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 1044, Mon Jan 28 21:36:08 2002 UTC revision 1394, Fri Sep 26 19:56:20 2003 UTC
# Line 13  Line 13 
13  Description:  Description:
14    
15  ----------------------------------------------------------------------  ----------------------------------------------------------------------
16    Name: Matthias Blume (blume (at) tti - c (dot) org)
17    Date: 2003/09/26 15:00:00 CDT
18    Tag: blume-20030926-110_43_3
19    Description:
20    
21    - additional cleanup
22    - version number bump, NEW BOOTFILES
23    
24    ----------------------------------------------------------------------
25    Name: Matthias Blume (blume (at) tti - c (dot) org)
26    Date: 2003/09/26 12:00:00 CDT
27    Tag: blume-20030926-ppautoload
28    Description:
29    
30    I modified the read-eval-print loop so that the autoloader gets
31    invoked whenever the prettyprinter tries to look up a symbol that
32    is not currently defined in the toplevel environment but which
33    appears in CM's autoload registry.  As a result, we see far fewer of
34    those ?.Foo.Bar.xxx names in the prettyprinter's output.
35    
36    In addition to this I tried to clean up some pieces of the Basis
37    implementation (e.g., Socket, Word8Array) in order to prevent other
38    instances of these ?.Foo.Bar.xxx names from being printed.
39    
40    The mechanism that picks names for types still needs some work, though.
41    (Right now it seems that if there is a type A.t which is defined to
42    be B.u, but B is unavailable at toplevel, then A.t gets printed as
43    "?.B.u" although the perhaps more sensible solution would be to use
44    "A.t" in this case.  In other words, the prettyprinter should follow
45    a chain of DEFtycs not farther than there are corresponding toplevel
46    names in the current environment.)
47    
48    ----------------------------------------------------------------------
49    Name: Matthias Blume (blume (at) tti - c (dot) org)
50    Date: 2003/09/24 16:31:00 CDT
51    Tag: blume-20030924-installer
52    Description:
53    
54    Another installer tweak:  All the ML code for the installer is now
55    compiled during CMB.make and put into a little library called
56    $smlnj/installer.cm.  The installation then simply invokes
57    
58       sml -m $smlnj/installer.cm
59    
60    and everything happens automagically.
61    
62    Win32:  ML code senses value of environment variable SMLNJ_HOME.
63    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
64            and BINDIR.
65    
66    The new scheme guarantees that the ML code responsible for the installation
67    is in sync with the APIs of the main system.  Also, the installer is
68    somewhat faster because the installer script is precompiled.
69    
70    ----------------------------------------------------------------------
71    Name: Matthias Blume (blume (at) tti - c (dot) org)
72    Date: 2003/09/24 15:35:00 CDT
73    Tag: blume-20030924-synsock
74    Description:
75    
76    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
77    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
78    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
79    implementations of non-blocking operations from CML's Socket
80    structure.
81    
82    ----------------------------------------------------------------------
83    Name: Matthias Blume (blume (at) tti - c (dot) org)
84    Date: 2003/09/24 15:10:05 CDT
85    Tag: blume-20030924-sockets
86    Description:
87    
88    1. Fixed SOCKET API and implementation to match Basis spec.
89       This required changing the internal representation of sockets to one
90       that remembers (for each socket file descriptor) whether it is currently
91       blocking or non-blocking.  This state is maintained lazily (i.e., a system
92       call is made only if the state actually needs to change).
93    
94    2. OS-specific details of sockets were moved into separate files, thus
95       making it possible to unify the bulk of the socket implementations
96       between Unix and Win32.
97    
98    3. CML's socket API changed accordingly.
99       (Note that we need to remove non-blocking functions from this API
100       since they are redundant in the case of CML!)
101    
102    4. CML's socket implementation now makes use of non-blocking functions
103       provided by Basis, thus removing all OS-dependent code from this part
104       of CML.
105    
106    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
107    
108    ----------------------------------------------------------------------
109    Name: Matthias Blume (blume (at) tti - c (dot) org)
110    Date: 2003/09/22 12:10:00 CDT
111    Tag: blume-20030922-110_43_2
112    Description:
113    
114    Made a new interim version and bootfiles for developer's bootstrapping
115    convenience.
116    
117       110.43.2 -- NEW BOOTFILES
118    
119    ----------------------------------------------------------------------
120    Name: Matthias Blume (blume (at) tti - c (dot) org)
121    Date: 2003/09/19 15:55:00 CDT
122    Tag: blume-20030919-cmdir
123    Description:
124    
125    1. new-install.sh -> install.sh
126    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
127    3. tweaked installer so that another name instead of .cm can be chosen
128       at install time (by setting the CM_DIR_ARC environment variable
129       during installation); once installation is complete, the name is
130       fixed
131    
132    ----------------------------------------------------------------------
133    Name: Matthias Blume (blume (at) tti - c (dot) org)
134    Date: 2003/09/18 16:00:00 CDT
135    Tag: blume-20030918-110_43_1
136    Description:
137    
138    Made a new interim version and bootfiles for developer's bootstrapping
139    convenience.
140    
141       110.43.1 -- NEW BOOTFILES
142    
143    ----------------------------------------------------------------------
144    Name: Matthias Blume (blume (at) tti - c (dot) org)
145    Date: 2003/09/18 15:20:00 CDT
146    Tag: blume-20030918-misc
147    Description:
148    
149    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
150       this seems to be controversial at the moment) and used those in
151       Posix.ProcEnv.times.
152    
153    2. Added Time.{from,to}Nanoseconds to Time.
154    
155    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
156       For example, fromLargeInt never needs to look at more than 3 "big
157       digits" to get its 53 bits of precision.
158    
159    ----------------------------------------------------------------------
160    Name: Matthias Blume (blume (at) tti - c (dot) org)
161    Date: 2003/09/17 16:30:00 CDT
162    Tag: blume-20030917-real32-slices
163    Description:
164    
165    Added an entry to the primitive environment
166    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
167    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
168    
169    Removed some of the "magic" constants in real64.sml and replaced them
170    with code that generates these values from their corresponding
171    integer counterparts.
172    
173    Made all(?) the slice-related changes to the Basis and made everything
174    compile again...
175    
176    ----------------------------------------------------------------------
177    Name: Matthias Blume (blume (at) tti - c (dot) org)
178    Date: 2003/09/15 17:45:00 CDT
179    Tag: blume-20030915-rbase
180    Description:
181    
182    Fixed bug in Real.fromLargeInt.
183    
184    ----------------------------------------------------------------------
185    Name: Matthias Blume (blume (at) tti - c (dot) org)
186    Date: 2003/09/13 18:11:00 CDT
187    Tag: blume-20030913-libinstall
188    Description:
189    
190    Minor bugfix in config/libinstall (set anchor with path to
191    standalone tool after installing it, otherwise libraries that
192    need ml-lex or ml-yacc won't compile the first time the installer
193    runs).
194    
195    ----------------------------------------------------------------------
196    Name: Matthias Blume (blume (at) tti - c (dot) org)
197    Date: 2003/09/12 11:45:00 CDT
198    Tag: blume-20030912-various
199    Description:
200    
201     - fixed bug in Real.toLargeInt
202     - fixed bug in Posix.ProcEnv.times
203     - changed inputLine functions to return an option
204     - minor installer improvements / bugfixes
205     - changed default @SMLalloc parameter for x86/celeron to 64k
206    
207    ----------------------------------------------------------------------
208    Name: Matthias Blume (blume (at) tti - c (dot) org)
209    Date: 2003/09/09 22:00:00 CDT
210    Tag: Release_110_43
211    Description:
212    
213        New working release 110.43.  New bootfiles.
214    
215    ----------------------------------------------------------------------
216    Name: Matthias Blume (blume (at) tti - c (dot) org)
217    Date: 2003/09/09 19:20:00 CDT
218    Tag: blume-20030909-installer
219    Description:
220    
221    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
222    Modified config/install.bat to take advantage of it.  Also modified
223    config/install.sh (and called it config/new-install.sh) to take advantage
224    of it on Unix systems.  (The SML code is (supposed to be) platform-
225    independent.)
226    
227    The installer can now install everything under Win32
228    as well as under *nix as long as it compiles.
229    
230    Other changes:
231    
232      - made CML compile again under Win32
233      - made eXene compile under Win32 (by providing a fake structure UnixSock
234        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
235      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
236        same as type int;  under Win32 it isn't
237      - fixed some slice-related problems in the win32-specific parts of CML
238      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
239        (under Win32, the volume name is case-insensitive, and the
240        OS.Path code compares volume names for equality)
241    
242    ----------------------------------------------------------------------
243    Name: Matthias Blume (blume (at) tti - c (dot) org)
244    Date: 2003/09/08 11:55:00 CDT
245    Tag: blume-20030908-fullpath
246    Description:
247    
248    Made Win32 version of OS.FileSys.fullPath return current directory
249    when given an empty string.  This is what the spec says, and incidentally,
250    CM depends on it.  (CM otherwise goes into an infinite loop in certain
251    cases when presented with the name of a non-existing .cm file.)
252    
253    ----------------------------------------------------------------------
254    Name: Matthias Blume (blume (at) tti - c (dot) org)
255    Date: 2003/09/04 16:30:00 CDT
256    Tag: blume-20030905-slices-etc
257    Description:
258    
259      1. Changed interface to vectors and arrays in Basis to match
260         (draft) Basis spec.
261      2. Added signatures and implementations of slices according to
262         Basis spec.
263      3. Edited source code throughout the system to make it compile again
264         under 1. and 2.  (In some cases code had to be added to have it
265         match the new signatures.)
266      4. MLRISC should be backward-compatible: the copies of the originals
267         of files that needed to change under 3. were retained, the .cm files
268         check the compiler version number and use old versions when
269         appropriate.
270      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
271         dirstream -> string option (in accordance with Basis spec).
272      6. When generating code that counts lines, ml-lex used function
273         CharVector.foldli, taking advantage of its old interface.
274         This has been replaced with the corresponding code from
275         CharVectorSlice.  (html-lex must be re-lexed!)
276      7. BitArray in smlnj-lib/Util has been extended/modified to match the
277         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
278      8. Removed temporary additions (fromInternal, toInternal) from the
279         (now obsolete) IntInf in smlnj-lib/Util.
