Date: 2001/07/10 17:30:00 EDT
3207    Tag: Release_110_34
3208    Description:
3209    
3210    Minor tweak to 110.34 (re-tagged):
3211    
3212      - README.html file added to CVS repository
3213      - runtime compiles properly under FreeBSD 3.X and 4.X
3214    
3215    ----------------------------------------------------------------------
3216    Name: Matthias Blume
3217    Date: 2001/07/10 17:30:00 EDT
3218    Tag: Release_110_34
3219    Description:
3220    
3221    New version number (110.34). New bootfiles.
3222    
3223    ----------------------------------------------------------------------
3224    Name: Matthias Blume
3225    Date: 2001/07/09 16:00:00 EDT
3226    Tag: blume-20010709-more-varargs
3227    Description:
3228    
3229    I changed the handling of varargs in ml-nlffigen again:
3230    The ellipsis ... will now simply be ignored (with an accompanying warning).
3231    
3232    The immediate effect is that you can actually call a varargs function
3233    from ML -- but you can't actually supply any arguments beyond the ones
3234    specified explicitly.  (For example, you can call printf with its format
3235    string, but you cannot pass additional arguments.)
3236    
3237    This behavior is only marginally more useful than the one before, but
3238    it has the advantage that a function or, more importantly, a function
3239    type never gets dropped on the floor, thus avoiding follow-up problems with
3240    other types that refer to the offending one.
3241    
3242    ----------------------------------------------------------------------
3243    Name: Matthias Blume
3244    Date: 2001/07/09 11:25:00 EDT
3245    Tag: blume-20010709-varargs
3246    Description:
3247    
3248    1. ckit-lib.cm now exports structure Error
3249    2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
3250       with a warning accompanied by a source location.  Moreover, it
3251       merely skips the offending function or type and proceeds with the
3252       rest of its work.u  As a result, one can safely feed C code containing
3253       "..." to ml-nlffigen.
3254    3. There are some internal improvements to CM, providing slightly
3255       more general string substitutions in the tools subsystem.
3256    
3257    ----------------------------------------------------------------------
3258    Name: Matthias Blume
3259    Date: 2001/06/27 15:10:00 EDT
3260    Tag: blume-20010627-concur
3261    Description:
3262    
3263    Fixed a small bug in CM's handling of parallel compilation.
3264    (You could observe the bug by Control-C-interrupting an ordinary
3265    CMB.make or CM.stabilize and then attaching some compile servers.
3266    The result was that all of a sudden the previously interrupted
3267    compilation would continue on its own.  This was because of
3268    an over-optimization: CM did not bother to clean out certain queues
3269    when no servers were attached "anyway", resulting in the contents
3270    of these queues to grab control when new servers did get attached.)
3271    
3272    There is also another minor update to the CM manual.
3273    
3274    ----------------------------------------------------------------------
3275    Name: Matthias Blume
3276    Date: 2001/06/26 16:15:00 EDT
3277    Tag: blume-20010626-cmdoc
3278    Description:
3279    
3280    Minor typo fixed in CM manual (syntax diagram for libraries).
3281    
3282    ----------------------------------------------------------------------
3283    Name: Matthias Blume
3284    Date: 2001/06/25 22:55:00 EDT
3285    Tag: blume-20010625-x86pc
3286    Description:
3287    
3288    Fixed a nasty bug in the X86 assembly code that caused signal
3289    handlers to fail (crash) randomly.
3290    
3291    ----------------------------------------------------------------------
3292    Name: Matthias Blume
3293    Date: 2001/06/25 12:05:00 EDT
3294    Tag: blume-20010625-nlffigen
3295    Description:
3296    
3297    This update fixes a number of minor bugs in ml-nlffigen as reported by
3298    Nick Carter <nbc@andrew.cmu.edu>.
3299    
3300      1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
3301      2. Default names for generated files are now derived from the name of
3302         the C file *without its directory*.  In particular, this causes generated
3303         files to be placed locally even if the C file is in some system directory.
