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 1049, Thu Jan 31 21:03:49 2002 UTC revision 1389, Wed Sep 24 20:09:40 2003 UTC
# Line 13  Line 13 
13  Description:  Description:
14    
15  ----------------------------------------------------------------------  ----------------------------------------------------------------------
16    Name: Matthias Blume (blume (at) tti - c (dot) org)
17    Date: 2003/09/24 15:10:05 CDT
18    Tag: blume-20030924-sockets
19    Description:
20    
21    1. Fixed SOCKET API and implementation to match Basis spec.
22       This required changing the internal representation of sockets to one
23       that remembers (for each socket file descriptor) whether it is currently
24       blocking or non-blocking.  This state is maintained lazily (i.e., a system
25       call is made only if the state actually needs to change).
26    
27    2. OS-specific details of sockets were moved into separate files, thus
28       making it possible to unify the bulk of the socket implementations
29       between Unix and Win32.
30    
31    3. CML's socket API changed accordingly.
32       (Note that we need to remove non-blocking functions from this API
33       since they are redundant in the case of CML!)
34    
35    4. CML's socket implementation now makes use of non-blocking functions
36       provided by Basis, thus removing all OS-dependent code from this part
37       of CML.
38    
39    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
40    
41    ----------------------------------------------------------------------
42    Name: Matthias Blume (blume (at) tti - c (dot) org)
43    Date: 2003/09/22 12:10:00 CDT
44    Tag: blume-20030922-110_43_2
45    Description:
46    
47    Made a new interim version and bootfiles for developer's bootstrapping
48    convenience.
49    
50       110.43.2 -- NEW BOOTFILES
51    
52    ----------------------------------------------------------------------
53    Name: Matthias Blume (blume (at) tti - c (dot) org)
54    Date: 2003/09/19 15:55:00 CDT
55    Tag: blume-20030919-cmdir
56    Description:
57    
58    1. new-install.sh -> install.sh
59    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
60    3. tweaked installer so that another name instead of .cm can be chosen
61       at install time (by setting the CM_DIR_ARC environment variable
62       during installation); once installation is complete, the name is
63       fixed
64    
65    ----------------------------------------------------------------------
66    Name: Matthias Blume (blume (at) tti - c (dot) org)
67    Date: 2003/09/18 16:00:00 CDT
68    Tag: blume-20030918-110_43_1
69    Description:
70    
71    Made a new interim version and bootfiles for developer's bootstrapping
72    convenience.
73    
74       110.43.1 -- NEW BOOTFILES
75    
76    ----------------------------------------------------------------------
77    Name: Matthias Blume (blume (at) tti - c (dot) org)
78    Date: 2003/09/18 15:20:00 CDT
79    Tag: blume-20030918-misc
80    Description:
81    
82    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
83       this seems to be controversial at the moment) and used those in
84       Posix.ProcEnv.times.
85    
86    2. Added Time.{from,to}Nanoseconds to Time.
87    
88    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
89       For example, fromLargeInt never needs to look at more than 3 "big
90       digits" to get its 53 bits of precision.
91    
92    ----------------------------------------------------------------------
93    Name: Matthias Blume (blume (at) tti - c (dot) org)
94    Date: 2003/09/17 16:30:00 CDT
95    Tag: blume-20030917-real32-slices
96    Description:
97    
98    Added an entry to the primitive environment
99    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
100    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
101    
102    Removed some of the "magic" constants in real64.sml and replaced them
103    with code that generates these values from their corresponding
104    integer counterparts.
105    
106    Made all(?) the slice-related changes to the Basis and made everything
107    compile again...
108    
109    ----------------------------------------------------------------------
110    Name: Matthias Blume (blume (at) tti - c (dot) org)
111    Date: 2003/09/15 17:45:00 CDT
112    Tag: blume-20030915-rbase
113    Description:
114    
115    Fixed bug in Real.fromLargeInt.
116    
117    ----------------------------------------------------------------------
118    Name: Matthias Blume (blume (at) tti - c (dot) org)
119    Date: 2003/09/13 18:11:00 CDT
120    Tag: blume-20030913-libinstall
121    Description:
122    
123    Minor bugfix in config/libinstall (set anchor with path to
124    standalone tool after installing it, otherwise libraries that
125    need ml-lex or ml-yacc won't compile the first time the installer
126    runs).
127    
128    ----------------------------------------------------------------------
129    Name: Matthias Blume (blume (at) tti - c (dot) org)
130    Date: 2003/09/12 11:45:00 CDT
131    Tag: blume-20030912-various
132    Description:
133    
134     - fixed bug in Real.toLargeInt
135     - fixed bug in Posix.ProcEnv.times
136     - changed inputLine functions to return an option
137     - minor installer improvements / bugfixes
138     - changed default @SMLalloc parameter for x86/celeron to 64k
139    
140    ----------------------------------------------------------------------
141    Name: Matthias Blume (blume (at) tti - c (dot) org)
142    Date: 2003/09/09 22:00:00 CDT
143    Tag: Release_110_43
144    Description:
145    
146        New working release 110.43.  New bootfiles.
147    
148    ----------------------------------------------------------------------
149    Name: Matthias Blume (blume (at) tti - c (dot) org)
150    Date: 2003/09/09 19:20:00 CDT
151    Tag: blume-20030909-installer
152    Description:
153    
154    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
155    Modified config/install.bat to take advantage of it.  Also modified
156    config/install.sh (and called it config/new-install.sh) to take advantage
157    of it on Unix systems.  (The SML code is (supposed to be) platform-
158    independent.)
159    
160    The installer can now install everything under Win32
161    as well as under *nix as long as it compiles.
162    
163    Other changes:
164    
165      - made CML compile again under Win32
166      - made eXene compile under Win32 (by providing a fake structure UnixSock
167        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
168      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
169        same as type int;  under Win32 it isn't
170      - fixed some slice-related problems in the win32-specific parts of CML
171      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
172        (under Win32, the volume name is case-insensitive, and the
173        OS.Path code compares volume names for equality)
174    
175    ----------------------------------------------------------------------
176    Name: Matthias Blume (blume (at) tti - c (dot) org)
177    Date: 2003/09/08 11:55:00 CDT
178    Tag: blume-20030908-fullpath
179    Description:
180    
181    Made Win32 version of OS.FileSys.fullPath return current directory
182    when given an empty string.  This is what the spec says, and incidentally,
183    CM depends on it.  (CM otherwise goes into an infinite loop in certain
184    cases when presented with the name of a non-existing .cm file.)
185    
186    ----------------------------------------------------------------------
187    Name: Matthias Blume (blume (at) tti - c (dot) org)
188    Date: 2003/09/04 16:30:00 CDT
189    Tag: blume-20030905-slices-etc
190    Description:
191    
192      1. Changed interface to vectors and arrays in Basis to match
193         (draft) Basis spec.
194      2. Added signatures and implementations of slices according to
195         Basis spec.
196      3. Edited source code throughout the system to make it compile again
197         under 1. and 2.  (In some cases code had to be added to have it
198         match the new signatures.)
199      4. MLRISC should be backward-compatible: the copies of the originals
200         of files that needed to change under 3. were retained, the .cm files
201         check the compiler version number and use old versions when
202         appropriate.
203      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
204         dirstream -> string option (in accordance with Basis spec).
205      6. When generating code that counts lines, ml-lex used function
206         CharVector.foldli, taking advantage of its old interface.
207         This has been replaced with the corresponding code from
208         CharVectorSlice.  (html-lex must be re-lexed!)
209      7. BitArray in smlnj-lib/Util has been extended/modified to match the
210         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
211      8. Removed temporary additions (fromInternal, toInternal) from the
212         (now obsolete) IntInf in smlnj-lib/Util.
213      9. Cleaned up structure Byte.
214     10. Added localOffset, scan, and fromString to Date (according to spec).
215         Cleaned/corrected implementation of Date.
216         (Still need to check for correctness; implement better canonicalizeDate.)
217     11. Added "scan" to signature IEEE_REAL.
218     12. Some improvements to IntInf [in particular: efficiency-hack for
219         mod and rem when second operand is 2 (for parity checks).]
220     13. Changed representation of type Time.time, using a single IntInf.int
221         value counting microseconds.  This considerably simplified the
222         implementation of structure Time.  We now support negative time
223         values; scan and fromString handle signs.
224     14. Functor PrimIO now takes two additional arguments (VectorSlice and
225         ArraySlice).
226    
227    ----------------------------------------------------------------------
228    Name: Matthias Blume (blume (at) tti - c (dot) org)
229    Date: 2003/08/28 17:00:00 CDT
230    Tag: blume-20030828-intinf
231    Description:
232    
233    This is a major update which comes with a version number bump
234    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
235    and an implementation of IntInf in the Basis.
236    
237    There are a fairly large number of related changes and updates throughout
238    the system:
239    
240    Basis:
241            - Implemented IntInf.
242            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
243            - Added some missing Real64 operations, most notably Real.toLargeInt.
244            - Added FixedInt as a synonym for Int32.
245    
246    compiler:
247            * Added support for a built-in intinf type.
248              - literals
249              - pattern matching
250              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
251              - overloading on literals and operations
252    
253            This required adding a primitive type intinf, some additional
254            primops, and implementations for several non-trivial intinf
255            operations in Core.  (The intinf type is completely abstract
256            to the compiler; all operations get delegated back to the Core.)
257    
258            * Intinf equality is handled by polyequal.  However, the compiler
259            does not print its usual warning in this case (since polyequal
260            is the right thing to do there).
261    
262            * Improved the organization of structure InlineT.