280      9. Cleaned up structure Byte.
281     10. Added localOffset, scan, and fromString to Date (according to spec).
282         Cleaned/corrected implementation of Date.
283         (Still need to check for correctness; implement better canonicalizeDate.)
284     11. Added "scan" to signature IEEE_REAL.
285     12. Some improvements to IntInf [in particular: efficiency-hack for
286         mod and rem when second operand is 2 (for parity checks).]
287     13. Changed representation of type Time.time, using a single IntInf.int
288         value counting microseconds.  This considerably simplified the
289         implementation of structure Time.  We now support negative time
290         values; scan and fromString handle signs.
291     14. Functor PrimIO now takes two additional arguments (VectorSlice and
292         ArraySlice).
293    
294    ----------------------------------------------------------------------
295    Name: Matthias Blume (blume (at) tti - c (dot) org)
296    Date: 2003/08/28 17:00:00 CDT
297    Tag: blume-20030828-intinf
298    Description:
299    
300    This is a major update which comes with a version number bump
301    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
302    and an implementation of IntInf in the Basis.
303    
304    There are a fairly large number of related changes and updates throughout
305    the system:
306    
307    Basis:
308            - Implemented IntInf.
309            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
310            - Added some missing Real64 operations, most notably Real.toLargeInt.
311            - Added FixedInt as a synonym for Int32.
312    
313    compiler:
314            * Added support for a built-in intinf type.
315              - literals
316              - pattern matching
317              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
318              - overloading on literals and operations
319    
320            This required adding a primitive type intinf, some additional
321            primops, and implementations for several non-trivial intinf
322            operations in Core.  (The intinf type is completely abstract
323            to the compiler; all operations get delegated back to the Core.)
324    
325            * Intinf equality is handled by polyequal.  However, the compiler
326            does not print its usual warning in this case (since polyequal
327            is the right thing to do there).
328    
329            * Improved the organization of structure InlineT.
330    
331            * A word about conversion primops:
332            If conversions involving intinf do not cancel out during
333            CPS contract, then the compiler must insert calls to Core functions.
334            Since all core access must be resolved already during the FLINT
335            translate phase, it would be too late a the time of CPS contract
336            to add new Core calls.  For this reason, conversion primops
337            for intinf carry two arguments:  1. the numeric argument that
338            they are supposed to convert, and 2. the Core function that
339            can help with this conversion if necessary.  If CPS contract
340            eliminates a primop, then the associated Core function becomes
341            dead and goes away.  Intinf conversion primops that do not get
342            eliminated by CPS contract get rewritten into calls of their
343            core functions by a separate, new phase.
344    
345    interactive system:
346            - Control.Print.intinfDepth controls max length of intinf constants
347            being printed. (Analogous to Control.Print.stringDepth.)
348            - Cleanup in printutil and pputil: got rid of unused stuff and
349            duplicates; replaced some of the code with code that makes better
350            use of library functionality.
351    
352    CM:
353            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
354            are no longer silent.
355    
356    CKIT:
357            Fixed mismatched uses of Int32 and LargeInt.  I always decided
358            in favor of LargeInt -- which is now the same as IntInf.
359            CKIT-knowledgable people should check whether this is what's
360            intended and otherwise change things back to using Int32 or
361            FixedInt.
362    
363    Throughout the code:
364            Started using IntInf.int literals and built-in operations
365            (e.g., comparison with 0) where this seems appropriate.
366    
367    
368    ----------------------------------------------------------------------
369    Name: Dave MacQueen (dbm@cs.uchicago.edu)
370    Date: 2003/08/13 11:36:00 CDT
371    Tag: dbm-20030813-mcz-merge1
372    Description:
373    
374    Merging changes from the mcz-branch development branch into trunk.
375    These changes involve replacement of the emulated old prettyprinter
376    interface with direct use of the SML/NJ Lib PP library, and fixing
377    of a couple of bugs (895, 1186) relating to error messages.  A new
378    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
379    has been added.
380    
381    ----------------------------------------------------------------------
382    Name: Matthias Blume (blume (at) tti - c (dot) org)
383    Date: 2003/08/11 15:45:00 CDT
384    Tag: blume-20030811-windows
385    Description:
386    
387    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
388    
389        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
390    
391    This patch restores SML/NJ's ability to run under win32.  There are a
392    number of changes, including fixes for several bugs that had gone
393    unnoticed until now:
394    
395      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
396        trying to run the win32 version from within cygwin.)
397    
398      - There are a number of simple .bat scripts that substitute for their
399        corresponding Unix shell-scripts.  (See below.)
400    
401      - The internals of ml-build have been modified slightly.  The main
402        difference is that instead of calling ".link-sml" (or link-sml.bat)
403        using OS.Process.system, the ML process delegates this task back
404        to the script.  Otherwise problems arise in mixed environments such
405        as Cygwin where scripts look and work like Unix scripts, but
406        where OS.Process.system cannot run them.
407    
408      - In CM, the srcpath pickler used native pathname syntax -- which
409        is incorrect in the case of cross-compilation.  The new pickle format
410        is independent of platform-specific naming conventions.
411    
412      - Path configuration files (such as lib/pathconfig) can now choose
413        between native and standard syntax.  Placing a line of the form
414    
415            standard!
416    
417        into the file causes all subsequent paths to be interpreted using
418        CM standard pathname syntax (= Unix conventions); a line
419    
420            native!
421    
422        switches back to native style.  This was needed so that
423        path config files can be written portably, see src/system/pathconfig.
424    
425      - Runtime system:
426    
427          - win32-filesys.c:  get_file_time and set_file_time now
428            access modification time, not creation time.
429    
430          - I/O code made aware of new array representation.
431    
432          - Bug fixes in X86.prim.masm.
433    
434       - src/system/makeml made aware of win32. (For use under cygwin
435            and other Unix-environments for windows.)
436    
437       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
438         which caused BinIO.inputAll to fail consistently.
439    
440    .bat scripts:
441    
442       Windows .bat scripts assume that SMLNJ_HOME is defined.
443    
444       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
445         applications (sml, ml-yacc, ml-lex).
446       - ml-build.bat: analogous to ml-build.
447       - config\install.bat:  Analogous to config/install.sh.  This requires
448         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
449         (nmake etc. must be on the path, and vcvars32 must have been run.)
450         Moreover, sources for ml-lex and ml-yacc need to exist under src,
451         and the bootfile hierarchy must have been unpacked under
452         sml.boot.x86-win32.
453         The script is very primitive and does a poor job at error checking.
454         It only installs the base system, ml-lex, and ml-yacc.  No other
455         libraries are being installed (i.e., you get only those that
456         are part of the compiler.)
457       - link-sml.bat:  analogous to .link-sml, but not currently used
458    
459    Unrelated bug fixes:
460    
461      - ml-nlffigen now exports structures ST_* corresponding to incomplete
462        types.
463      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
464        otherwise.)
465    
466    ----------------------------------------------------------------------
467    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
468    Date: 2003/06/17
469    Tag: macqueen-20030617-bug895
470    Description:
471    
472    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
473    Tag will be used for new development branch (mcz-branch) for
474    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
475    
476    ----------------------------------------------------------------------
477    Name: Matthias Blume (blume (at) tti - c (dot) org)
478    Date: 2003/05/27 16:55:00 CDT
479    Tag: blume-20030527-polyeq
480    Description:
481    
482    Tried to eliminated most cases of polymorphic equality.
483    
484    ----------------------------------------------------------------------
485    Name: Matthias Blume (blume (at) tti - c (dot) org)
486    Date: 2003/05/21 17:45:00 CDT
487    Tag: blume-20030517-complete
488    Description:
489    
490    Two changes:
491    
492      1. Added a flag for controlling whether non-exhaustive bindings will
493         be treated as errors (default is false).
494      2. Cleaned up the *entire* source tree so that CMB.make goes through
495         without a single non-exhaustive match- or bind warning.
496    
497    ----------------------------------------------------------------------
498    Name: Matthias Blume (blume (at) tti - c (dot) org)
499    Date: 2003/05/17 10:20:00 CDT
500    Tag: blume-20030517-absyn
501    Description:
502    
503    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
504    
505       This mainly affects the quality of error messages.  However, some
506       of the code is now more straightforward than before. (Treatment of
507       the above four constructs in translate.sml is much simpler than
508       the "macro-expansion" that was going on before.  Plus, the mach-
509       compiler no longer gets invoked just to be able to compile an
510       if-expression.)
511    
512    2. The ErrorMsg.Error exception is now caught and absorbed by the
513       interactive loop.
514    
515    ----------------------------------------------------------------------
516    Name: Allen Leung
517    Date: 2003/05/16 13:05:00 CDT
518    Tag: leunga-20030516-cygwin-runtime
519    Description:
520    
521        Ported the runtime system to cygwin, which uses the unix
522    x86-unix bin files.  Missing/buggy features:
523    
524       o getnetbyname, getnetbyaddr: these functions seem to be missing in
525         the Cygwin library.
526       o Ctrl-C handling may be flaky.
527       o Windows system calls and Windows I/O are not supported.
528    
529       A new set of binfiles is located at:
530    
531         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
532    
533       This is only needed for bootstrapping the cygwin version of smlnj.
534       Other x86 versions can use the existing binfiles.
535    
536    ----------------------------------------------------------------------
537    Name: Matthias Blume
538    Date: 2003/04/08 15:42:00 CDT
539    Tag: blume-20030408-listpair
540    Description:
541    
542    1. Added a target 'mlrisc' to installer.
543    
544    2. Added missing elements to structure ListPair.
545    
546    ----------------------------------------------------------------------
547    Name: Allen Leung
548    Date: 2003/01/07 10:40:00 EST
549    Tag: leunga-20030107-int-rem
550    Description:
551    
552    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
553    The arguments to the SUBL instruction were swapped.
554    
555    ----------------------------------------------------------------------
556    Name: Matthias Blume
557    Date: 2002/12/12 16:25:00 EST
558    Tag: blume-20021212-risc-ra
559    Description:
560    
561    Fixed a serious bug in the rewrite code for FP spilling/reloading that
562    sent the RA into an infinite loop when floating point registers get
563    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
564    110.37 and 110.38.)