3304      3. Default names for generated signatures and structures are also derived
3305         from the C file name without its directory.  This avoids silly things
3306         like "structure GL/GL".
3307         (Other silly names are still possible because ml-nlffigen does not do
3308          a thorough check of whether generated names are legal ML identifiers.
3309          When in doubt, use command line arguments to force particular names.)
3310    
3311    ----------------------------------------------------------------------
3312    Name: Matthias Blume
3313    Date: 2001/06/21 12:25:00 EDT
3314    Tag: blume-20010621-eXene
3315    Description:
3316    
3317    eXene now compiles and (sort of) works again.
3318    
3319    The library name (for version > 110.33) is $/eXene.cm.
3320    
3321    I also added an new example in src/eXene/examples/nbody.  See the
3322    README file there for details.
3323    
3324    ----------------------------------------------------------------------
3325    Name: Matthias Blume
3326    Date: 2001/06/20 16:40:00 EDT
3327    Tag: blume-20010620-cml
3328    Description:
3329    
3330    CML now compiles and works again.
3331    
3332    Libraries (for version > 110.33):
3333    
3334      $cml/cml.cm            Main CML library.
3335      $cml/basis.cm          CML's version of $/basis.cm.
3336      $cml/cml-internal.cm   Internal helper library.
3337      $cml/core-cml.cm       Internal helper library.
3338      $cml-lib/trace-cml.cm  Tracing facility.
3339      $cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
3340    
3341    The installer (config/install.sh) has been taught how to properly
3342    install this stuff.
3343    
3344    ----------------------------------------------------------------------
3345    Name: Matthias Blume
3346    Date: 2001/06/19 17:55:00 EDT
3347    Tag: blume-20010619-instantiate
3348    Description:
3349    
3350    This un-breaks the fix for bug 1432.
3351    (The bug was originally fixed in 110.9 but I broke it again some
3352    time after that.)
3353    
3354    ----------------------------------------------------------------------
3355    Name: Matthias Blume
3356    Date: 2001/06/19 17:25:00 EDT
3357    Tag: blume-20010619-signals
3358    Description:
3359    
3360    This should (hopefully) fix the long-standing signal handling bug.
3361    (The runtime system was constructing a continuation record with an
3362    incorrect descriptor which would cause the GC to drop data on the floor...)
3363    
3364    ----------------------------------------------------------------------
3365    Name: Matthias Blume
3366    Date: 2001/06/15 15:05:00 EDT
3367    Tag: blume-20010615-moresparc
3368    Description:
3369    
3370    Here is a short late-hour update related to Sparc c-calls:
3371    
3372     -- made handling of double-word arguments a bit smarter
3373    
3374     -- instruction selection phase tries to collapse certain clumsily
3375        constructed ML-Trees; typical example:
3376    
3377            ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
3378    
3379        This currently has no further impact on SML/NJ since mlriscGen does
3380        not seem to generate such patterns in the first place, and c-calls
3381        (which did generate them in the beginning) has meanwhile been fixed
3382        so as to avoid them as well.
3383    
3384    ----------------------------------------------------------------------
3385    Name: Matthias Blume
3386    Date: 2001/06/15 15:05:00 EDT
3387    Tag: blume-20010615-sparc
3388    Description:
3389    
3390    The purpose of this update is to provide an implementation of NLFFI
3391    on Sparc machines.
3392    
3393    Here are the changes in detail:
3394    
3395       * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
3396       the Sparc implementation of the c-calls API.
3397       * The Sparc backend of SML/NJ has been modified to uniformely use %fp
3398       for accessing the ML frame.  Thus, we have a real frame pointer and
3399       can freely modify %sp without need for an omit-frame-ptr phase.
3400       The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
3401       this case.
3402       * ml-nlffigen has been taught to produce code for different architectures
3403       and calling conventions.
3404       * In a way similar to what was done in the x86 case, the Sparc
3405       backend uses its own specific extension to mltree.  (For example,
3406       it needs to be able to generate UNIMP instructions which are part
3407       of the calling convention.)