263    
264            * A word about conversion primops:
265            If conversions involving intinf do not cancel out during
266            CPS contract, then the compiler must insert calls to Core functions.
267            Since all core access must be resolved already during the FLINT
268            translate phase, it would be too late a the time of CPS contract
269            to add new Core calls.  For this reason, conversion primops
270            for intinf carry two arguments:  1. the numeric argument that
271            they are supposed to convert, and 2. the Core function that
272            can help with this conversion if necessary.  If CPS contract
273            eliminates a primop, then the associated Core function becomes
274            dead and goes away.  Intinf conversion primops that do not get
275            eliminated by CPS contract get rewritten into calls of their
276            core functions by a separate, new phase.
277    
278    interactive system:
279            - Control.Print.intinfDepth controls max length of intinf constants
280            being printed. (Analogous to Control.Print.stringDepth.)
281            - Cleanup in printutil and pputil: got rid of unused stuff and
282            duplicates; replaced some of the code with code that makes better
283            use of library functionality.
284    
285    CM:
286            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
287            are no longer silent.
288    
289    CKIT:
290            Fixed mismatched uses of Int32 and LargeInt.  I always decided
291            in favor of LargeInt -- which is now the same as IntInf.
292            CKIT-knowledgable people should check whether this is what's
293            intended and otherwise change things back to using Int32 or
294            FixedInt.
295    
296    Throughout the code:
297            Started using IntInf.int literals and built-in operations
298            (e.g., comparison with 0) where this seems appropriate.
299    
300    
301    ----------------------------------------------------------------------
302    Name: Dave MacQueen (dbm@cs.uchicago.edu)
303    Date: 2003/08/13 11:36:00 CDT
304    Tag: dbm-20030813-mcz-merge1
305    Description:
306    
307    Merging changes from the mcz-branch development branch into trunk.
308    These changes involve replacement of the emulated old prettyprinter
309    interface with direct use of the SML/NJ Lib PP library, and fixing
310    of a couple of bugs (895, 1186) relating to error messages.  A new
311    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
312    has been added.
313    
314    ----------------------------------------------------------------------
315    Name: Matthias Blume (blume (at) tti - c (dot) org)
316    Date: 2003/08/11 15:45:00 CDT
317    Tag: blume-20030811-windows
318    Description:
319    
320    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
321    
322        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
323    
324    This patch restores SML/NJ's ability to run under win32.  There are a
325    number of changes, including fixes for several bugs that had gone
326    unnoticed until now:
327    
328      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
329        trying to run the win32 version from within cygwin.)
330    
331      - There are a number of simple .bat scripts that substitute for their
332        corresponding Unix shell-scripts.  (See below.)
333    
334      - The internals of ml-build have been modified slightly.  The main
335        difference is that instead of calling ".link-sml" (or link-sml.bat)
336        using OS.Process.system, the ML process delegates this task back
337        to the script.  Otherwise problems arise in mixed environments such
338        as Cygwin where scripts look and work like Unix scripts, but
339        where OS.Process.system cannot run them.
340    
341      - In CM, the srcpath pickler used native pathname syntax -- which
342        is incorrect in the case of cross-compilation.  The new pickle format
343        is independent of platform-specific naming conventions.
344    
345      - Path configuration files (such as lib/pathconfig) can now choose
346        between native and standard syntax.  Placing a line of the form
347    
348            standard!
349    
350        into the file causes all subsequent paths to be interpreted using
351        CM standard pathname syntax (= Unix conventions); a line
352    
353            native!
354    
355        switches back to native style.  This was needed so that
356        path config files can be written portably, see src/system/pathconfig.
357    
358      - Runtime system:
359    
360          - win32-filesys.c:  get_file_time and set_file_time now
361            access modification time, not creation time.
362    
363          - I/O code made aware of new array representation.
364    
365          - Bug fixes in X86.prim.masm.
366    
367       - src/system/makeml made aware of win32. (For use under cygwin
368            and other Unix-environments for windows.)
369    
370       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
371         which caused BinIO.inputAll to fail consistently.
372    
373    .bat scripts:
374    
375       Windows .bat scripts assume that SMLNJ_HOME is defined.
376    
377       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
378         applications (sml, ml-yacc, ml-lex).
379       - ml-build.bat: analogous to ml-build.
380       - config\install.bat:  Analogous to config/install.sh.  This requires
381         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
382         (nmake etc. must be on the path, and vcvars32 must have been run.)
383         Moreover, sources for ml-lex and ml-yacc need to exist under src,
384         and the bootfile hierarchy must have been unpacked under
385         sml.boot.x86-win32.
386         The script is very primitive and does a poor job at error checking.
387         It only installs the base system, ml-lex, and ml-yacc.  No other
388         libraries are being installed (i.e., you get only those that
389         are part of the compiler.)
390       - link-sml.bat:  analogous to .link-sml, but not currently used
391    
392    Unrelated bug fixes:
393    
394      - ml-nlffigen now exports structures ST_* corresponding to incomplete
395        types.
396      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
397        otherwise.)
398    
399    ----------------------------------------------------------------------
400    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
401    Date: 2003/06/17
402    Tag: macqueen-20030617-bug895
403    Description:
404    
405    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
406    Tag will be used for new development branch (mcz-branch) for
407    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
408    
409    ----------------------------------------------------------------------
410    Name: Matthias Blume (blume (at) tti - c (dot) org)
411    Date: 2003/05/27 16:55:00 CDT
412    Tag: blume-20030527-polyeq
413    Description:
414    
415    Tried to eliminated most cases of polymorphic equality.
416    
417    ----------------------------------------------------------------------
418    Name: Matthias Blume (blume (at) tti - c (dot) org)
419    Date: 2003/05/21 17:45:00 CDT
420    Tag: blume-20030517-complete
421    Description:
422    
423    Two changes:
424    
425      1. Added a flag for controlling whether non-exhaustive bindings will
426         be treated as errors (default is false).
427      2. Cleaned up the *entire* source tree so that CMB.make goes through
428         without a single non-exhaustive match- or bind warning.
429    
430    ----------------------------------------------------------------------
431    Name: Matthias Blume (blume (at) tti - c (dot) org)
432    Date: 2003/05/17 10:20:00 CDT
433    Tag: blume-20030517-absyn
434    Description:
435    
436    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
437    
438       This mainly affects the quality of error messages.  However, some
439       of the code is now more straightforward than before. (Treatment of
440       the above four constructs in translate.sml is much simpler than
441       the "macro-expansion" that was going on before.  Plus, the mach-
442       compiler no longer gets invoked just to be able to compile an
443       if-expression.)
444    
445    2. The ErrorMsg.Error exception is now caught and absorbed by the
446       interactive loop.
447    
448    ----------------------------------------------------------------------
449    Name: Allen Leung
450    Date: 2003/05/16 13:05:00 CDT
451    Tag: leunga-20030516-cygwin-runtime
452    Description:
453    
454        Ported the runtime system to cygwin, which uses the unix
455    x86-unix bin files.  Missing/buggy features:
456    
457       o getnetbyname, getnetbyaddr: these functions seem to be missing in
458         the Cygwin library.
459       o Ctrl-C handling may be flaky.
460       o Windows system calls and Windows I/O are not supported.
461    
462       A new set of binfiles is located at:
463    
464         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
465    
466       This is only needed for bootstrapping the cygwin version of smlnj.
467       Other x86 versions can use the existing binfiles.
468    
469    ----------------------------------------------------------------------
470    Name: Matthias Blume
471    Date: 2003/04/08 15:42:00 CDT
472    Tag: blume-20030408-listpair
473    Description:
474    
475    1. Added a target 'mlrisc' to installer.
476    
477    2. Added missing elements to structure ListPair.
478    
479    ----------------------------------------------------------------------
480    Name: Allen Leung
481    Date: 2003/01/07 10:40:00 EST
482    Tag: leunga-20030107-int-rem
483    Description:
484    
485    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
486    The arguments to the SUBL instruction were swapped.
487    
488    ----------------------------------------------------------------------
489    Name: Matthias Blume
490    Date: 2002/12/12 16:25:00 EST
491    Tag: blume-20021212-risc-ra
492    Description:
493    
494    Fixed a serious bug in the rewrite code for FP spilling/reloading that
495    sent the RA into an infinite loop when floating point registers get
496    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
497    110.37 and 110.38.)
498    There was another set of potential problems related to the handling of
499    MLRISC annotations (but those did not yet cause real problems, apparently).
500    
501    ----------------------------------------------------------------------
502    Name: Matthias Blume
503    Date: 2002/12/06 22:40:00 EST
504    Tag: blume-20021206-cm-fileid
505    Description:
506    
507    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
508    This fixes the problem of CM getting confused by files that suddenly
509    change their identity (e.g., by getting unlinked and recreated by some
510    text editor such as vi).  There might be a better/cheaper/cleaner way
511    of doing this, but for now this will have to do.
512    
513    ----------------------------------------------------------------------
514    Name: Matthias Blume
515    Date: 2002/10/28 09:50:00 EST
516    Tag: blume-20021028-typecheck
517    Description:
518    
519    Exported structure Typecheck from $smlnj/viscomp/core.cm.
520    
521    ----------------------------------------------------------------------
522    Name: Matthias Blume
523    Date: 2002/10/17 09:10:00 EDT
524    Tag: Release_110_42
525    Description:
526    
527    In good old tradition, there has been a slight hiccup so that we have
528    to patch 110.42 after the fact.   The old release tag has been replaced
529    (see below).
530    
531    The change solves a problem with two competing approaches the
532    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
533    each other's way.
534    
535    This change only affects the runtime system code and the installer script.
536    (No new bootfiles.)