565    There was another set of potential problems related to the handling of
566    MLRISC annotations (but those did not yet cause real problems, apparently).
567    
568    ----------------------------------------------------------------------
569    Name: Matthias Blume
570    Date: 2002/12/06 22:40:00 EST
571    Tag: blume-20021206-cm-fileid
572    Description:
573    
574    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
575    This fixes the problem of CM getting confused by files that suddenly
576    change their identity (e.g., by getting unlinked and recreated by some
577    text editor such as vi).  There might be a better/cheaper/cleaner way
578    of doing this, but for now this will have to do.
579    
580    ----------------------------------------------------------------------
581    Name: Matthias Blume
582    Date: 2002/10/28 09:50:00 EST
583    Tag: blume-20021028-typecheck
584    Description:
585    
586    Exported structure Typecheck from $smlnj/viscomp/core.cm.
587    
588    ----------------------------------------------------------------------
589    Name: Matthias Blume
590    Date: 2002/10/17 09:10:00 EDT
591    Tag: Release_110_42
592    Description:
593    
594    In good old tradition, there has been a slight hiccup so that we have
595    to patch 110.42 after the fact.   The old release tag has been replaced
596    (see below).
597    
598    The change solves a problem with two competing approaches the
599    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
600    each other's way.
601    
602    This change only affects the runtime system code and the installer script.
603    (No new bootfiles.)
604    
605    ----------------------------------------------------------------------
606    Name: Matthias Blume
607    Date: 2002/10/16 12:00:00 EDT
608    Tag: Release_110_42_removed
609    Description:
610    
611    New working release.  New bootfiles.
612    
613    ----------------------------------------------------------------------
614    Name: Matthias Blume
615    Date: 2002/10/10 13:10:00 EDT
616    Tag: blume-20021010-ppc-divs
617    Description:
618    
619    The mltree operator DIVS must be implemented with an overflow check on
620    the PPC because the hardware indicates divide-by-zero using "overflow" as
621    well.
622    
623    ----------------------------------------------------------------------
624    Name: Matthias Blume
625    Date: 2002/07/23 11:45:00
626    Tag: blume-20020723-smlnj-home
627    Description:
628    
629    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
630    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
631    CM_PATHCONFIG is also set), the path configuration file is assumed
632    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
633    the entire tree to some other place and everything will "just work".
634    
635    (Companion commands such as ml-build and ml-makedepend also sense this
636    variable.)
637    
638    ----------------------------------------------------------------------
639    Name: Matthias Blume
640    Date: 2002/07/12 21:19:00 EDT
641    Tag: blume-20020712-liveness
642    Description:
643    
644    Exported two useful "step" functions from liveness module (MLRISC).
645    
646    ----------------------------------------------------------------------
647    Name: Matthias Blume
648    Date: 2002/07/05 16:00 EDT
649    Tag: Release_110_41
650    Description:
651    
652    New working release.  New bootfiles.
653    
654    ----------------------------------------------------------------------
655    Name: Matthias Blume
656    Date: 2002/07/05 10:25:00 EDT
657    Tag: blume-20020705-btimp
658    Description:
659    
660    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
661    other clients can set up backtracing support.
662    
663    ----------------------------------------------------------------------
664    Name: Matthias Blume
665    Date: 2002/06/25 17:23:00 EDT
666    Tag: blume-20020625-fpmax
667    Description:
668    
669    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
670    case.  (The sense of the isNaN test was reversed -- which made min and
671    max always return their first argument.)
672    
673    ----------------------------------------------------------------------
674    Name: Matthias Blume
675    Date: 2002/06/11
676    Tag: blume-20020611-unixpath
677    Description:
678    
679    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
680    
681    ----------------------------------------------------------------------
682    Name: Matthias Blume
683    Date: 2002/06/10 16:35:00 EDT
684    Tag: blume-20020610-ieeereal
685    Description:
686    
687    I back-ported my implementation of IEEEReal.fromString from the
688    idlbasis-devel branch so that we can test it.
689    
690    Another small change is that ppDec tries to give more information
691    than just "<sig>" in the case of functors.  However, this code is
692    broken in some mysterious way if the functor's body's signature
693    has not been declared by ascription but gets inferred from the
694    implementation.  This needs fixing...
695    
696    ----------------------------------------------------------------------
697    Name: Matthias Blume
698    Date: 2002/05/31
699    Tag: blume-20020531-btrace-mode
700    Description:
701    
702    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
703    the wayside when I switched over to using Controls everywhere.)
704    
705    ----------------------------------------------------------------------
706    Name: Lal George
707    Date: 2002/05/23 12:21:40 EDT
708    Tag: george-20020523-visual-labels
709    Description:
710    
711    Labels are now displayed in the graphical output to make
712    the fall-through and target blocks obvious.
713    
714    ----------------------------------------------------------------------
715    Name: Matthias Blume
716    Date: 2002/05/22 11:03:00 EDT
717    Tag: blume-20020522-shrink
718    Description:
719    
720    John tweaked yesterday's fix for 1131 to handle an out-of-memory
721    situation that comes up when allocating huge arrays.
722    
723    ----------------------------------------------------------------------
724    Name: Matthias Blume
725    Date: 2002/05/21 16:00:00 EDT
726    Tag: Release_110_40
727    Description:
728    
729    New working release (110.40).  New bootfiles.
730    
731    [Also: John Reppy fixed GC bug 1131.]
732    
733    ----------------------------------------------------------------------
734    Name: Matthias Blume
735    Date: 2002/05/21 12:35:00 EDT
736    Tag: blume-20020521-cmdoc
737    Description:
738    
739    CM documentation update.
740    
741    ----------------------------------------------------------------------
742    Name: Matthias Blume
743    Date: 2002/05/21 10:55:00 EDT
744    Tag: blume-20020521-misc
745    Description:
746    
747    - John tweaked runtime to be silent on heap export (except when
748      GC messages are on).
749    - I added a few more things (cross-compiling versions of CMB) to
750      config/preloads (as suggestions).
751    
752    ----------------------------------------------------------------------
753    Name: Matthias Blume
754    Date: 2002/05/20 22:25:00 EDT
755    Tag: blume-20020520-controls
756    Description:
757    
758    - Added ControlUtil structure to control-lib.cm.
759    - Use it throughout.
760    - Used Controls facility to define MLRISC controls (as opposed to
761      registering MLRISC control ref cells with Controls after the fact)
762    - Fixed messed-up controls priorities.
763    
764    * Removed again all the stuff from config/preloads that one wouldn't
765      be able to preload at the time the initial heap image is built.
766      (Many libraries, e.g., CML, do not exist yet at this time.  The
767       only libraries that can be preloaded via config/preloads are those
768       that come bundled with the bootfiles.)
769    
770    ----------------------------------------------------------------------
771    Name: Matthias Blume
772    Date: 2002/05/20 10:59:00 EDT
773    Tag: blume-20020520-preloads
774    Description:
775    
776    Added a lot of commented-out suggestions for things to be included
777    in config/preloads.
778    
779    ----------------------------------------------------------------------
780    Name: Allen Leung
781    Date: 2002/05/18 14:20:00 EDT
782    Tag: leunga-20020518-mdl
783    Description:
784    
785    o Made the mdl tool stuff compile and run again.
786    o I've disabled all the stuff that depends on RTL specifications; they
787      are all badly broken anyway.
788    
789    ----------------------------------------------------------------------
790    Name: Matthias Blume
791    Date: 2002/05/17 16:49:00 EDT
792    Tag: blume-20020517-controls
793    Description:
794    
795    0. John Reppy made several modifications to the SML/NJ library.
796       In particular, there is a shiny new controls-lib.cm.
797    
798    1. Pushed new controls interface through compiler so that everything
799       compiles again.
800    
801    2. Added FormatComb and FORMAT_COMB to the CML version of the
802       SML/NJ library (so that CML compiles again).
803    
804    3. Modified init scripts because XXX_DEFAULT environment variables
805       are no longer with us.  (Boot-time initialization is now done
806       using the same environment variables that are also used for
807       startup-time initialization of controls.)
808    
809    ----------------------------------------------------------------------
810    Name: Lal George
811    Date: 2002/05/15 09:20:10 EDT
812    Tag: george-20020515-pseudo-op-decls
813    Description:
814    
815            All pseudo-ops emitted before the first segment declaration
816    such as TEXT, DATA, and BSS directives are assumed to be global
817    declarations and are emitted first in the assembly file. This is
818    useful in a number of situations where one has pseudo-ops that are not
819    specific to any segment, and also works around the constraint that one
820    cannot have client pseudo-ops in the TEXT segment.
821    
822            Because no segment is associated with these declarations it is
823    an error to allocate any space or objects before the first segment
824    directive and an exception will be raised. However, we cannot make
825    this check for client pseudo-ops.
826    
827            These top level declarations are a field in the CFG graph_info.
828    In theory you can continue to add to this field after the CFG has been
829    built -- provided you know what you are doing;-)
830    
831    ----------------------------------------------------------------------
832    Name: Matthias Blume
833    Date: 2002/05/13 16:40:00 EDT
834    Tag: blume-20020513-pp-etc
835    Description:
836    
837    A few minor bugfixes:
838    
839      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
840        (Bogus printouts for redefined bindings still occur.  Compiler
841        bug should no longer occur now.  We need to redo the prettyprinter
842        from scratch.)
843    
844      - CM pathname printer now also adds escape sequences for ( and )
845    
846      - commend and docu fixes for ml-nlffi
847    
848    ----------------------------------------------------------------------
849    Name: Matthias Blume
850    Date: 2002/05/10 16:40:00 EDT
851    Tag: blume-20020510-erg-textio
852    Description:
853    
854    Applied the following bugfix provided by Emden Gansner:
855    
856        Output is corrupted when outputSubstr is used rather than output.
857    
858        The problem occurs when a substring
859    
860            ss = (s, dataStart, dataLen)
861    
862        where dataStart > 0, fills a stream buffer with avail bytes left.