3408       * ml-nlffi-lib was reorganized to make it more modular (in particular,
3409       to make it easier to plug in new machine- and os-dependent parts).
3410    
3411    There are some other fairly unrelated bug fixes and cleanups as well:
3412    
3413       * I further hacked the .cm files for MLRISC tools (like MDLGen) so
3414       that they properly share their libraries with existing SML/NJ libraries.
3415       * I fixed a minor cosmetic bug in CM, supressing certain spurious
3416       follow-up error messages.
3417       * Updates to CM/CMB documentation.
3418    
3419    TODO items:
3420    
3421       * MLRISC should use a different register as its asmTemp on the Sparc.
3422         (The current %o2 is a really bad choice because it is part of the
3423         calling conventions, so things might interfere in unexpected ways.)
3424    
3425    ----------------------------------------------------------------------
3426    Name: Matthias Blume
3427    Date: 2001/06/07
3428    Tag: blume-20010607-calls
3429    Description:
3430    
3431    A number of internal changes related to C calls and calling conventions:
3432    
3433    1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
3434       number of bytes popped implicitly (by the callee).  In most cases
3435       this field is 0 but on x86/win32 it is some non-zero value.  This
3436       is information provided for the benefit of the "omit-frameptr" pass.
3437    2. The CALL instruction on the x86 carries a similar "pops" field.
3438       The instruction selection phase copies its value from the ML-Tree
3439       CALL statement.
3440    3. On all other architectures, the instruction selection phase checks
3441       whether "pops=0" and complains if not.
3442    4. The c-calls implementation for x86 now accepts two calling conventions:
3443       "ccall" and "stdcall".  When "ccall" is selected, the caller cleans
3444       up after the call and pops is set to 0.  For "stdcall", the caller
3445       does nothing, leaving the cleanup to the callee; pops is set to
3446       the number of bytes that were pushed onto the stack.
3447    5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
3448       distinguish between "ccall" and "stdcall".
3449    6. The UNIMP instruction has been added to the supported Sparc instruction
3450       set. (This is needed for implementing the official C calling convention
3451       on this architecture.)
3452    7. I fixed some of the .cm files under src/MLRISC/Tools to make them
3453       work with the latest CM.
3454    
3455    ----------------------------------------------------------------------
3456    Name: Matthias Blume
3457    Date: 2001/06/05 15:10:00 EDT
3458    Tag: blume-20010605-cm-index
3459    Description:
3460    
3461    0. The "lambdasplit" parameter for class "sml" in CM has been documented.
3462    
3463    1. CM can now generate "index files".  These are human-readable files
3464       that list on a per-.cm-file basis each toplevel symbol defined or
3465       imported.  The location of the index file for
3466       <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
3467       To enable index-file generation, set CM.Control.generate_index to true
3468       or export an environment-symbol: export CM_GENERATE_INDEX=true.
3469    
3470       The CM manual has been updated accordingly.
3471    
3472    2. I made some slight modifications to the c-calls API in MLRISC.
3473    
3474         a) There is now a callback to support saving/restoring of
3475            dedicated but caller-save registers around the actual call
3476            instruction.
3477         b) One can optionally specify a comment-annotation for the
3478            call instruction.
3479    
3480    3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
3481       (For example, the comment annotation shows the C prototype of
3482        the function being called.)
3483    
3484    ----------------------------------------------------------------------
3485    Name: Matthias Blume
3486    Date: 2001/06/01 13:30:00 EDT
3487    Tag: blume-20010601-nlffi-cleanup
3488    Description:
3489    
3490    This is mostly a cleanup of MLFFI stuff:
3491    
3492       - some signature files have been put into a more exposed place
3493       - the ugly 'f type parameter is gone (simplifies types tremendously!)