537    
538    ----------------------------------------------------------------------
539    Name: Matthias Blume
540    Date: 2002/10/16 12:00:00 EDT
541    Tag: Release_110_42_removed
542    Description:
543    
544    New working release.  New bootfiles.
545    
546    ----------------------------------------------------------------------
547    Name: Matthias Blume
548    Date: 2002/10/10 13:10:00 EDT
549    Tag: blume-20021010-ppc-divs
550    Description:
551    
552    The mltree operator DIVS must be implemented with an overflow check on
553    the PPC because the hardware indicates divide-by-zero using "overflow" as
554    well.
555    
556    ----------------------------------------------------------------------
557    Name: Matthias Blume
558    Date: 2002/07/23 11:45:00
559    Tag: blume-20020723-smlnj-home
560    Description:
561    
562    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
563    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
564    CM_PATHCONFIG is also set), the path configuration file is assumed
565    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
566    the entire tree to some other place and everything will "just work".
567    
568    (Companion commands such as ml-build and ml-makedepend also sense this
569    variable.)
570    
571    ----------------------------------------------------------------------
572    Name: Matthias Blume
573    Date: 2002/07/12 21:19:00 EDT
574    Tag: blume-20020712-liveness
575    Description:
576    
577    Exported two useful "step" functions from liveness module (MLRISC).
578    
579    ----------------------------------------------------------------------
580    Name: Matthias Blume
581    Date: 2002/07/05 16:00 EDT
582    Tag: Release_110_41
583    Description:
584    
585    New working release.  New bootfiles.
586    
587    ----------------------------------------------------------------------
588    Name: Matthias Blume
589    Date: 2002/07/05 10:25:00 EDT
590    Tag: blume-20020705-btimp
591    Description:
592    
593    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
594    other clients can set up backtracing support.
595    
596    ----------------------------------------------------------------------
597    Name: Matthias Blume
598    Date: 2002/06/25 17:23:00 EDT
599    Tag: blume-20020625-fpmax
600    Description:
601    
602    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
603    case.  (The sense of the isNaN test was reversed -- which made min and
604    max always return their first argument.)
605    
606    ----------------------------------------------------------------------
607    Name: Matthias Blume
608    Date: 2002/06/11
609    Tag: blume-20020611-unixpath
610    Description:
611    
612    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
613    
614    ----------------------------------------------------------------------
615    Name: Matthias Blume
616    Date: 2002/06/10 16:35:00 EDT
617    Tag: blume-20020610-ieeereal
618    Description:
619    
620    I back-ported my implementation of IEEEReal.fromString from the
621    idlbasis-devel branch so that we can test it.
622    
623    Another small change is that ppDec tries to give more information
624    than just "<sig>" in the case of functors.  However, this code is
625    broken in some mysterious way if the functor's body's signature
626    has not been declared by ascription but gets inferred from the
627    implementation.  This needs fixing...
628    
629    ----------------------------------------------------------------------
630    Name: Matthias Blume
631    Date: 2002/05/31
632    Tag: blume-20020531-btrace-mode
633    Description:
634    
635    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
636    the wayside when I switched over to using Controls everywhere.)
637    
638    ----------------------------------------------------------------------
639    Name: Lal George
640    Date: 2002/05/23 12:21:40 EDT
641    Tag: george-20020523-visual-labels
642    Description:
643    
644    Labels are now displayed in the graphical output to make
645    the fall-through and target blocks obvious.
646    
647    ----------------------------------------------------------------------
648    Name: Matthias Blume
649    Date: 2002/05/22 11:03:00 EDT
650    Tag: blume-20020522-shrink
651    Description:
652    
653    John tweaked yesterday's fix for 1131 to handle an out-of-memory
654    situation that comes up when allocating huge arrays.
655    
656    ----------------------------------------------------------------------
657    Name: Matthias Blume
658    Date: 2002/05/21 16:00:00 EDT
659    Tag: Release_110_40
660    Description:
661    
662    New working release (110.40).  New bootfiles.
663    
664    [Also: John Reppy fixed GC bug 1131.]
665    
666    ----------------------------------------------------------------------
667    Name: Matthias Blume
668    Date: 2002/05/21 12:35:00 EDT
669    Tag: blume-20020521-cmdoc
670    Description:
671    
672    CM documentation update.
673    
674    ----------------------------------------------------------------------
675    Name: Matthias Blume
676    Date: 2002/05/21 10:55:00 EDT
677    Tag: blume-20020521-misc
678    Description:
679    
680    - John tweaked runtime to be silent on heap export (except when
681      GC messages are on).
682    - I added a few more things (cross-compiling versions of CMB) to
683      config/preloads (as suggestions).
684    
685    ----------------------------------------------------------------------
686    Name: Matthias Blume
687    Date: 2002/05/20 22:25:00 EDT
688    Tag: blume-20020520-controls
689    Description:
690    
691    - Added ControlUtil structure to control-lib.cm.
692    - Use it throughout.
693    - Used Controls facility to define MLRISC controls (as opposed to
694      registering MLRISC control ref cells with Controls after the fact)
695    - Fixed messed-up controls priorities.
696    
697    * Removed again all the stuff from config/preloads that one wouldn't
698      be able to preload at the time the initial heap image is built.
699      (Many libraries, e.g., CML, do not exist yet at this time.  The
700       only libraries that can be preloaded via config/preloads are those
701       that come bundled with the bootfiles.)
702    
703    ----------------------------------------------------------------------
704    Name: Matthias Blume
705    Date: 2002/05/20 10:59:00 EDT
706    Tag: blume-20020520-preloads
707    Description:
708    
709    Added a lot of commented-out suggestions for things to be included
710    in config/preloads.
711    
712    ----------------------------------------------------------------------
713    Name: Allen Leung
714    Date: 2002/05/18 14:20:00 EDT
715    Tag: leunga-20020518-mdl
716    Description:
717    
718    o Made the mdl tool stuff compile and run again.
719    o I've disabled all the stuff that depends on RTL specifications; they
720      are all badly broken anyway.
721    
722    ----------------------------------------------------------------------
723    Name: Matthias Blume
724    Date: 2002/05/17 16:49:00 EDT
725    Tag: blume-20020517-controls
726    Description:
727    
728    0. John Reppy made several modifications to the SML/NJ library.
729       In particular, there is a shiny new controls-lib.cm.
730    
731    1. Pushed new controls interface through compiler so that everything
732       compiles again.
733    
734    2. Added FormatComb and FORMAT_COMB to the CML version of the
735       SML/NJ library (so that CML compiles again).
736    
737    3. Modified init scripts because XXX_DEFAULT environment variables
738       are no longer with us.  (Boot-time initialization is now done
739       using the same environment variables that are also used for
740       startup-time initialization of controls.)
741    
742    ----------------------------------------------------------------------
743    Name: Lal George
744    Date: 2002/05/15 09:20:10 EDT
745    Tag: george-20020515-pseudo-op-decls
746    Description:
747    
748            All pseudo-ops emitted before the first segment declaration
749    such as TEXT, DATA, and BSS directives are assumed to be global
750    declarations and are emitted first in the assembly file. This is
751    useful in a number of situations where one has pseudo-ops that are not
752    specific to any segment, and also works around the constraint that one
753    cannot have client pseudo-ops in the TEXT segment.
754    
755            Because no segment is associated with these declarations it is
756    an error to allocate any space or objects before the first segment
757    directive and an exception will be raised. However, we cannot make
758    this check for client pseudo-ops.
759    
760            These top level declarations are a field in the CFG graph_info.
761    In theory you can continue to add to this field after the CFG has been
762    built -- provided you know what you are doing;-)
763    
764    ----------------------------------------------------------------------
765    Name: Matthias Blume
766    Date: 2002/05/13 16:40:00 EDT
767    Tag: blume-20020513-pp-etc
768    Description:
769    
770    A few minor bugfixes:
771    
772      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
773        (Bogus printouts for redefined bindings still occur.  Compiler
774        bug should no longer occur now.  We need to redo the prettyprinter
775        from scratch.)
776    
777      - CM pathname printer now also adds escape sequences for ( and )
778    
779      - commend and docu fixes for ml-nlffi
780    
781    ----------------------------------------------------------------------
782    Name: Matthias Blume
783    Date: 2002/05/10 16:40:00 EDT
784    Tag: blume-20020510-erg-textio
785    Description:
786    
787    Applied the following bugfix provided by Emden Gansner:
788    
789        Output is corrupted when outputSubstr is used rather than output.
790    
791        The problem occurs when a substring
792    
793            ss = (s, dataStart, dataLen)
794    
795        where dataStart > 0, fills a stream buffer with avail bytes left.
796        avail bytes of s, starting at index dataStart, are copied into the
797        buffer, the buffer is flushed, and then the remaining dataLen-avail
798        bytes of ss are copied into the beginning of the buffer. Instead of
799        starting this copy at index dataStart+avail in s, the current code
800        starts the copy at index avail.
801    
802        Fix:
803        In text-io-fn.sml, change line 695 from
804             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
805        to
806             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
807    
808    ----------------------------------------------------------------------
809    Name: Matthias Blume
810    Date: 2002/04/12 13:55:00 EDT
811    Tag: blume-20020412-assyntax
812    Description:
813    
814    1. Grabbed newer assyntax.h from the XFree86 project.
815    2. Fiddled with how to compile X86.prim.asm without warnings.
816    3. (Very) Minor cleanup in CM.