863        avail bytes of s, starting at index dataStart, are copied into the
864        buffer, the buffer is flushed, and then the remaining dataLen-avail
865        bytes of ss are copied into the beginning of the buffer. Instead of
866        starting this copy at index dataStart+avail in s, the current code
867        starts the copy at index avail.
868    
869        Fix:
870        In text-io-fn.sml, change line 695 from
871             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
872        to
873             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
874    
875    ----------------------------------------------------------------------
876    Name: Matthias Blume
877    Date: 2002/04/12 13:55:00 EDT
878    Tag: blume-20020412-assyntax
879    Description:
880    
881    1. Grabbed newer assyntax.h from the XFree86 project.
882    2. Fiddled with how to compile X86.prim.asm without warnings.
883    3. (Very) Minor cleanup in CM.
884    
885    ----------------------------------------------------------------------
886    Name: Matthias Blume
887    Date: 2002/04/01 (no joke!) 17:07:00 EST
888    Tag: blume-20020401-x86div
889    Description:
890    
891    Added full support for div/mod/rem/quot on the x86, using the machine
892    instruction's two results (without clumsily recomputing the remainder)
893    directly where appropriate.
894    
895    Some more extensive power-of-two support was added to the x86 instruction
896    selector (avoiding expensive divs, mods, and muls where they can be
897    replaced with cheaper shifts and masks).  However, this sort of thing
898    ought to be done earlier, e.g., within the CPS optimizer so that
899    all architectures benefit from it.
900    
901    The compiler compiles to a fixed point, but changes might be somewhat
902    fragile nevertheless.  Please, report any strange things that you might
903    see wrt. div/mod/quot/rem...
904    
905    ----------------------------------------------------------------------
906    Name: Matthias Blume
907    Date: 2002/03/29 17:22:00
908    Tag: blume-20020329-div
909    Description:
910    
911    Fixed my broken div/mod logic.  Unfortunately, this means that the
912    inline code for div/mod now has one more comparison than before.
913    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
914    The problem was with quotient = 0, because that alone does not tell
915    us which way the rounding went.  One then has to look at whether
916    remainder and divisor have the same sign...  :(
917    
918    Anyway, I replaced the bootfiles with fresh ones...
919    
920    ----------------------------------------------------------------------
921    Name: Matthias Blume
922    Date: 2002/03/29 14:10:00 EST
923    Tag: blume-20020329-inlprims
924    Description:
925    
926    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
927    
928    Primops have changed. This means that the bin/boot-file formats have
929    changed as well.
930    
931    To make sure that there is no confusion, I made a new version.
932    
933    
934    CHANGES:
935    
936    * removed REMT from mltree (remainder should never overflow).
937    
938    * added primops to deal with divisions of all flavors to the frontend
939    
940    * handled these primops all the way through so they map to their respective
941      MLRISC support
942    
943    * used these primops in the implementation of Int, Int32, Word, Word32
944    
945    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
946    
947    * parameterized INLMIN, INLMAX, and INLABS by a numkind
948    
949    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
950      floating point
951    
952    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
953      and Real (but Real.abs maps to a separate floating-point-only primop)
954    
955    
956    TODO items:
957    
958    * Hacked Alpha32 instruction selection, disabling the selection of REMx
959      instructions because the machine instruction encoder cannot handle
960      them.  (Hppa, PPC, and Sparc instruction selection did not handle
961      REM in the first place, and REM is supported by the x86 machine coder.)
962    
963    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
964      selection phase.  (The two can be streamlined because the hardware
965      delivers both quotient and remainder at the same time anyway.)
966    
967    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
968      (Currently the behavior is inconsistent both across architectures and
969      wrt. the draft Basis spec.)
970    
971    * Word8 should eventually be handled natively, too.
972    
973    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
974      as if there currently is no execution path that could trigger it in
975      SML/NJ.  (The assumptions underlying functions arith and promotable do not
976      hold for things like multiplication and division.)
977    
978    ----------------------------------------------------------------------
979    Name: Matthias Blume
980    Date: 2002/03/27 16:27:00 EST
981    Tag: blume-20020327-mlrisc-divisions
982    Description:
983    
984    Added support for all four division operations (ML's div, mod, quot,
985    and rem) to MLRISC.  In the course of doing so, I also rationalized
986    the naming (no more annoying switch-around of DIV and QUOT), by
987    parameterizing the operation by div_rounding_mode (which can be either
988    DIV_TO_ZERO or DIV_TO_NEGINF).
989    
990    The generic MLTreeGen functor takes care of compiling all four
991    operations down to only round-to-zero div.
992    
993    Missing pieces:
994    
995      * Doing something smarter than relying on MLTreeGen on architectures
996        like, e.g., the x86 where hardware division delivers both quotient and
997        remainder at the same time.  With this, the implementation of the
998        round-to-neginf operations could be further streamlined.
999    
1000      * Remove inlining support for div/mod/rem from the frontend and replace it
1001        with primops that get carried through to the backend.  Do this for all
1002        int and word types.
1003    
1004    ----------------------------------------------------------------------
1005    Name: Matthias Blume
1006    Date: 2002/03/25 17:25:00 EST
1007    Tag: blume-20020325-divmod
1008    Description:
1009    
1010    I improved (hopefully without breaking them) the implementation of Int.div,
1011    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1012    advantage of the following observations:
1013    
1014      Let  q = x quot y      r = x rem y
1015           d = x div  y      m = x mod y
1016    
1017    where "quot" is the round-to-zero version of integer division that
1018    hardware usually provides.  Then we have:
1019    
1020         r = x - q * y        where neither the * nor the - will overflow
1021         d = if q >= 0 orelse x = q * y then q else q - 1
1022                              where neither the * nor the - will overflow
1023         m = if q >= 0 orelse r = 0 then r else r + y
1024                              where the + will not overflow
1025    
1026    This results in substantial simplification of the generated code.
1027    The following table shows the number of CFG nodes and edges generated
1028    for
1029            fun f (x, y) = x OPER y
1030            (* with OPER \in div, mod, quot, rem *)
1031    
1032    
1033        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1034        --------------------------------------------------------
1035         div |         24 |         39 |         12 |         16
1036         mod |         41 |         71 |         12 |         16
1037        quot |          8 |         10 |          8 |         10
1038         rem |         10 |         14 |          8 |         10
1039    
1040    
1041    ----------------------------------------------------------------------
1042    Name: Matthias Blume
1043    Date: 2002/03/25 22:06:00 EST
1044    Tag: blume-20020325-cprotobug
1045    Description:
1046    
1047    Fixed a bug in cproto (c prototype decoder).
1048    
1049    ----------------------------------------------------------------------
1050    Name: Matthias Blume
1051    Date: 2002/03/25 16:00:00 EST
1052    Tag: blume-20020325-raw-primops
1053    Description:
1054    
1055    I did some cleanup to Allen's new primop code and
1056    replaced yesterday's bootfiles with new ones.
1057    (But they are stored in the same place.)
1058    
1059    ----------------------------------------------------------------------
1060    Name: Matthias Blume
1061    Date: 2002/03/24 22:40:00 EST
1062    Tag: blume-20020324-bootfiles
1063    Description:
1064    
1065    Made the bootfiles that Allen asked for.
1066    
1067    ----------------------------------------------------------------------
1068    Name: Allen Leung
1069    Date: 2002/03/23 15:50:00 EST
1070    Tag: leunga-20020323-flint-cps-rcc-primops
1071    Description:
1072    
1073      1. Changes to FLINT primops:
1074    
1075        (* make a call to a C-function;
1076         * The primop carries C function prototype information and specifies
1077         * which of its (ML-) arguments are floating point. C prototype
1078         * information is for use by the backend, ML information is for
1079         * use by the CPS converter. *)
1080      | RAW_CCALL of { c_proto: CTypes.c_proto,
1081                       ml_args: ccall_type list,
1082                       ml_res_opt: ccall_type option,
1083                       reentrant : bool
1084                     } option
1085       (* Allocate uninitialized storage on the heap.
1086        * The record is meant to hold short-lived C objects, i.e., they
1087        * are not ML pointers.  With the tag, the representation is
1088        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1089        *)
1090      | RAW_RECORD of {tag:bool,sz:int}
1091      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1092    
1093      2.  These CPS primops are now overloaded:
1094    
1095           rawload of {kind:numkind}
1096           rawstore of {kind:numkind}
1097    
1098          The one argument form is:
1099    
1100             rawload {kind} address
1101    
1102          The two argument form is:
1103    
1104             rawload {kind} [ml object, byte-offset]
1105    
1106      3. RAW_CCALL/RCC now takes two extra arguments:
1107    
1108         a. The first is whether the C call is reentrant, i.e., whether
1109            ML state should be saved and restored.
1110         b. The second argument is a string argument specifying the name of
1111            library and the C function.
1112    
1113         These things are currently not handled in the code generator, yet.
1114    
1115      4. In CProto,
1116    
1117         An encoding type of "bool" means "ml object" and is mapped into
1118         C prototype of PTR.  Note that "bool" is different than "string",
1119         even though "string" is also mapped into PTR, because "bool"
1120         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1121    
1122      5. Pickler/unpicker
1123    
1124         Changed to handle RAW_RECORD and newest RAW_CCALL
1125    
1126      6. MLRiscGen,
1127    
1128         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1129         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1130            in the file CodeGen/cpscompile/cps-c-calls.sml
1131    
1132      7. Added the conditional move operator
1133    
1134             condmove of branch
1135    
1136         to cps.  Generation of this is still buggy so it is currently
1137         disabled.
1138    
1139    ----------------------------------------------------------------------
1140    Name: Lal George
1141    Date: 2002/03/22 14:18:25 EST
1142    Tag: george-20020322-cps-branch-prob
1143    Description:
1144    
1145    Implemented the Ball-Larus branch prediction-heuristics, and
1146    incorporated graphical viewers for control flow graphs.