3494       - ml-nlffigen changed accordingly
3495       - tutorial updated
3496    
3497    Other changes:
3498    
3499       - author's affiliation in CM manual(s) updated
3500       - some more recognized keywords added to Allen's sml.sty
3501    
3502    ----------------------------------------------------------------------
3503    Name: Matthias Blume
3504    Date: 2001/05/25 15:30:00 EDT
3505    Tag: blume-20010525-iptr
3506    Description:
3507    
3508      - put the official 110.33-README (as it appears on the ftp server) under
3509        CVS
3510      - fixed a small bug related to incomplete pointer types in
3511        ml-nlffigen
3512      - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
3513        (it does not need the 'f type parameter)
3514    
3515    ----------------------------------------------------------------------
3516    Name: Matthias Blume
3517    Date: 2001/05/23 14:30:00 EDT
3518    Tag: Release_110_33
3519    Description:
3520    
3521    New version number (110.33).  New bootfiles.
3522    
3523    ----------------------------------------------------------------------
3524    Name: Matthias Blume
3525    Date: 2001/05/22 18:06:00 EDT
3526    Tag: blume-20010522-targets
3527    Description:
3528    
3529    Made install.sh use file config/targets.customized if it exists, falling
3530    back to config/targets if it doesn't.  This way one can have a customized
3531    version of the targets file without touching the "real thing", thus
3532    eliminating the constant fear of accidentally checking something bogus
3533    back into the CVS repository...  (File config/targets.customized must
3534    not be added to the repository!)
3535    
3536    ----------------------------------------------------------------------
3537    Name: Matthias Blume
3538    Date: 2001/05/22 16:30:00 EDT
3539    Tag: blume-20010522-minitut
3540    Description:
3541    
3542    1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
3543       struct returns.
3544    2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
3545       incomplete, preliminary documentation for NLFFI.
3546    
3547    ----------------------------------------------------------------------
3548    Name: Matthias Blume
3549    Date: 2001/05/14 11:30:00 EDT
3550    Tag: blume-20010514-script
3551    Description:
3552    
3553    Some bugs in install script fixed.
3554    
3555    In addition to that I also made a slight change to the NLFFI API:
3556    Functors generated by ml-nlffigen now take the dynamic library as a
3557    straight functor argument, not as a suspended one.  (The original
3558    functor code used to force the suspension right away anyway, so there
3559    was nothing gained by this complication of the interface.)
3560    
3561    ----------------------------------------------------------------------
3562    Name: Matthias Blume
3563    Date: 2001/05/11 14:35:00 EDT
3564    Tag: blume-20010511-ml-nlffi
3565    Description:
3566    
3567    I finally took the plunge and added my new FFI code to the main
3568    repository.  For x86-linux it is now ready for prime-time.
3569    
3570    There are two new subdirectories of "src":
3571    
3572      - ml-nlffi-lib:
3573           The utility library for programs using the FFI interface.
3574           Here is the implementation of $/c.cm and its associated low-level
3575           partners $/c-int.cm and $/memory.cm.
3576      - ml-nlffigen:
3577           A stand-alone program for generating ML glue code from C source
3578           code.
3579    
3580    Building ml-nlffigen requires $/ckit-lib.cm.
3581    
3582    The config/install.sh script has been updates to do the Right Thing
3583    (hopefully).
3584    
3585    Notice that the source tree for the C-Kit will not be put under "src"
3586    but directly under the installation root directory.  (This is the
3587    structure that currently exists on the CVS server when you check out
3588    module "sml".)  Fortunately, config/install.sh knows about this oddity.
3589    
3590    Bugs: No documentation yet.
3591    
3592    ----------------------------------------------------------------------
3593    Name: Matthias Blume
3594    Date: 2001/05/09 16:35:00 EDT
3595    Tag: blume-20010509-cpscontract
3596    Description:
3597    
3598    Fixed a bug in the accounting code in cpsopt/contract.sml.  (The
3599    wrapper/unwrapper elimination did not decrement usage counts and some
3600    dead variables got overlooked by the dead-up logic.)
3601    
3602    ----------------------------------------------------------------------
3603    Name: Lal George
3604    Date: 2001/05/08  17:26:09 EDT
3605    Tag: george-20010508-omit-frameptr
3606    Description:
3607    
3608    Changes to implement the omit-frame-pointer optimization to support
3609    raw C calls. For now, there is only support on the Intel x86, but
3610