817    
818    ----------------------------------------------------------------------
819    Name: Matthias Blume
820    Date: 2002/04/01 (no joke!) 17:07:00 EST
821    Tag: blume-20020401-x86div
822    Description:
823    
824    Added full support for div/mod/rem/quot on the x86, using the machine
825    instruction's two results (without clumsily recomputing the remainder)
826    directly where appropriate.
827    
828    Some more extensive power-of-two support was added to the x86 instruction
829    selector (avoiding expensive divs, mods, and muls where they can be
830    replaced with cheaper shifts and masks).  However, this sort of thing
831    ought to be done earlier, e.g., within the CPS optimizer so that
832    all architectures benefit from it.
833    
834    The compiler compiles to a fixed point, but changes might be somewhat
835    fragile nevertheless.  Please, report any strange things that you might
836    see wrt. div/mod/quot/rem...
837    
838    ----------------------------------------------------------------------
839    Name: Matthias Blume
840    Date: 2002/03/29 17:22:00
841    Tag: blume-20020329-div
842    Description:
843    
844    Fixed my broken div/mod logic.  Unfortunately, this means that the
845    inline code for div/mod now has one more comparison than before.
846    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
847    The problem was with quotient = 0, because that alone does not tell
848    us which way the rounding went.  One then has to look at whether
849    remainder and divisor have the same sign...  :(
850    
851    Anyway, I replaced the bootfiles with fresh ones...
852    
853    ----------------------------------------------------------------------
854    Name: Matthias Blume
855    Date: 2002/03/29 14:10:00 EST
856    Tag: blume-20020329-inlprims
857    Description:
858    
859    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
860    
861    Primops have changed. This means that the bin/boot-file formats have
862    changed as well.
863    
864    To make sure that there is no confusion, I made a new version.
865    
866    
867    CHANGES:
868    
869    * removed REMT from mltree (remainder should never overflow).
870    
871    * added primops to deal with divisions of all flavors to the frontend
872    
873    * handled these primops all the way through so they map to their respective
874      MLRISC support
875    
876    * used these primops in the implementation of Int, Int32, Word, Word32
877    
878    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
879    
880    * parameterized INLMIN, INLMAX, and INLABS by a numkind
881    
882    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
883      floating point
884    
885    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
886      and Real (but Real.abs maps to a separate floating-point-only primop)
887    
888    
889    TODO items:
890    
891    * Hacked Alpha32 instruction selection, disabling the selection of REMx
892      instructions because the machine instruction encoder cannot handle
893      them.  (Hppa, PPC, and Sparc instruction selection did not handle
894      REM in the first place, and REM is supported by the x86 machine coder.)
895    
896    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
897      selection phase.  (The two can be streamlined because the hardware
898      delivers both quotient and remainder at the same time anyway.)
899    
900    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
901      (Currently the behavior is inconsistent both across architectures and
902      wrt. the draft Basis spec.)
903    
904    * Word8 should eventually be handled natively, too.
905    
906    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
907      as if there currently is no execution path that could trigger it in
908      SML/NJ.  (The assumptions underlying functions arith and promotable do not
909      hold for things like multiplication and division.)
910    
911    ----------------------------------------------------------------------
912    Name: Matthias Blume
913    Date: 2002/03/27 16:27:00 EST
914    Tag: blume-20020327-mlrisc-divisions
915    Description:
916    
917    Added support for all four division operations (ML's div, mod, quot,
918    and rem) to MLRISC.  In the course of doing so, I also rationalized
919    the naming (no more annoying switch-around of DIV and QUOT), by
920    parameterizing the operation by div_rounding_mode (which can be either
921    DIV_TO_ZERO or DIV_TO_NEGINF).
922    
923    The generic MLTreeGen functor takes care of compiling all four
924    operations down to only round-to-zero div.
925    
926    Missing pieces:
927    
928      * Doing something smarter than relying on MLTreeGen on architectures
929        like, e.g., the x86 where hardware division delivers both quotient and
930        remainder at the same time.  With this, the implementation of the
931        round-to-neginf operations could be further streamlined.
932    
933      * Remove inlining support for div/mod/rem from the frontend and replace it
934        with primops that get carried through to the backend.  Do this for all
935        int and word types.
936    
937    ----------------------------------------------------------------------
938    Name: Matthias Blume
939    Date: 2002/03/25 17:25:00 EST
940    Tag: blume-20020325-divmod
941    Description:
942    
943    I improved (hopefully without breaking them) the implementation of Int.div,
944    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
945    advantage of the following observations:
946    
947      Let  q = x quot y      r = x rem y
948           d = x div  y      m = x mod y
949    
950    where "quot" is the round-to-zero version of integer division that
951    hardware usually provides.  Then we have:
952    
953         r = x - q * y        where neither the * nor the - will overflow
954         d = if q >= 0 orelse x = q * y then q else q - 1
955                              where neither the * nor the - will overflow
956         m = if q >= 0 orelse r = 0 then r else r + y
957                              where the + will not overflow
958    
959    This results in substantial simplification of the generated code.
960    The following table shows the number of CFG nodes and edges generated
961    for
962            fun f (x, y) = x OPER y
963            (* with OPER \in div, mod, quot, rem *)
964    
965    
966        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
967        --------------------------------------------------------
968         div |         24 |         39 |         12 |         16
969         mod |         41 |         71 |         12 |         16
970        quot |          8 |         10 |          8 |         10
971         rem |         10 |         14 |          8 |         10
972    
973    
974    ----------------------------------------------------------------------
975    Name: Matthias Blume
976    Date: 2002/03/25 22:06:00 EST
977    Tag: blume-20020325-cprotobug
978    Description:
979    
980    Fixed a bug in cproto (c prototype decoder).
981    
982    ----------------------------------------------------------------------
983    Name: Matthias Blume
984    Date: 2002/03/25 16:00:00 EST
985    Tag: blume-20020325-raw-primops
986    Description:
987    
988    I did some cleanup to Allen's new primop code and
989    replaced yesterday's bootfiles with new ones.
990    (But they are stored in the same place.)
991    
992    ----------------------------------------------------------------------
993    Name: Matthias Blume
994    Date: 2002/03/24 22:40:00 EST
995    Tag: blume-20020324-bootfiles
996    Description:
997    
998    Made the bootfiles that Allen asked for.
999    
1000    ----------------------------------------------------------------------
1001    Name: Allen Leung
1002    Date: 2002/03/23 15:50:00 EST
1003    Tag: leunga-20020323-flint-cps-rcc-primops
1004    Description:
1005    
1006      1. Changes to FLINT primops:
1007    
1008        (* make a call to a C-function;
1009         * The primop carries C function prototype information and specifies
1010         * which of its (ML-) arguments are floating point. C prototype
1011         * information is for use by the backend, ML information is for
1012         * use by the CPS converter. *)
1013      | RAW_CCALL of { c_proto: CTypes.c_proto,
1014                       ml_args: ccall_type list,
1015                       ml_res_opt: ccall_type option,
1016                       reentrant : bool
1017                     } option
1018       (* Allocate uninitialized storage on the heap.
1019        * The record is meant to hold short-lived C objects, i.e., they
1020        * are not ML pointers.  With the tag, the representation is
1021        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1022        *)
1023      | RAW_RECORD of {tag:bool,sz:int}
1024      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1025    
1026      2.  These CPS primops are now overloaded:
1027    
1028           rawload of {kind:numkind}
1029           rawstore of {kind:numkind}
1030    
1031          The one argument form is:
1032    
1033             rawload {kind} address
1034    
1035          The two argument form is:
1036    
1037             rawload {kind} [ml object, byte-offset]
1038    
1039      3. RAW_CCALL/RCC now takes two extra arguments:
1040    
1041         a. The first is whether the C call is reentrant, i.e., whether
1042            ML state should be saved and restored.
1043         b. The second argument is a string argument specifying the name of
1044            library and the C function.
1045    
1046         These things are currently not handled in the code generator, yet.
1047    
1048      4. In CProto,
1049    
1050         An encoding type of "bool" means "ml object" and is mapped into
1051         C prototype of PTR.  Note that "bool" is different than "string",
1052         even though "string" is also mapped into PTR, because "bool"
1053         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1054    
1055      5. Pickler/unpicker
1056    
1057         Changed to handle RAW_RECORD and newest RAW_CCALL
1058    
1059      6. MLRiscGen,
1060    
1061         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1062         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1063            in the file CodeGen/cpscompile/cps-c-calls.sml
1064    
1065      7. Added the conditional move operator
1066    
1067             condmove of branch
1068    
1069         to cps.  Generation of this is still buggy so it is currently
1070         disabled.
1071    
1072    ----------------------------------------------------------------------
1073    Name: Lal George
1074    Date: 2002/03/22 14:18:25 EST
1075    Tag: george-20020322-cps-branch-prob
1076    Description:
1077    
1078    Implemented the Ball-Larus branch prediction-heuristics, and
1079    incorporated graphical viewers for control flow graphs.
1080    
1081    Ball-Larus Heuristics:
1082    ---------------------
1083    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1084    
1085    By design it uses the Dempster-Shafer theory for combining
1086    probabilities.  For example, in the function:
1087    
1088        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1089    
1090    the ball-larus heuristics predicts that the n=0 is unlikely
1091    (OH-heuristic), and the 'then' branch is unlikely because of the
1092    RH-heuristic -- giving the 'then' branch an even lower combined
1093    probability using the Dempster-Shafer theory.
1094    
1095    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1096    probability of the 'then' branch because of the loop in the else
1097    branch.