1147    
1148    Ball-Larus Heuristics:
1149    ---------------------
1150    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1151    
1152    By design it uses the Dempster-Shafer theory for combining
1153    probabilities.  For example, in the function:
1154    
1155        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1156    
1157    the ball-larus heuristics predicts that the n=0 is unlikely
1158    (OH-heuristic), and the 'then' branch is unlikely because of the
1159    RH-heuristic -- giving the 'then' branch an even lower combined
1160    probability using the Dempster-Shafer theory.
1161    
1162    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1163    probability of the 'then' branch because of the loop in the else
1164    branch.
1165    
1166    
1167    Graphical Viewing:
1168    ------------------
1169    I merely plugged in Allen's graphical viewers into the compiler. The
1170    additional code is not much. At the top level, saying:
1171    
1172            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1173    
1174    will display the graphical view of the control flow graph just before
1175    back-patching.  daVinci must be in your path for this to work. If
1176    daVinci is not available, then the default viewer can be changed
1177    using:
1178    
1179            Control.MLRISC.getString "viewer"
1180    
1181    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1182    course, these viewers must be in your path.
1183    
1184    The above will display the compilation unit at the level of clusters,
1185    many of which are small, boring, and un-interesting. Also setting:
1186    
1187            Control.MLRISC.getInt "cfg-graphical-view_size"
1188    
1189    will display clusters that are larger than the value set by the above.
1190    
1191    
1192    ----------------------------------------------------------------------
1193    Name: Matthias Blume
1194    Date: 2002/03/21 22:20:00 EST
1195    Tag: blume-20020321-kmp-bugfix
1196    Description:
1197    
1198    Changed the interface to the KMP routine in PreString and fixed
1199    a minor bug in one place where it was used.
1200    
1201    ----------------------------------------------------------------------
1202    Name: Allen Leung
1203    Date: 2002/03/21 20:30:00 EST
1204    Tag: leunga-20020321-cfg
1205    Description:
1206    
1207      Fixed a potential problem in cfg edge splitting.
1208    
1209    ----------------------------------------------------------------------
1210    Name: Allen Leung
1211    Date: 2002/03/21 17:15:00 EST
1212    Tag: leunga-20020321-x86-fp-cfg
1213    Description:
1214    
1215      1. Recoded the buggy parts of x86-fp.
1216    
1217         a. All the block reordering code has been removed.
1218            We now depend on the block placement phases to do this work.
1219    
1220         b. Critical edge splitting code has been simplified and moved into the
1221            CFG modules, as where they belong.
1222    
1223         Both of these were quite buggy and complex.  The code is now much, much
1224         simpler.
1225    
1226      2. X86 backend.
1227    
1228         a. Added instructions for 64-bit support.  Instruction selection for
1229            64-bit has not been committed, however, since that
1230            requires changes to MLTREE which haven't been approved by
1231            Lal and John.
1232    
1233         b. Added support for FUCOMI and FUCOMIP when generating code for
1234            PentiumPro and above.  We only generate these instructions in
1235            the fast-fp mode.
1236    
1237         c. Added cases for JP and JNP in X86FreqProps.
1238    
1239      3. CFG
1240    
1241         CFG now has a bunch of methods for edge splitting and merging.
1242    
1243      4. Machine description.
1244    
1245         John's simplification of MLTREE_BASIS.fcond broke a few machine
1246         description things:
1247    
1248         rtl-build.{sig,sml} and hppa.mdl fixed.
1249    
1250         NOTE: the machine description stuff in the repository is still broken.
1251               Again, I can't put my fixes in because that involves
1252               changes to MLTREE.
1253    
1254    ----------------------------------------------------------------------
1255    Name: Matthias Blume
1256    Date: 2002/03/20 15:55:00 EST
1257    Tag: blume-20020320-kmp
1258    Description:
1259    
1260    Implemented Knuth-Morris-Pratt string matching in PreString and used
1261    it for String.isSubstring, Substring.isSubstring, and
1262    Substring.position.
1263    
1264    (Might need some stress-testing.  Simple examples worked fine.)
1265    
1266    ----------------------------------------------------------------------
1267    Name: Matthias Blume
1268    Date: 2002/03/19 16:37:00 EST
1269    Tag: blume-20020319-witnesses
1270    Description:
1271    
1272    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1273    
1274    This implements a generic mechanism for changing constness qualifiers
1275    anywhere within big C types without resorting to outright "casts".
1276    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1277    modify the constness at the outermost level.)
1278    The implementation of "convert" is based on the idea of "witness"
1279    values -- values that are not used by the operation but whose types
1280    "testify" to their applicability.  On the implementation side, "convert"
1281    is simply a projection (returning its second curried argument).  With
1282    cross-module inlining, it should not result in any machine code being
1283    generated.
1284    
1285    ----------------------------------------------------------------------
1286    Name: Matthias Blume
1287    Date: 2002/03/15 16:40:00 EST
1288    Tag: blume-20020315-basis
1289    Description:
1290    
1291    Provided (preliminary?) implementations for
1292    
1293      {String,Substring}.{concatWith,isSuffix,isSubstring}
1294    
1295    and
1296    
1297      Substring.full
1298    
1299    Those are in the Basis spec but they were missing in SML/NJ.
1300    
1301    ----------------------------------------------------------------------
1302    Name: Matthias Blume
1303    Date: 2002/03/14 21:30:00 EST
1304    Tag: blume-20020314-controls
1305    Description:
1306    
1307    Controls:
1308    ---------
1309    
1310    1. Factored out the recently-added Controls : CONTROLS stuff and put
1311       it into its own library $/controls-lib.cm.  The source tree for
1312       this is under src/smlnj-lib/Controls.
1313    
1314    2. Changed the names of types and functions in this interface, so they
1315       make a bit more "sense":
1316    
1317          module -> registry
1318          'a registry -> 'a group
1319    
1320    3. The interface now deals in ref cells only.  The getter/setter interface
1321       is (mostly) gone.
1322    
1323    4. Added a function that lets one register an already-existing ref cell.
1324    
1325    5. Made the corresponding modifications to the rest of the code so that
1326       everything compiles again.
1327    
1328    6. Changed the implementation of Controls.MLRISC back to something closer
1329       to the original.  In particular, this module (and therefore MLRISC)
1330       does not depend on Controls.  There now is some link-time code in
1331       int-sys.sml that registers the MLRISC controls with the Controls
1332       module.
1333    
1334    CM:
1335    ---
1336    
1337      * One can now specify the lambda-split aggressiveness in init.cmi.
1338    
1339    ----------------------------------------------------------------------
1340    Name: Allen Leung
1341    Date: 2002/03/13 17:30:00 EST
1342    Tag: leunga-20020313-x86-fp-unary
1343    Description:
1344    
1345    Bug fix for:
1346    
1347    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1348    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1349    > - fun f(x,(y,z)) = Real.~ y;
1350    > [autoloading]
1351    > [autoloading done]
1352    >       fchsl   (%eax), 184(%esp)
1353    > Error: MLRisc bug: X86MCEmitter.emitInstr
1354    >
1355    > uncaught exception Error
1356    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1357    
1358    The problem was that the code generator did not generate any fp registers
1359    in this case, and the ra didn't know that it needed to run the X86FP phase to
1360    translate the pseudo fp instruction.   This only happened with unary fp
1361    operators in certain situations.
1362    
1363    ----------------------------------------------------------------------
1364    Name: Matthias Blume
1365    Date: 2002/03/13 14:00:00 EST
1366    Tag: blume-20020313-overload-etc
1367    Description:
1368    
1369    1. Added _overload as a synonym for overload for backward compatibility.
1370       (Control.overloadKW must be true for either version to be accepted.)
1371    
1372    2. Fixed bug in install script that caused more things to be installed
1373       than what was requested in config/targets.
1374    
1375    3. Made CM aware of the (_)overload construct so that autoloading
1376       works.
1377    
1378    ----------------------------------------------------------------------
1379    Name: Matthias Blume
1380    Date: 2002/03/12 22:03:00 EST
1381    Tag: blume-20020312-url
1382    Description:
1383    
1384    Forgot to update BOOT and srcarchiveurl.
1385    
1386    ----------------------------------------------------------------------
1387    Name: Matthias Blume
1388    Date: 2002/03/12 17:30:00 EST
1389    Tag: blume-20020312-version110392
1390    Description:
1391    
1392    Yet another version number bump (because of small changes to the
1393    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1394    
1395    Changes:
1396    
1397      The new pid generation scheme described a few weeks ago was overly
1398      complicated.  I implemented a new mechanism that is simpler and
1399      provides a bit more "stability":  Once CM has seen a compilation
1400      unit, it keeps its identity constant (as long as you do not delete
1401      those crucial CM/GUID/* files).  This means that when you change
1402      an interface, compile, then go back to the old interface, and
1403      compile again, you arrive at the original pid.
1404    
1405      There now also is a mechanism that instructs CM to use the plain
1406      environment hash as a module's pid (effectively making its GUID
1407      the empty string).  For this, "noguid" must be specified as an
1408      option to the .sml file in question within its .cm file.
1409      This is most useful for code that is being generated by tools such
1410      as ml-nlffigen (because during development programmers tend to
1411      erase the tool's entire output directory tree including CM's cached
1412      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1413      revert to the old, broken behavior of SML/NJ, but in specific cases
1414      where there is no danger of interface confusion, its use is ok
1415      (I think).
1416    
1417      ml-nlffigen by default generates "noguid" annotations.  They can be
1418      turned off by specifying -guid in its command line.
1419    
1420    ----------------------------------------------------------------------
1421    Name: Lal George
1422    Date: 2002/03/12 12 14:42:36 EST
1423    Tag: george-20020312-frequency-computation
1424    Description:
1425    
1426    Integrated jump chaining and static block frequency into the
1427    compiler. More details and numbers later.
1428    
1429    ----------------------------------------------------------------------
1430    Name: Lal George
1431    Date: 2002/03/11 11 22:38:53 EST
1432    Tag: george-20020311-jump-chain-elim
1433    Description:
1434    
1435    Tested the jump chain elimination on all architectures (except the
1436    hppa).  This is on by default right now and is profitable for the
1437    alpha and x86, however, it may not be profitable for the sparc and ppc
1438    when compiling the compiler.