1098    
1099    
1100    Graphical Viewing:
1101    ------------------
1102    I merely plugged in Allen's graphical viewers into the compiler. The
1103    additional code is not much. At the top level, saying:
1104    
1105            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1106    
1107    will display the graphical view of the control flow graph just before
1108    back-patching.  daVinci must be in your path for this to work. If
1109    daVinci is not available, then the default viewer can be changed
1110    using:
1111    
1112            Control.MLRISC.getString "viewer"
1113    
1114    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1115    course, these viewers must be in your path.
1116    
1117    The above will display the compilation unit at the level of clusters,
1118    many of which are small, boring, and un-interesting. Also setting:
1119    
1120            Control.MLRISC.getInt "cfg-graphical-view_size"
1121    
1122    will display clusters that are larger than the value set by the above.
1123    
1124    
1125    ----------------------------------------------------------------------
1126    Name: Matthias Blume
1127    Date: 2002/03/21 22:20:00 EST
1128    Tag: blume-20020321-kmp-bugfix
1129    Description:
1130    
1131    Changed the interface to the KMP routine in PreString and fixed
1132    a minor bug in one place where it was used.
1133    
1134    ----------------------------------------------------------------------
1135    Name: Allen Leung
1136    Date: 2002/03/21 20:30:00 EST
1137    Tag: leunga-20020321-cfg
1138    Description:
1139    
1140      Fixed a potential problem in cfg edge splitting.
1141    
1142    ----------------------------------------------------------------------
1143    Name: Allen Leung
1144    Date: 2002/03/21 17:15:00 EST
1145    Tag: leunga-20020321-x86-fp-cfg
1146    Description:
1147    
1148      1. Recoded the buggy parts of x86-fp.
1149    
1150         a. All the block reordering code has been removed.
1151            We now depend on the block placement phases to do this work.
1152    
1153         b. Critical edge splitting code has been simplified and moved into the
1154            CFG modules, as where they belong.
1155    
1156         Both of these were quite buggy and complex.  The code is now much, much
1157         simpler.
1158    
1159      2. X86 backend.
1160    
1161         a. Added instructions for 64-bit support.  Instruction selection for
1162            64-bit has not been committed, however, since that
1163            requires changes to MLTREE which haven't been approved by
1164            Lal and John.
1165    
1166         b. Added support for FUCOMI and FUCOMIP when generating code for
1167            PentiumPro and above.  We only generate these instructions in
1168            the fast-fp mode.
1169    
1170         c. Added cases for JP and JNP in X86FreqProps.
1171    
1172      3. CFG
1173    
1174         CFG now has a bunch of methods for edge splitting and merging.
1175    
1176      4. Machine description.
1177    
1178         John's simplification of MLTREE_BASIS.fcond broke a few machine
1179         description things:
1180    
1181         rtl-build.{sig,sml} and hppa.mdl fixed.
1182    
1183         NOTE: the machine description stuff in the repository is still broken.
1184               Again, I can't put my fixes in because that involves
1185               changes to MLTREE.
1186    
1187    ----------------------------------------------------------------------
1188    Name: Matthias Blume
1189    Date: 2002/03/20 15:55:00 EST
1190    Tag: blume-20020320-kmp
1191    Description:
1192    
1193    Implemented Knuth-Morris-Pratt string matching in PreString and used
1194    it for String.isSubstring, Substring.isSubstring, and
1195    Substring.position.
1196    
1197    (Might need some stress-testing.  Simple examples worked fine.)
1198    
1199    ----------------------------------------------------------------------
1200    Name: Matthias Blume
1201    Date: 2002/03/19 16:37:00 EST
1202    Tag: blume-20020319-witnesses
1203    Description:
1204    
1205    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1206    
1207    This implements a generic mechanism for changing constness qualifiers
1208    anywhere within big C types without resorting to outright "casts".
1209    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1210    modify the constness at the outermost level.)
1211    The implementation of "convert" is based on the idea of "witness"
1212    values -- values that are not used by the operation but whose types
1213    "testify" to their applicability.  On the implementation side, "convert"
1214    is simply a projection (returning its second curried argument).  With
1215    cross-module inlining, it should not result in any machine code being
1216    generated.
1217    
1218    ----------------------------------------------------------------------
1219    Name: Matthias Blume
1220    Date: 2002/03/15 16:40:00 EST
1221    Tag: blume-20020315-basis
1222    Description:
1223    
1224    Provided (preliminary?) implementations for
1225    
1226      {String,Substring}.{concatWith,isSuffix,isSubstring}
1227    
1228    and
1229    
1230      Substring.full
1231    
1232    Those are in the Basis spec but they were missing in SML/NJ.
1233    
1234    ----------------------------------------------------------------------
1235    Name: Matthias Blume
1236    Date: 2002/03/14 21:30:00 EST
1237    Tag: blume-20020314-controls
1238    Description:
1239    
1240    Controls:
1241    ---------
1242    
1243    1. Factored out the recently-added Controls : CONTROLS stuff and put
1244       it into its own library $/controls-lib.cm.  The source tree for
1245       this is under src/smlnj-lib/Controls.
1246    
1247    2. Changed the names of types and functions in this interface, so they
1248       make a bit more "sense":
1249    
1250          module -> registry
1251          'a registry -> 'a group
1252    
1253    3. The interface now deals in ref cells only.  The getter/setter interface
1254       is (mostly) gone.
1255    
1256    4. Added a function that lets one register an already-existing ref cell.
1257    
1258    5. Made the corresponding modifications to the rest of the code so that
1259       everything compiles again.
1260    
1261    6. Changed the implementation of Controls.MLRISC back to something closer
1262       to the original.  In particular, this module (and therefore MLRISC)
1263       does not depend on Controls.  There now is some link-time code in
1264       int-sys.sml that registers the MLRISC controls with the Controls
1265       module.
1266    
1267    CM:
1268    ---
1269    
1270      * One can now specify the lambda-split aggressiveness in init.cmi.
1271    
1272    ----------------------------------------------------------------------
1273    Name: Allen Leung
1274    Date: 2002/03/13 17:30:00 EST
1275    Tag: leunga-20020313-x86-fp-unary
1276    Description:
1277    
1278    Bug fix for:
1279    
1280    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1281    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1282    > - fun f(x,(y,z)) = Real.~ y;
1283    > [autoloading]
1284    > [autoloading done]
1285    >       fchsl   (%eax), 184(%esp)
1286    > Error: MLRisc bug: X86MCEmitter.emitInstr
1287    >
1288    > uncaught exception Error
1289    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1290    
1291    The problem was that the code generator did not generate any fp registers
1292    in this case, and the ra didn't know that it needed to run the X86FP phase to
1293    translate the pseudo fp instruction.   This only happened with unary fp
1294    operators in certain situations.
1295    
1296    ----------------------------------------------------------------------
1297    Name: Matthias Blume
1298    Date: 2002/03/13 14:00:00 EST
1299    Tag: blume-20020313-overload-etc
1300    Description:
1301    
1302    1. Added _overload as a synonym for overload for backward compatibility.
1303       (Control.overloadKW must be true for either version to be accepted.)
1304    
1305    2. Fixed bug in install script that caused more things to be installed
1306       than what was requested in config/targets.
1307    
1308    3. Made CM aware of the (_)overload construct so that autoloading
1309       works.
1310    
1311    ----------------------------------------------------------------------
1312    Name: Matthias Blume
1313    Date: 2002/03/12 22:03:00 EST
1314    Tag: blume-20020312-url
1315    Description:
1316    
1317    Forgot to update BOOT and srcarchiveurl.
1318    
1319    ----------------------------------------------------------------------
1320    Name: Matthias Blume
1321    Date: 2002/03/12 17:30:00 EST
1322    Tag: blume-20020312-version110392
1323    Description:
1324    
1325    Yet another version number bump (because of small changes to the
1326    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1327    
1328    Changes:
1329    
1330      The new pid generation scheme described a few weeks ago was overly
1331      complicated.  I implemented a new mechanism that is simpler and
1332      provides a bit more "stability":  Once CM has seen a compilation
1333      unit, it keeps its identity constant (as long as you do not delete
1334      those crucial CM/GUID/* files).  This means that when you change
1335      an interface, compile, then go back to the old interface, and
1336      compile again, you arrive at the original pid.
1337    
1338      There now also is a mechanism that instructs CM to use the plain
1339      environment hash as a module's pid (effectively making its GUID
1340      the empty string).  For this, "noguid" must be specified as an
1341      option to the .sml file in question within its .cm file.
1342      This is most useful for code that is being generated by tools such
1343      as ml-nlffigen (because during development programmers tend to
1344      erase the tool's entire output directory tree including CM's cached
1345      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1346      revert to the old, broken behavior of SML/NJ, but in specific cases
1347      where there is no danger of interface confusion, its use is ok
1348      (I think).
1349    
1350      ml-nlffigen by default generates "noguid" annotations.  They can be
1351      turned off by specifying -guid in its command line.
1352    
1353    ----------------------------------------------------------------------
1354    Name: Lal George
1355    Date: 2002/03/12 12 14:42:36 EST
1356    Tag: george-20020312-frequency-computation
1357    Description:
1358    
1359    Integrated jump chaining and static block frequency into the
1360    compiler. More details and numbers later.
1361    
1362    ----------------------------------------------------------------------
1363    Name: Lal George
1364    Date: 2002/03/11 11 22:38:53 EST
1365    Tag: george-20020311-jump-chain-elim
1366    Description:
1367    
1368    Tested the jump chain elimination on all architectures (except the
1369    hppa).  This is on by default right now and is profitable for the
1370    alpha and x86, however, it may not be profitable for the sparc and ppc
1371    when compiling the compiler.