1439    
1440    The gc test will typically jump to a label at the end of the cluster,
1441    where there is another jump to an external cluster containing the actual
1442    code to invoke gc. This is to allow factoring of common gc invocation
1443    sequences. That is to say, we generate:
1444    
1445            f:
1446               testgc
1447               ja   L1      % jump if above to L1
1448    
1449            L1:
1450               jmp L2
1451    
1452    
1453    After jump chain elimination the 'ja L1' instructions is converted to
1454    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1455    up being implemented in their long form (if L2 is far away) using:
1456    
1457            jbe     L3      % jump if below or equal to L3
1458            jmp     L2
1459         L3:
1460            ...
1461    
1462    
1463    For large compilation units L2  may be far away.
1464    
1465    
1466    ----------------------------------------------------------------------
1467    Name: Matthias Blume
1468    Date: 2002/03/11 13:30:00 EST
1469    Tag: blume-20020311-mltreeeval
1470    Description:
1471    
1472    A functor parameter was missing.
1473    
1474    ----------------------------------------------------------------------
1475    Name: Allen Leung
1476    Date: 2002/03/11 10:30:00 EST
1477    Tag: leunga-20020311-runtime-string0
1478    Description:
1479    
1480       The representation of the empty string now points to a
1481    legal null terminated C string instead of unit.  It is now possible
1482    to convert an ML string into C string with InlineT.CharVector.getData.
1483    This compiles into one single machine instruction.
1484    
1485    ----------------------------------------------------------------------
1486    Name: Allen Leung
1487    Date: 2002/03/10 23:55:00 EST
1488    Tag: leunga-20020310-x86-call
1489    Description:
1490    
1491       Added machine generation for CALL instruction (relative displacement mode)
1492    
1493    ----------------------------------------------------------------------
1494    Name: Matthias Blume
1495    Date: 2002/03/08 16:05:00
1496    Tag: blume-20020308-entrypoints
1497    Description:
1498    
1499    Version number bumped to 110.39.1.  NEW BOOTFILES!
1500    
1501    Entrypoints: non-zero offset into a code object where execution should begin.
1502    
1503    - Added the notion of an entrypoint to CodeObj.
1504    - Added reading/writing of entrypoint info to Binfile.
1505    - Made runtime system bootloader aware of entrypoints.
1506    - Use the address of the label of the first function given to mlriscGen
1507      as the entrypoint.  This address is currently always 0, but it will
1508      not be 0 once we turn on block placement.
1509    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
1510      with entry points) from mlriscGen.
1511    
1512    ----------------------------------------------------------------------
1513    Name: Allen Leung
1514    Date: 2002/03/07 20:45:00 EST
1515    Tag: leunga-20020307-x86-cmov
1516    Description:
1517    
1518       Bug fixes for CMOVcc on x86.
1519    
1520       1. Added machine code generation for CMOVcc
1521       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1522       3. CMOVcc cannot have an immediate operand as argument.
1523    
1524    ----------------------------------------------------------------------
1525    Name: Matthias Blume
1526    Date: 2002/03/07 16:15:00 EST
1527    Tag: blume-20020307-controls
1528    Description:
1529    
1530    This is a very large but mostly boring patch which makes (almost)
1531    every tuneable compiler knob (i.e., pretty much everything under
1532    Control.* plus a few other things) configurable via both the command
1533    line and environment variables in the style CM did its configuration
1534    until now.
1535    
1536    Try starting sml with '-h' (or, if you are brave, '-H')
1537    
1538    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1539    implements the underlying generic mechanism.
1540    
1541    The interface to some of the existing such facilities has changed somewhat.
1542    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1543    (The getFoo interface is still there for backward-compatibility, but its
1544    use is deprecated.)
1545    
1546    The ml-build script passes -Cxxx=yyy command-line arguments through so
1547    that one can now twiddle the compiler settings when using this "batch"
1548    compiler.
1549    
1550    TODO items:
1551    
1552    We should go through and throw out all controls that are no longer
1553    connected to anything.  Moreover, we should go through and provide
1554    meaningful (and correct!) documentation strings for those controls
1555    that still are connected.
1556    
1557    Currently, multiple calls to Controls.new are accepted (only the first
1558    has any effect).  Eventually we should make sure that every control
1559    is being made (via Controls.new) exactly once.  Future access can then
1560    be done using Controls.acc.
1561    
1562    Finally, it would probably be a good idea to use the getter-setter
1563    interface to controls rather than ref cells.  For the time being, both
1564    styles are provided by the Controls module, but getter-setter pairs are
1565    better if thread-safety is of any concern because they can be wrapped.
1566    
1567    *****************************************
1568    
1569    One bug fix: The function blockPlacement in three of the MLRISC
1570    backpatch files used to be hard-wired to one of two possibilities at
1571    link time (according to the value of the placementFlag).  But (I
1572    think) it should rather sense the flag every time.
1573    
1574    *****************************************
1575    
1576    Other assorted changes (by other people who did not supply a HISTORY entry):
1577    
1578    1. the cross-module inliner now works much better (Monnier)
1579    2. representation of weights, frequencies, and probabilities in MLRISC
1580       changed in preparation of using those for weighted block placement
1581       (Reppy, George)
1582    
1583    ----------------------------------------------------------------------
1584    Name: Lal George
1585    Date: 2002/03/07 14:44:24 EST 2002
1586    Tag: george-20020307-weighted-block-placement
1587    
1588    Tested the weighted block placement optimization on all architectures
1589    (except the hppa) using AMPL to generate the block and edge frequencies.
1590    Changes were required in the machine properties to correctly
1591    categorize trap instructions. There is an MLRISC flag
1592    "weighted-block-placement" that can be used to enable weighted block
1593    placement, but this will be ineffective without block/edge
1594    frequencies (coming soon).
1595    
1596    
1597    ----------------------------------------------------------------------
1598    Name: Lal George
1599    Date: 2002/03/05 17:24:48 EST
1600    Tag: george-20020305-linkage-cluster
1601    
1602    In order to support the block placement optimization, a new cluster
1603    is generated as the very first cluster (called the linkage cluster).
1604    It contains a single jump to the 'real' entry point for the compilation
1605    unit. Block placement has no effect on the linkage cluster itself, but
1606    all the other clusters  have full freedom in the manner in which they
1607    reorder blocks or functions.
1608    
1609    On the x86 the typical linkage code that is generated is:
1610       ----------------------
1611            .align 2
1612       L0:
1613            addl    $L1-L0, 72(%esp)
1614            jmp     L1
1615    
1616    
1617            .align  2
1618       L1:
1619       ----------------------
1620    
1621    72(%esp) is the memory location for the stdlink register. This
1622    must contain the address of the CPS function being called. In the
1623    above example, it contains the address of  L0; before
1624    calling L1 (the real entry point for the compilation unit), it
1625    must contain the address for L1, and hence
1626    
1627            addl $L1-L0, 72(%esp)
1628    
1629    I have tested this on all architectures except the hppa.The increase
1630    in code size is of course negligible
1631    
1632    ----------------------------------------------------------------------
1633    Name: Allen Leung
1634    Date: 2002/03/03 13:20:00 EST
1635    Tag: leunga-20020303-mlrisc-tools
1636    
1637      Added #[ ... ] expressions to mlrisc tools
1638    
1639    ----------------------------------------------------------------------
1640    Name: Matthias Blume
1641    Date: 2002/02/27 12:29:00 EST
1642    Tag: blume-20020227-cdebug
1643    Description:
1644    
1645    - made types in structure C and C_Debug to be equal
1646    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1647    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1648    
1649    ----------------------------------------------------------------------
1650    Name: Matthias Blume
1651    Date: 2002/02/26 12:00:00 EST
1652    Tag: blume-20020226-ffi
1653    Description:
1654    
1655    1. Fixed a minor bug in CM's "noweb" tool:
1656       If numbering is turned off, then truly don't number (i.e., do not
1657       supply the -L option to noweb).  The previous behavior was to supply
1658       -L'' -- which caused noweb to use the "default" line numbering scheme.
1659       Thanks to Chris Richards for pointing this out (and supplying the fix).
1660    
1661    2. Once again, I reworked some aspects of the FFI:
1662    
1663       A. The incomplete/complete type business:
1664    
1665       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
1666         gone!
1667       - ML types representing an incomplete type are now *equal* to
1668         ML types representing their corresponding complete types (just like
1669         in C).  This is still safe because ml-nlffigen will not generate
1670         RTTI for incomplete types, nor will it generate functions that
1671         require access to such RTTI.   But when ML code generated from both
1672         incomplete and complete versions of the C type meet, the ML types
1673         are trivially interoperable.
1674    
1675         NOTE:  These changes restore the full generality of the translation
1676         (which was previously lost when I eliminated functorization)!
1677    
1678       B. Enum types:
1679    
1680       - Structure C now has a type constructor "enum" that is similar to
1681         how the "su" constructor works.  However, "enum" is not a phantom
1682         type because each "T enum" has values (and is isomorphic to
1683         MLRep.Signed.int).
1684       - There are generic access operations for enum objects (using
1685         MLRep.Signed.int).
1686       - ml-nlffigen will generate a structure E_foo for each "enum foo".
1687         * The structure contains the definition of type "mlrep" (the ML-side
1688         representation type of the enum).  Normally, mlrep is the same
1689         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
1690         then mlrep will be defined as a datatype -- thus facilitating
1691         pattern matching on mlrep values.
1692         ("-ec" will be suppressed if there are duplicate values in an
1693          enumeration.)
1694         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
1695         will be generated for each C enum constant xxx.
1696         * Conversion functions m2i and i2m convert between mlrep and
1697         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
1698         * Coversion functions c and ml convert between mlrep and "tag enum".
1699         * Access functions (get/set) fetch and store mlrep values.
1700       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
1701         enumerations are merged into one single enumeration represented by
1702         structure E_'.
1703    
1704    ----------------------------------------------------------------------
1705    Name: Allen Leung
1706    Date: 2002/02/25 04:45:00 EST
1707    Tag: leunga-20020225-cps-spill
1708    
1709    This is a new implementation of the CPS spill phase.
1710    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
1711    In case of problems, replace it with the old file spill.sml
1712    
1713    The current compiler runs into some serious performance problems when
1714    constructing a large record.  This can happen when we try to compile a
1715    structure with many items.  Even a very simple structure like the following
1716    makes the compiler slow down.
1717    
1718        structure Foo = struct
1719           val x_1 = 0w1 : Word32.int
1720           val x_2 = 0w2 : Word32.int
1721           val x_3 = 0w3 : Word32.int
1722           ...
1723           val x_N = 0wN : Word32.int
1724        end
1725    
1726    The following table shows the compile time, from N=1000 to N=4000,
1727    with the old compiler:
1728    
1729    N
1730    1000   CPS 100 spill                           0.04u  0.00s  0.00g
1731           MLRISC ra                               0.06u  0.00s  0.05g
1732              (spills = 0 reloads = 0)
1733           TOTAL                                   0.63u  0.07s  0.21g
1734    
1735    1100   CPS 100 spill                           8.25u  0.32s  0.64g
1736           MLRISC ra                               5.68u  0.59s  3.93g
1737              (spills = 0 reloads = 0)
1738           TOTAL                                   14.71u  0.99s  4.81g
1739    
1740    1500   CPS 100 spill                           58.55u  2.34s  1.74g
1741           MLRISC ra                               5.54u  0.65s  3.91g
1742              (spills = 543 reloads = 1082)
1743           TOTAL                                   65.40u  3.13s  6.00g
1744    
1745    2000   CPS 100 spill                           126.69u  4.84s  3.08g
1746           MLRISC ra                               0.80u  0.10s  0.55g
1747              (spills = 42 reloads = 84)
1748           TOTAL                                   129.42u  5.10s  4.13g
1749    
1750    3000   CPS 100 spill                           675.59u  19.03s  11.64g
1751           MLRISC ra                               2.69u  0.27s  1.38g
1752              (spills = 62 reloads = 124)
1753           TOTAL                                   682.48u  19.61s  13.99g
1754    
1755    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
1756           MLRISC ra                               4.96u  0.27s  2.72g
1757              (spills = 85 reloads = 170)
1758           TOTAL                                   2375.26u  57.21s  48.00g
1759    
1760    As you can see the old cps spill module suffers from some serious
1761    performance problem.  But since I cannot decipher the old code fully,
1762    instead of patching the problems up, I'm reimplementing it
1763    with a different algorithm.  The new code is more modular,
1764    smaller when compiled, and substantially faster
1765    (O(n log n) time and O(n) space).  Timing of the new spill module:
1766    
1767    4000  CPS 100 spill                           0.02u  0.00s  0.00g
1768          MLRISC ra                               0.25u  0.02s  0.15g
1769             (spills=1 reloads=3)
1770          TOTAL                                   7.74u  0.34s  1.62g
1771    
1772    Implementation details:
1773    
1774    As far as I can tell, the purpose of the CPS spill module is to make sure the
1775    number of live variables at any program point (the bandwidth)
1776    does not exceed a certain limit, which is determined by the
1777    size of the spill area.
1778    
1779    When the bandwidth is too large, we decrease the register pressure by
1780    packing live variables into spill records.  How we achieve this is
1781    completely different than what we did in the old code.
1782    
1783    First, there is something about the MLRiscGen code generator
1784    that we should be aware of:
1785    
1786    o MLRiscGen performs code motion!
1787    
1788       In particular, it will move floating point computations and
1789       address computations involving only the heap pointer to
1790       their use sites (if there is only a single use).
1791       What this means is that if we have a CPS record construction
1792       statement
1793    
1794           RECORD(k,vl,w,e)
1795    
1796       we should never count the new record address w as live if w
1797       has only one use (which is often the case).
1798    
1799       We should do something similar to floating point, but the transformation
1800       there is much more complex, so I won't deal with that.
1801    
1802    Secondly, there are now two new cps primops at our disposal:
1803    
1804     1. rawrecord of record_kind option
1805        This pure operator allocates some uninitialized storage from the heap.
1806        There are two forms:
1807    
1808         rawrecord NONE [INT n]  allocates a tagless record of length n
1809         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
1810                                     and initializes the tag.
1811    
1812     2. rawupdate of cty
1813          rawupdate cty (v,i,x)
1814          Assigns to x to the ith component of record v.
1815          The storelist is not updated.
1816    
1817    We use these new primops for both spilling and increment record construction.
1818    
1819     1. Spilling.
1820    
1821        This is implemented with a linear scan algorithm (but generalized
1822        to trees).  The algorithm will create a single spill record at the
1823        beginning of the cps function and use rawupdate to spill to it,
1824        and SELECT or SELp to reload from it.  So both spills and reloads
1825        are fine-grain operations.  In contrast, in the old algorithm
1826        "spills" have to be bundled together in records.
1827    
1828        Ideally, we should sink the spill record construction to where
1829        it is needed.  We can even split the spill record into multiple ones
1830        at the places where they are needed.  But CPS is not a good
1831        representation for global code motion, so I'll keep it simple and
1832        am not attempting this.
1833    
1834     2. Incremental record construction (aka record splitting).
1835    
1836        Long records with many component values which are simulatenously live
1837        (recall that single use record addresses are not considered to
1838         be live) are constructed with rawrecord and rawupdate.
1839        We allocate space on the heap with rawrecord first, then gradually
1840        fill it in with rawupdate.  This is the technique suggested to me
1841        by Matthias.
1842    
1843        Some restrictions on when this is applicable:
1844        1. It is not a VECTOR record.  The code generator currently does not handle
1845           this case. VECTOR record uses double indirection like arrays.
1846        2. All the record component values are defined in the same "basic block"
1847           as the record constructor.  This is to prevent speculative
1848           record construction.
1849    
1850    ----------------------------------------------------------------------
1851    Name: Allen Leung
1852    Date: 2002/02/22 01:02:00 EST
1853    Tag: leunga-20020222-mlrisc-tools
1854    
1855    Minor bug fixes in the parser and rewriter
1856    
1857    ----------------------------------------------------------------------
1858    Name: Allen Leung
1859    Date: 2002/02/21 20:20:00 EST
1860    Tag: leunga-20020221-peephole
1861    
1862    Regenerated the peephole files.  Some contained typos in the specification
1863    and some didn't compile because of pretty printing bugs in the old version
1864    of 'nowhere'.
1865    
1866    ----------------------------------------------------------------------
1867    Name: Allen Leung
1868    Date: 2002/02/19 20:20:00 EST
1869    Tag: leunga-20020219-mlrisc-tools
1870    Description:
1871    
1872       Minor bug fixes to the mlrisc-tools library:
1873    
1874       1.  Fixed up parsing colon suffixed keywords
1875       2.  Added the ability to shut the error messages up
1876       3.  Reimplemented the pretty printer and fixed up/improved
1877           the pretty printing of handle and -> types.
1878       4.  Fixed up generation of literal symbols in the nowhere tool.
1879       5.  Added some SML keywords to to sml.sty
1880    
1881    ----------------------------------------------------------------------
1882    Name: Matthias Blume
1883    Date: 2002/02/19 16:20:00 EST
1884    Tag: blume-20020219-cmffi
1885    Description:
1886    
1887    A wild mix of changes, some minor, some major:
1888    
1889    * All C FFI-related libraries are now anchored under $c:
1890        $/c.cm      --> $c/c.cm
1891        $/c-int.cm  --> $c/internals/c-int.cm
1892        $/memory.cm --> $c/memory/memory.cm
1893    
1894    * "make" tool (in CM) now treats its argument pathname slightly
1895      differently:
1896        1. If the native expansion is an absolute name, then before invoking
1897           the "make" command on it, CM will apply OS.Path.mkRelative
1898           (with relativeTo = OS.FileSys.getDir()) to it.
1899        2. The argument will be passed through to subsequent phases of CM
1900           processing without "going native".  In particular, if the argument
1901           was an anchored path, then "make" will not lose track of that anchor.
1902    
1903    * Compiler backends now "know" their respective C calling conventions
1904      instead of having to be told about it by ml-nlffigen.  This relieves
1905      ml-nlffigen from one of its burdens.
1906    
1907    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
1908    
1909    * Export C_DEBUG and C_Debug from $c/c.cm.
1910    
1911    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
1912      subtyping relationship between incomplete pointers and their complete
1913      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
1914      with the convention of instantiating 'o with ('t, 'c) obj whenever
1915      the pointer target type is complete.  In the incomplete case, 'o
1916      will be instantiated with some "'c iobj" -- a type obtained by
1917      using one of the functors PointerToIncompleteType or PointerToCompleteType.
1918    
1919      Operations that work on both incomplete and complete pointer types are
1920      typed as taking an 'o ptr while operations that require the target to
1921      be known are typed as taking some ('t, 'c) obj ptr.
1922    
1923      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
1924      where void is an eqtype without any values.  This makes it possible
1925      to work on voidptr values using functions meant to operate on light
1926      incomplete pointers.
1927    
1928    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
1929      been vastly simplified.
1930    
1931    ----------------------------------------------------------------------
1932    Name: Matthias Blume
1933    Date: 2002/02/19 10:48:00 EST
1934    Tag: blume-20020219-pqfix
1935    Description:
1936    
1937    Applied Chris Okasaki's bug fix for priority queues.
1938    
1939    ----------------------------------------------------------------------
1940    Name: Matthias Blume
1941    Date: 2002/02/15 17:05:00
1942    Tag: Release_110_39
1943    Description:
1944    
1945    Last-minute retagging is becoming a tradition... :-(
1946    
1947    This is the working release 110.39.
1948    
1949    ----------------------------------------------------------------------
1950    Name: Matthias Blume
1951    Date: 2002/02/15 16:00:00 EST
1952    Tag: Release_110_39-orig
1953    Description:
1954    
1955    Working release 110.39.  New bootfiles.
1956    
1957    (Update: There was a small bug in the installer so it wouldn't work
1958    with all shells.  So I retagged. -Matthias)
1959    
1960    ----------------------------------------------------------------------
1961    Name: Matthias Blume
1962    Date: 2002/02/15 14:17:00 EST
1963    Tag: blume-20020215-showbindings
1964    Description:
1965    
1966    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
1967    the latter can be useful for exploring what bindings are available at
1968    the interactive prompt.  (The first function returns only the list
1969    of symbols that are really bound, the second prints those but also the
1970    ones that CM's autoloading mechanism knows about.)