1372    
1373    The gc test will typically jump to a label at the end of the cluster,
1374    where there is another jump to an external cluster containing the actual
1375    code to invoke gc. This is to allow factoring of common gc invocation
1376    sequences. That is to say, we generate:
1377    
1378            f:
1379               testgc
1380               ja   L1      % jump if above to L1
1381    
1382            L1:
1383               jmp L2
1384    
1385    
1386    After jump chain elimination the 'ja L1' instructions is converted to
1387    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1388    up being implemented in their long form (if L2 is far away) using:
1389    
1390            jbe     L3      % jump if below or equal to L3
1391            jmp     L2
1392         L3:
1393            ...
1394    
1395    
1396    For large compilation units L2  may be far away.
1397    
1398    
1399    ----------------------------------------------------------------------
1400    Name: Matthias Blume
1401    Date: 2002/03/11 13:30:00 EST
1402    Tag: blume-20020311-mltreeeval
1403    Description:
1404    
1405    A functor parameter was missing.
1406    
1407    ----------------------------------------------------------------------
1408    Name: Allen Leung
1409    Date: 2002/03/11 10:30:00 EST
1410    Tag: leunga-20020311-runtime-string0
1411    Description:
1412    
1413       The representation of the empty string now points to a
1414    legal null terminated C string instead of unit.  It is now possible
1415    to convert an ML string into C string with InlineT.CharVector.getData.
1416    This compiles into one single machine instruction.
1417    
1418    ----------------------------------------------------------------------
1419    Name: Allen Leung
1420    Date: 2002/03/10 23:55:00 EST
1421    Tag: leunga-20020310-x86-call
1422    Description:
1423    
1424       Added machine generation for CALL instruction (relative displacement mode)
1425    
1426    ----------------------------------------------------------------------
1427    Name: Matthias Blume
1428    Date: 2002/03/08 16:05:00
1429    Tag: blume-20020308-entrypoints
1430    Description:
1431    
1432    Version number bumped to 110.39.1.  NEW BOOTFILES!
1433    
1434    Entrypoints: non-zero offset into a code object where execution should begin.
1435    
1436    - Added the notion of an entrypoint to CodeObj.
1437    - Added reading/writing of entrypoint info to Binfile.
1438    - Made runtime system bootloader aware of entrypoints.
1439    - Use the address of the label of the first function given to mlriscGen
1440      as the entrypoint.  This address is currently always 0, but it will
1441      not be 0 once we turn on block placement.
1442    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
1443      with entry points) from mlriscGen.
1444    
1445    ----------------------------------------------------------------------
1446    Name: Allen Leung
1447    Date: 2002/03/07 20:45:00 EST
1448    Tag: leunga-20020307-x86-cmov
1449    Description:
1450    
1451       Bug fixes for CMOVcc on x86.
1452    
1453       1. Added machine code generation for CMOVcc
1454       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1455       3. CMOVcc cannot have an immediate operand as argument.
1456    
1457    ----------------------------------------------------------------------
1458    Name: Matthias Blume
1459    Date: 2002/03/07 16:15:00 EST
1460    Tag: blume-20020307-controls
1461    Description:
1462    
1463    This is a very large but mostly boring patch which makes (almost)
1464    every tuneable compiler knob (i.e., pretty much everything under
1465    Control.* plus a few other things) configurable via both the command
1466    line and environment variables in the style CM did its configuration
1467    until now.
1468    
1469    Try starting sml with '-h' (or, if you are brave, '-H')
1470    
1471    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1472    implements the underlying generic mechanism.
1473    
1474    The interface to some of the existing such facilities has changed somewhat.
1475    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1476    (The getFoo interface is still there for backward-compatibility, but its
1477    use is deprecated.)
1478    
1479    The ml-build script passes -Cxxx=yyy command-line arguments through so
1480    that one can now twiddle the compiler settings when using this "batch"
1481    compiler.
1482    
1483    TODO items:
1484    
1485    We should go through and throw out all controls that are no longer
1486    connected to anything.  Moreover, we should go through and provide
1487    meaningful (and correct!) documentation strings for those controls
1488    that still are connected.
1489    
1490    Currently, multiple calls to Controls.new are accepted (only the first
1491    has any effect).  Eventually we should make sure that every control
1492    is being made (via Controls.new) exactly once.  Future access can then
1493    be done using Controls.acc.
1494    
1495    Finally, it would probably be a good idea to use the getter-setter
1496    interface to controls rather than ref cells.  For the time being, both
1497    styles are provided by the Controls module, but getter-setter pairs are
1498    better if thread-safety is of any concern because they can be wrapped.
1499    
1500    *****************************************
1501    
1502    One bug fix: The function blockPlacement in three of the MLRISC
1503    backpatch files used to be hard-wired to one of two possibilities at
1504    link time (according to the value of the placementFlag).  But (I
1505    think) it should rather sense the flag every time.
1506    
1507    *****************************************
1508    
1509    Other assorted changes (by other people who did not supply a HISTORY entry):
1510    
1511    1. the cross-module inliner now works much better (Monnier)
1512    2. representation of weights, frequencies, and probabilities in MLRISC
1513       changed in preparation of using those for weighted block placement
1514       (Reppy, George)
1515    
1516    ----------------------------------------------------------------------
1517    Name: Lal George
1518    Date: 2002/03/07 14:44:24 EST 2002
1519    Tag: george-20020307-weighted-block-placement
1520    
1521    Tested the weighted block placement optimization on all architectures
1522    (except the hppa) using AMPL to generate the block and edge frequencies.
1523    Changes were required in the machine properties to correctly
1524    categorize trap instructions. There is an MLRISC flag
1525    "weighted-block-placement" that can be used to enable weighted block
1526    placement, but this will be ineffective without block/edge
1527    frequencies (coming soon).
1528    
1529    
1530    ----------------------------------------------------------------------
1531    Name: Lal George
1532    Date: 2002/03/05 17:24:48 EST
1533    Tag: george-20020305-linkage-cluster
1534    
1535    In order to support the block placement optimization, a new cluster
1536    is generated as the very first cluster (called the linkage cluster).
1537    It contains a single jump to the 'real' entry point for the compilation
1538    unit. Block placement has no effect on the linkage cluster itself, but
1539    all the other clusters  have full freedom in the manner in which they
1540    reorder blocks or functions.
1541    
1542    On the x86 the typical linkage code that is generated is:
1543       ----------------------
1544            .align 2
1545       L0:
1546            addl    $L1-L0, 72(%esp)
1547            jmp     L1
1548    
1549    
1550            .align  2
1551       L1:
1552       ----------------------
1553    
1554    72(%esp) is the memory location for the stdlink register. This
1555    must contain the address of the CPS function being called. In the
1556    above example, it contains the address of  L0; before
1557    calling L1 (the real entry point for the compilation unit), it
1558    must contain the address for L1, and hence
1559    
1560            addl $L1-L0, 72(%esp)
1561    
1562    I have tested this on all architectures except the hppa.The increase
1563    in code size is of course negligible
1564    
1565    ----------------------------------------------------------------------
1566    Name: Allen Leung
1567    Date: 2002/03/03 13:20:00 EST
1568    Tag: leunga-20020303-mlrisc-tools
1569    
1570      Added #[ ... ] expressions to mlrisc tools
1571    
1572    ----------------------------------------------------------------------
1573    Name: Matthias Blume
1574    Date: 2002/02/27 12:29:00 EST
1575    Tag: blume-20020227-cdebug
1576    Description:
1577    
1578    - made types in structure C and C_Debug to be equal
1579    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1580    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1581    
1582    ----------------------------------------------------------------------
1583    Name: Matthias Blume
1584    Date: 2002/02/26 12:00:00 EST
1585    Tag: blume-20020226-ffi
1586    Description:
1587    
1588    1. Fixed a minor bug in CM's "noweb" tool:
1589       If numbering is turned off, then truly don't number (i.e., do not
1590       supply the -L option to noweb).  The previous behavior was to supply
1591       -L'' -- which caused noweb to use the "default" line numbering scheme.
1592       Thanks to Chris Richards for pointing this out (and supplying the fix).
1593    
1594    2. Once again, I reworked some aspects of the FFI:
1595    
1596       A. The incomplete/complete type business:
1597    
1598       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
1599         gone!
1600       - ML types representing an incomplete type are now *equal* to
1601         ML types representing their corresponding complete types (just like
1602         in C).  This is still safe because ml-nlffigen will not generate
1603         RTTI for incomplete types, nor will it generate functions that
1604         require access to such RTTI.   But when ML code generated from both
1605         incomplete and complete versions of the C type meet, the ML types
1606         are trivially interoperable.
1607    
1608         NOTE:  These changes restore the full generality of the translation
1609         (which was previously lost when I eliminated functorization)!
1610    
1611       B. Enum types:
1612    
1613       - Structure C now has a type constructor "enum" that is similar to
1614         how the "su" constructor works.  However, "enum" is not a phantom
1615         type because each "T enum" has values (and is isomorphic to
1616         MLRep.Signed.int).
1617       - There are generic access operations for enum objects (using
1618         MLRep.Signed.int).
1619       - ml-nlffigen will generate a structure E_foo for each "enum foo".
1620         * The structure contains the definition of type "mlrep" (the ML-side
1621         representation type of the enum).  Normally, mlrep is the same
1622         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
1623         then mlrep will be defined as a datatype -- thus facilitating
1624         pattern matching on mlrep values.
1625         ("-ec" will be suppressed if there are duplicate values in an
1626          enumeration.)
1627         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
1628         will be generated for each C enum constant xxx.
1629         * Conversion functions m2i and i2m convert between mlrep and
1630         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
1631         * Coversion functions c and ml convert between mlrep and "tag enum".