1971    
1972    ----------------------------------------------------------------------
1973    Name: Matthias Blume
1974    Date: 2002/02/15 12:08:00 EST
1975    Tag: blume-20020215-iptrs
1976    Description:
1977    
1978    Two improvements to ml-nlffigen:
1979    
1980      1. Write files only if they do not exist or if their current contents
1981         do not coincide with what's being written.  (That is, avoid messing
1982         with the time stamps unless absolutely necessary.)
1983    
1984      2. Implement a "repository" mechanism for generated files related
1985         to "incomplete pointer types".   See the README file for details.
1986    
1987    ----------------------------------------------------------------------
1988    Name: Matthias Blume
1989    Date: 2002/02/14 11:50:00 EST
1990    Tag: blume-20020214-quote
1991    Description:
1992    
1993    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
1994    because of the new and improved tag generation scheme.  (Thanks to Allen
1995    Leung for pointing it out.)
1996    
1997    ----------------------------------------------------------------------
1998    Name: Lal George
1999    Date: 2002/02/14 09:55:27 EST 2002
2000    Tag: george-20020214-isabelle-bug
2001    Description:
2002    
2003    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2004    of Isabelle on the x86.
2005    
2006    From Allen:
2007    -----------
2008     I've found the problem:
2009    
2010         in ra-core.sml, I use the counter "blocked" to keep track of the
2011         true number of elements in the freeze queue.  When the counter goes
2012         to zero, I skip examining the queue.  But I've messed up the
2013         bookkeeping in combine():
2014    
2015             else ();
2016             case !ucol of
2017               PSEUDO => (if !cntv > 0 then
2018                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2019                                        ^^^^^^^^^^^^^^^^^^^^^^^
2020                      moveu := mergeMoveList(!movev, !moveu)
2021                     )
2022                  else ();
2023    
2024         combine() is called to coalesce two nodes u and v.
2025         I think I was thinking that if the move counts of u and v are both
2026         greater than zero then after they are coalesced then one node is
2027         removed from the freeze queue.  Apparently I was thinking that
2028         both u and v are of low degree, but that's clearly not necessarily true.
2029    
2030    
2031    02/12/2002:
2032        Here's the patch.  HOL now compiles.
2033    
2034        I don't know how this impact on performance (compile
2035        time or runtime).  This bug caused the RA (especially on the x86)
2036        to go thru the potential spill phase when there are still nodes on the
2037        freeze queue.
2038    
2039    
2040    
2041    
2042    ----------------------------------------------------------------------
2043    Name: Matthias Blume
2044    Date: 2002/02/13 22:40:00 EST
2045    Tag: blume-20020213-fptr-rtti
2046    Description:
2047    
2048    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2049    updates.
2050    
2051    ----------------------------------------------------------------------
2052    Name: Matthias Blume
2053    Date: 2002/02/13 16:41:00 EST
2054    Tag: blume-20020213-cmlpq
2055    Description:
2056    
2057    Added new priority queue export symbols (which have just been added to
2058    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2059    not compile and the installer would choke.)
2060    
2061    ----------------------------------------------------------------------
2062    Name: Matthias Blume
2063    Date: 2002/02/13 16:15:00 EST
2064    Tag: blume-20020213-various
2065    Description:
2066    
2067    1. More tweaks to ml-nlffigen:
2068    
2069       - better internal datastructures (resulting in slight speedup)
2070       - "-match" option requires exact match
2071       - "localized" gensym counters (untagged structs/unions nested within
2072         other structs/unions or within typedefs get a fresh counter; their
2073         tag will be prefixed by a concatenation of their parents' tags)
2074       - bug fixes (related to calculation of transitive closure of types
2075         to be included in the output)
2076    
2077    2. Minor Basis updates:
2078    
2079       - added implementations for List.collate and Option.app
2080    
2081    ----------------------------------------------------------------------
2082    Name: Matthias Blume
2083    Date: 2002/02/11 15:55:00 EST
2084    Tag: blume-20020211-gensym
2085    Description:
2086    
2087    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2088    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2089    names (ML structure names that correspond to unnamed C structs, unions,
2090    and enums), so that separate runs of ml-nlffigen do not clash.
2091    
2092    ----------------------------------------------------------------------
2093    Name: Matthias Blume
2094    Date: 2002/02/11 12:05:00 EST
2095    Tag: blume-20020211-gensml
2096    Description:
2097    
2098    A quick fix for a problem with GenSML (in the pgraph-util library):
2099    Make generation of toplevel "local" optional.  (Strictly speaking,
2100    signature definitions within "local" are not legal SML.)
2101    
2102    Other than that: updates to INSTALL and cm/TODO.
2103    
2104    ----------------------------------------------------------------------
2105    Name: Matthias Blume
2106    Date: 2002/02/08 15:00:00 EST
2107    Tag: blume-20020208-uniquepid
2108    Description:
2109    
2110    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2111    
2112    1. The installer (config/install.sh) has gotten smarter:
2113    
2114         - Configuration options are a bit easier to specify now
2115           (in config/targets).
2116         - Bug in recognizing .tar.bz2 files fixed.
2117         - Installer automatically resolves dependencies between
2118           configuration options (e.g., if you ask for eXene, you will
2119           also get cml -- regardless whether you asked for it or not).
2120         - Installer can run in "quieter mode" by setting the environment
2121           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2122           "completely silent", though.
2123         - Build HashCons library as part of smlnj-lib.
2124    
2125    2. A new scheme for assigning persistent identifiers to compilation
2126       units (and, by extension, to types etc.) has been put into place.
2127       This fixes a long-standing bug where types and even dynamic values
2128       can get internally confused, thereby compromising type safety
2129       (abstraction) and dynamic correctness.  See
2130    
2131         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2132    
2133       for an example of how things could go wrong until now.
2134    
2135       The downside of the new scheme is that pids are not quite as
2136       persistent as they used to be: CM will generate a fresh pid
2137       for every compilation unit that it thinks it sees for the first
2138       time.  That means that if you compile starting from a clean, fresh
2139       source tree at two different times, you end up with different
2140       binaries.
2141    
2142       Cutoff recompilation, however, has not been compromised because
2143       CM keeps pid information in special caches between runs.
2144    
2145    ----------------------------------------------------------------------
2146    Name: Lal George
2147    Date: 2002/02/07 15:34:13 EST 2002
2148    Tag: <none>
2149    Description:
2150    
2151    Compilers that generate assembly code may produce  global labels
2152    whose value is resolved at link time. The various peephole optimization
2153    modules did not take this in account.
2154    
2155    TODO. The Labels.addrOf function should really return an option
2156    type so that clients are forced to deal with this issue, rather
2157    than an exception being raised.
2158    
2159    ----------------------------------------------------------------------
2160    Name: Lal George
2161    Date: 2002/02/06 13:55:02 EST
2162    Tag: george-20020206-ra-breakup
2163    Description:
2164    
2165    1. A bug fix from Allen.
2166    
2167        A typo causes extra fstp %st(0)'s to be generated at compensation
2168        edges, which might cause stack underflow traps at runtime.  This
2169        occurs in fft where there are extraneous fstps right before the 'into'
2170        trap instruction (in this case they are harmless since none of the
2171        integers overflow.)
2172    
2173    2. Pulled out various utility modules that were embedded in the modules
2174       of the register allocator. I need these modules for other purposes, but
2175       they are not complete enough to put into a library (just yet).
2176    ----------------------------------------------------------------------
2177    Name: Matthias Blume
2178    Date: 2002/01/31 16:05:00 EST
2179    Tag: blume-20020131-sparc-ccalls
2180    Description:
2181    
2182    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2183       of extra stack space by mistake.  Fixed.
2184    
2185    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2186    
2187    ----------------------------------------------------------------------
2188    Name: Allen Leung
2189    Date: 2002/01/30
2190    Tag: leunga-20020130-nowhere-bug-fix
2191    Description:
2192    
2193       MLRISC bug fixes:
2194       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2195       2. MachineInt.fromString was negating its value.
2196    
2197    ----------------------------------------------------------------------
2198    Name: Matthias Blume
2199    Date: 2002/01/29
2200    Tag: blume-20020129-INSTALL
2201    Description:
2202    
2203    - Added somewhat detailed installation instructions (file INSTALL).
2204    - Fixed curl-detection bug in config/install.sh.
2205    - It is now possible to select the URL getter using the URLGETTER
2206      environment variable:
2207    
2208          not set / "unknown"      --> automatic detection (script tries wget,
2209                                       curl, and lynx)
2210          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2211                                       how to properly invoke them)
2212          other                    --> use $URLGETTER directly, it must take
2213                                       precisely two command-line arguments
2214                                       (source URL and destination file name)
2215    
2216    ----------------------------------------------------------------------
2217    Name: Matthias Blume
2218    Date: 2002/01/28
2219    Tag: blume-20020128-sparc-ccalls
2220    Description:
2221    
2222    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2223    - Make use of the allocParam argument in sparc-c-calls.
2224    
2225    ----------------------------------------------------------------------
2226  Name: Matthias Blume  Name: Matthias Blume
2227  Date: 2002/01/28  Date: 2002/01/28
2228  Tag: blume-20020128-allocParam  Tag: blume-20020128-allocParam
# Line 584  Line 2794 
2794    
2795  ----------------------------------------------------------------------  ----------------------------------------------------------------------
2796  Name: Matthias Blume  Name: Matthias Blume
 >>>>>>> 1.169  
2797  Date: 2001/09/18 15:35:00 EDT  Date: 2001/09/18 15:35:00 EDT
2798  Tag: blume-20010918-readme11036  Tag: blume-20010918-readme11036
2799  Description:  Description:

Legend:
Removed from v.1044  
changed lines
  Added in v.1394

root@smlnj-gforge.cs.uchicago.edu
ViewVC Help
Powered by ViewVC 1.0.0