1632         * Access functions (get/set) fetch and store mlrep values.
1633       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
1634         enumerations are merged into one single enumeration represented by
1635         structure E_'.
1636    
1637    ----------------------------------------------------------------------
1638    Name: Allen Leung
1639    Date: 2002/02/25 04:45:00 EST
1640    Tag: leunga-20020225-cps-spill
1641    
1642    This is a new implementation of the CPS spill phase.
1643    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
1644    In case of problems, replace it with the old file spill.sml
1645    
1646    The current compiler runs into some serious performance problems when
1647    constructing a large record.  This can happen when we try to compile a
1648    structure with many items.  Even a very simple structure like the following
1649    makes the compiler slow down.
1650    
1651        structure Foo = struct
1652           val x_1 = 0w1 : Word32.int
1653           val x_2 = 0w2 : Word32.int
1654           val x_3 = 0w3 : Word32.int
1655           ...
1656           val x_N = 0wN : Word32.int
1657        end
1658    
1659    The following table shows the compile time, from N=1000 to N=4000,
1660    with the old compiler:
1661    
1662    N
1663    1000   CPS 100 spill                           0.04u  0.00s  0.00g
1664           MLRISC ra                               0.06u  0.00s  0.05g
1665              (spills = 0 reloads = 0)
1666           TOTAL                                   0.63u  0.07s  0.21g
1667    
1668    1100   CPS 100 spill                           8.25u  0.32s  0.64g
1669           MLRISC ra                               5.68u  0.59s  3.93g
1670              (spills = 0 reloads = 0)
1671           TOTAL                                   14.71u  0.99s  4.81g
1672    
1673    1500   CPS 100 spill                           58.55u  2.34s  1.74g
1674           MLRISC ra                               5.54u  0.65s  3.91g
1675              (spills = 543 reloads = 1082)
1676           TOTAL                                   65.40u  3.13s  6.00g
1677    
1678    2000   CPS 100 spill                           126.69u  4.84s  3.08g
1679           MLRISC ra                               0.80u  0.10s  0.55g
1680              (spills = 42 reloads = 84)
1681           TOTAL                                   129.42u  5.10s  4.13g
1682    
1683    3000   CPS 100 spill                           675.59u  19.03s  11.64g
1684           MLRISC ra                               2.69u  0.27s  1.38g
1685              (spills = 62 reloads = 124)
1686           TOTAL                                   682.48u  19.61s  13.99g
1687    
1688    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
1689           MLRISC ra                               4.96u  0.27s  2.72g
1690              (spills = 85 reloads = 170)
1691           TOTAL                                   2375.26u  57.21s  48.00g
1692    
1693    As you can see the old cps spill module suffers from some serious
1694    performance problem.  But since I cannot decipher the old code fully,
1695    instead of patching the problems up, I'm reimplementing it
1696    with a different algorithm.  The new code is more modular,
1697    smaller when compiled, and substantially faster
1698    (O(n log n) time and O(n) space).  Timing of the new spill module:
1699    
1700    4000  CPS 100 spill                           0.02u  0.00s  0.00g
1701          MLRISC ra                               0.25u  0.02s  0.15g
1702             (spills=1 reloads=3)
1703          TOTAL                                   7.74u  0.34s  1.62g
1704    
1705    Implementation details:
1706    
1707    As far as I can tell, the purpose of the CPS spill module is to make sure the
1708    number of live variables at any program point (the bandwidth)
1709    does not exceed a certain limit, which is determined by the
1710    size of the spill area.
1711    
1712    When the bandwidth is too large, we decrease the register pressure by
1713    packing live variables into spill records.  How we achieve this is
1714    completely different than what we did in the old code.
1715    
1716    First, there is something about the MLRiscGen code generator
1717    that we should be aware of:
1718    
1719    o MLRiscGen performs code motion!
1720    
1721       In particular, it will move floating point computations and
1722       address computations involving only the heap pointer to
1723       their use sites (if there is only a single use).
1724       What this means is that if we have a CPS record construction
1725       statement
1726    
1727           RECORD(k,vl,w,e)
1728    
1729       we should never count the new record address w as live if w
1730       has only one use (which is often the case).
1731    
1732       We should do something similar to floating point, but the transformation
1733       there is much more complex, so I won't deal with that.
1734    
1735    Secondly, there are now two new cps primops at our disposal:
1736    
1737     1. rawrecord of record_kind option
1738        This pure operator allocates some uninitialized storage from the heap.
1739        There are two forms:
1740    
1741         rawrecord NONE [INT n]  allocates a tagless record of length n
1742         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
1743                                     and initializes the tag.
1744    
1745     2. rawupdate of cty
1746          rawupdate cty (v,i,x)
1747          Assigns to x to the ith component of record v.
1748          The storelist is not updated.
1749    
1750    We use these new primops for both spilling and increment record construction.
1751    
1752     1. Spilling.
1753    
1754        This is implemented with a linear scan algorithm (but generalized
1755        to trees).  The algorithm will create a single spill record at the
1756        beginning of the cps function and use rawupdate to spill to it,
1757        and SELECT or SELp to reload from it.  So both spills and reloads
1758        are fine-grain operations.  In contrast, in the old algorithm
1759        "spills" have to be bundled together in records.
1760    
1761        Ideally, we should sink the spill record construction to where
1762        it is needed.  We can even split the spill record into multiple ones
1763        at the places where they are needed.  But CPS is not a good
1764        representation for global code motion, so I'll keep it simple and
1765        am not attempting this.
1766    
1767     2. Incremental record construction (aka record splitting).
1768    
1769        Long records with many component values which are simulatenously live
1770        (recall that single use record addresses are not considered to
1771         be live) are constructed with rawrecord and rawupdate.
1772        We allocate space on the heap with rawrecord first, then gradually
1773        fill it in with rawupdate.  This is the technique suggested to me
1774        by Matthias.
1775    
1776        Some restrictions on when this is applicable:
1777        1. It is not a VECTOR record.  The code generator currently does not handle
1778           this case. VECTOR record uses double indirection like arrays.
1779        2. All the record component values are defined in the same "basic block"
1780           as the record constructor.  This is to prevent speculative
1781           record construction.
1782    
1783    ----------------------------------------------------------------------
1784    Name: Allen Leung
1785    Date: 2002/02/22 01:02:00 EST
1786    Tag: leunga-20020222-mlrisc-tools
1787    
1788    Minor bug fixes in the parser and rewriter
1789    
1790    ----------------------------------------------------------------------
1791    Name: Allen Leung
1792    Date: 2002/02/21 20:20:00 EST
1793    Tag: leunga-20020221-peephole
1794    
1795    Regenerated the peephole files.  Some contained typos in the specification
1796    and some didn't compile because of pretty printing bugs in the old version
1797    of 'nowhere'.
1798    
1799    ----------------------------------------------------------------------
1800    Name: Allen Leung
1801    Date: 2002/02/19 20:20:00 EST
1802    Tag: leunga-20020219-mlrisc-tools
1803    Description:
1804    
1805       Minor bug fixes to the mlrisc-tools library:
1806    
1807       1.  Fixed up parsing colon suffixed keywords
1808       2.  Added the ability to shut the error messages up
1809       3.  Reimplemented the pretty printer and fixed up/improved
1810           the pretty printing of handle and -> types.
1811       4.  Fixed up generation of literal symbols in the nowhere tool.
1812       5.  Added some SML keywords to to sml.sty
1813    
1814    ----------------------------------------------------------------------
1815    Name: Matthias Blume
1816    Date: 2002/02/19 16:20:00 EST
1817    Tag: blume-20020219-cmffi
1818    Description:
1819    
1820    A wild mix of changes, some minor, some major:
1821    
1822    * All C FFI-related libraries are now anchored under $c:
1823        $/c.cm      --> $c/c.cm
1824        $/c-int.cm  --> $c/internals/c-int.cm
1825        $/memory.cm --> $c/memory/memory.cm
1826    
1827    * "make" tool (in CM) now treats its argument pathname slightly
1828      differently:
1829        1. If the native expansion is an absolute name, then before invoking
1830           the "make" command on it, CM will apply OS.Path.mkRelative
1831           (with relativeTo = OS.FileSys.getDir()) to it.
1832        2. The argument will be passed through to subsequent phases of CM
1833           processing without "going native".  In particular, if the argument
1834           was an anchored path, then "make" will not lose track of that anchor.
1835    
1836    * Compiler backends now "know" their respective C calling conventions
1837      instead of having to be told about it by ml-nlffigen.  This relieves
1838      ml-nlffigen from one of its burdens.
1839    
1840    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
1841    
1842    * Export C_DEBUG and C_Debug from $c/c.cm.
1843    
1844    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
1845      subtyping relationship between incomplete pointers and their complete
1846      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
1847      with the convention of instantiating 'o with ('t, 'c) obj whenever
1848      the pointer target type is complete.  In the incomplete case, 'o
1849      will be instantiated with some "'c iobj" -- a type obtained by
1850      using one of the functors PointerToIncompleteType or PointerToCompleteType.
1851    
1852      Operations that work on both incomplete and complete pointer types are
1853      typed as taking an 'o ptr while operations that require the target to
1854      be known are typed as taking some ('t, 'c) obj ptr.
1855    
1856      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
1857      where void is an eqtype without any values.  This makes it possible
1858      to work on voidptr values using functions meant to operate on light
1859      incomplete pointers.
1860    
1861    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
1862      been vastly simplified.
1863    
1864    ----------------------------------------------------------------------
1865    Name: Matthias Blume
1866    Date: 2002/02/19 10:48:00 EST
1867    Tag: blume-20020219-pqfix
1868    Description:
1869    
1870    Applied Chris Okasaki's bug fix for priority queues.
1871    
1872    ----------------------------------------------------------------------
1873    Name: Matthias Blume
1874    Date: 2002/02/15 17:05:00
1875    Tag: Release_110_39
1876    Description:
1877    
1878    Last-minute retagging is becoming a tradition... :-(
1879    
1880    This is the working release 110.39.
1881    
1882    ----------------------------------------------------------------------
1883    Name: Matthias Blume
1884    Date: 2002/02/15 16:00:00 EST
1885    Tag: Release_110_39-orig
1886    Description:
1887    
1888    Working release 110.39.  New bootfiles.
1889    
1890    (Update: There was a small bug in the installer so it wouldn't work
1891    with all shells.  So I retagged. -Matthias)
1892    
1893    ----------------------------------------------------------------------
1894    Name: Matthias Blume
1895    Date: 2002/02/15 14:17:00 EST
1896    Tag: blume-20020215-showbindings
1897    Description:
1898    
1899    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
1900    the latter can be useful for exploring what bindings are available at
1901    the interactive prompt.  (The first function returns only the list
1902    of symbols that are really bound, the second prints those but also the
1903    ones that CM's autoloading mechanism knows about.)
1904    
1905    ----------------------------------------------------------------------
1906    Name: Matthias Blume
1907    Date: 2002/02/15 12:08:00 EST
1908    Tag: blume-20020215-iptrs
1909    Description:
1910    
1911    Two improvements to ml-nlffigen:
1912    
1913      1. Write files only if they do not exist or if their current contents
1914         do not coincide with what's being written.  (That is, avoid messing
1915         with the time stamps unless absolutely necessary.)
1916    
1917      2. Implement a "repository" mechanism for generated files related
1918         to "incomplete pointer types".   See the README file for details.
1919    
1920    ----------------------------------------------------------------------
1921    Name: Matthias Blume
1922    Date: 2002/02/14 11:50:00 EST
1923    Tag: blume-20020214-quote
1924    Description:
1925    
1926    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
1927    because of the new and improved tag generation scheme.  (Thanks to Allen
1928    Leung for pointing it out.)
1929    
1930    ----------------------------------------------------------------------
1931    Name: Lal George
1932    Date: 2002/02/14 09:55:27 EST 2002
1933    Tag: george-20020214-isabelle-bug
1934    Description:
1935    
1936    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
1937    of Isabelle on the x86.
1938    
1939    From Allen:
1940    -----------
1941     I've found the problem:
1942    
1943         in ra-core.sml, I use the counter "blocked" to keep track of the
1944         true number of elements in the freeze queue.  When the counter goes
1945         to zero, I skip examining the queue.  But I've messed up the
1946         bookkeeping in combine():
1947    
1948             else ();
1949             case !ucol of
1950               PSEUDO => (if !cntv > 0 then
1951                     (if !cntu > 0 then blocked := !blocked - 1 else ();
1952                                        ^^^^^^^^^^^^^^^^^^^^^^^
1953                      moveu := mergeMoveList(!movev, !moveu)
1954                     )
1955                  else ();
1956    
1957         combine() is called to coalesce two nodes u and v.
1958         I think I was thinking that if the move counts of u and v are both
1959         greater than zero then after they are coalesced then one node is
1960         removed from the freeze queue.  Apparently I was thinking that
1961         both u and v are of low degree, but that's clearly not necessarily true.
1962    
1963    
1964    02/12/2002:
1965        Here's the patch.  HOL now compiles.
1966    
1967        I don't know how this impact on performance (compile
1968        time or runtime).  This bug caused the RA (especially on the x86)
1969        to go thru the potential spill phase when there are still nodes on the
1970        freeze queue.
1971    
1972    
1973    
1974    
1975    ----------------------------------------------------------------------
1976    Name: Matthias Blume
1977    Date: 2002/02/13 22:40:00 EST
1978    Tag: blume-20020213-fptr-rtti
1979    Description:
1980    
1981    Fixed a bug in ml-nlffigen that was introduced with one of the previous
1982    updates.
1983    
1984    ----------------------------------------------------------------------
1985    Name: Matthias Blume
1986    Date: 2002/02/13 16:41:00 EST
1987    Tag: blume-20020213-cmlpq
1988    Description:
1989    
1990    Added new priority queue export symbols (which have just been added to
1991    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
1992    not compile and the installer would choke.)
1993    
1994    ----------------------------------------------------------------------
1995    Name: Matthias Blume
1996    Date: 2002/02/13 16:15:00 EST
1997    Tag: blume-20020213-various
1998    Description:
1999    
2000    1. More tweaks to ml-nlffigen:
2001    
2002       - better internal datastructures (resulting in slight speedup)
2003       - "-match" option requires exact match
2004       - "localized" gensym counters (untagged structs/unions nested within
2005         other structs/unions or within typedefs get a fresh counter; their
2006         tag will be prefixed by a concatenation of their parents' tags)
2007       - bug fixes (related to calculation of transitive closure of types
2008         to be included in the output)
2009    
2010    2. Minor Basis updates:
2011    
2012       - added implementations for List.collate and Option.app
2013    
2014    ----------------------------------------------------------------------
2015    Name: Matthias Blume
2016    Date: 2002/02/11 15:55:00 EST
2017    Tag: blume-20020211-gensym
2018    Description:
2019    
2020    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2021    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2022    names (ML structure names that correspond to unnamed C structs, unions,
2023    and enums), so that separate runs of ml-nlffigen do not clash.
2024    
2025    ----------------------------------------------------------------------
2026    Name: Matthias Blume
2027    Date: 2002/02/11 12:05:00 EST
2028    Tag: blume-20020211-gensml
2029    Description:
2030    
2031    A quick fix for a problem with GenSML (in the pgraph-util library):
2032    Make generation of toplevel "local" optional.  (Strictly speaking,
2033    signature definitions within "local" are not legal SML.)
2034    
2035    Other than that: updates to INSTALL and cm/TODO.
2036    
2037    ----------------------------------------------------------------------
2038    Name: Matthias Blume
2039    Date: 2002/02/08 15:00:00 EST
2040    Tag: blume-20020208-uniquepid
2041    Description:
2042    
2043    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2044    
2045    1. The installer (config/install.sh) has gotten smarter:
2046    
2047         - Configuration options are a bit easier to specify now
2048           (in config/targets).
2049         - Bug in recognizing .tar.bz2 files fixed.
2050         - Installer automatically resolves dependencies between
2051           configuration options (e.g., if you ask for eXene, you will
2052           also get cml -- regardless whether you asked for it or not).
2053         - Installer can run in "quieter mode" by setting the environment
2054           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2055           "completely silent", though.
2056         - Build HashCons library as part of smlnj-lib.
2057    
2058    2. A new scheme for assigning persistent identifiers to compilation
2059       units (and, by extension, to types etc.) has been put into place.
2060       This fixes a long-standing bug where types and even dynamic values
2061       can get internally confused, thereby compromising type safety
2062       (abstraction) and dynamic correctness.  See
2063    
2064         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2065    
2066       for an example of how things could go wrong until now.
2067    
2068       The downside of the new scheme is that pids are not quite as
2069       persistent as they used to be: CM will generate a fresh pid
2070       for every compilation unit that it thinks it sees for the first
2071       time.  That means that if you compile starting from a clean, fresh
2072       source tree at two different times, you end up with different
2073       binaries.
2074    
2075       Cutoff recompilation, however, has not been compromised because
2076       CM keeps pid information in special caches between runs.
2077    
2078    ----------------------------------------------------------------------
2079    Name: Lal George
2080    Date: 2002/02/07 15:34:13 EST 2002
2081    Tag: <none>
2082    Description:
2083    
2084    Compilers that generate assembly code may produce  global labels
2085    whose value is resolved at link time. The various peephole optimization
2086    modules did not take this in account.
2087    
2088    TODO. The Labels.addrOf function should really return an option
2089    type so that clients are forced to deal with this issue, rather
2090    than an exception being raised.
2091    
2092    ----------------------------------------------------------------------
2093    Name: Lal George
2094    Date: 2002/02/06 13:55:02 EST
2095    Tag: george-20020206-ra-breakup
2096    Description:
2097    
2098    1. A bug fix from Allen.
2099    
2100        A typo causes extra fstp %st(0)'s to be generated at compensation
2101        edges, which might cause stack underflow traps at runtime.  This
2102        occurs in fft where there are extraneous fstps right before the 'into'
2103        trap instruction (in this case they are harmless since none of the
2104        integers overflow.)
2105    
2106    2. Pulled out various utility modules that were embedded in the modules
2107       of the register allocator. I need these modules for other purposes, but
2108       they are not complete enough to put into a library (just yet).
2109    ----------------------------------------------------------------------
2110  Name: Matthias Blume  Name: Matthias Blume
2111  Date: 2002/01/31 16:05:00 EST  Date: 2002/01/31 16:05:00 EST
2112  Tag: blume-20020131-sparc-ccalls  Tag: blume-20020131-sparc-ccalls
# Line 633  Line 2727 
2727    
2728  ----------------------------------------------------------------------  ----------------------------------------------------------------------
2729  Name: Matthias Blume  Name: Matthias Blume
 >>>>>>> 1.169  
2730  Date: 2001/09/18 15:35:00 EDT  Date: 2001/09/18 15:35:00 EDT
2731  Tag: blume-20010918-readme11036  Tag: blume-20010918-readme11036
2732  Description:  Description:

Legend:
Removed from v.1049  
changed lines
  Added in v.1389

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