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 1078, Tue Feb 19 21:26:48 2002 UTC revision 1399, Tue Sep 30 21:10:05 2003 UTC
# Line 13  Line 13 
13  Description:  Description:
14    
15  ----------------------------------------------------------------------  ----------------------------------------------------------------------
16    Name: Matthias Blume (blume (at) tti - c (dot) org)
17    Date: 2003/09/30 16:10:00 CDT
18    Tag: blume-20030930-primio-bat
19    Description:
20    
21    1. Added openVector, nullRd, and nullWr to PRIM_IO.
22    2. Improved .bat files (for Win32 port) to make things work under Win95.
23       (thanks to Aaron S. Hawley for this one)
24    
25    ----------------------------------------------------------------------
26    Name: Matthias Blume (blume (at) tti - c (dot) org)
27    Date: 2003/09/26 16:05:00 CDT
28    Tag: blume-20030926-wrappriv
29    Description:
30    
31    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
32    
33    ----------------------------------------------------------------------
34    Name: Matthias Blume (blume (at) tti - c (dot) org)
35    Date: 2003/09/26 15:00:00 CDT
36    Tag: blume-20030926-110_43_3
37    Description:
38    
39    - additional cleanup
40    - version number bump, NEW BOOTFILES
41    
42    ----------------------------------------------------------------------
43    Name: Matthias Blume (blume (at) tti - c (dot) org)
44    Date: 2003/09/26 12:00:00 CDT
45    Tag: blume-20030926-ppautoload
46    Description:
47    
48    I modified the read-eval-print loop so that the autoloader gets
49    invoked whenever the prettyprinter tries to look up a symbol that
50    is not currently defined in the toplevel environment but which
51    appears in CM's autoload registry.  As a result, we see far fewer of
52    those ?.Foo.Bar.xxx names in the prettyprinter's output.
53    
54    In addition to this I tried to clean up some pieces of the Basis
55    implementation (e.g., Socket, Word8Array) in order to prevent other
56    instances of these ?.Foo.Bar.xxx names from being printed.
57    
58    The mechanism that picks names for types still needs some work, though.
59    (Right now it seems that if there is a type A.t which is defined to
60    be B.u, but B is unavailable at toplevel, then A.t gets printed as
61    "?.B.u" although the perhaps more sensible solution would be to use
62    "A.t" in this case.  In other words, the prettyprinter should follow
63    a chain of DEFtycs not farther than there are corresponding toplevel
64    names in the current environment.)
65    
66    ----------------------------------------------------------------------
67    Name: Matthias Blume (blume (at) tti - c (dot) org)
68    Date: 2003/09/24 16:31:00 CDT
69    Tag: blume-20030924-installer
70    Description:
71    
72    Another installer tweak:  All the ML code for the installer is now
73    compiled during CMB.make and put into a little library called
74    $smlnj/installer.cm.  The installation then simply invokes
75    
76       sml -m $smlnj/installer.cm
77    
78    and everything happens automagically.
79    
80    Win32:  ML code senses value of environment variable SMLNJ_HOME.
81    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
82            and BINDIR.
83    
84    The new scheme guarantees that the ML code responsible for the installation
85    is in sync with the APIs of the main system.  Also, the installer is
86    somewhat faster because the installer script is precompiled.
87    
88    ----------------------------------------------------------------------
89    Name: Matthias Blume (blume (at) tti - c (dot) org)
90    Date: 2003/09/24 15:35:00 CDT
91    Tag: blume-20030924-synsock
92    Description:
93    
94    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
95    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
96    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
97    implementations of non-blocking operations from CML's Socket
98    structure.
99    
100    ----------------------------------------------------------------------
101    Name: Matthias Blume (blume (at) tti - c (dot) org)
102    Date: 2003/09/24 15:10:05 CDT
103    Tag: blume-20030924-sockets
104    Description:
105    
106    1. Fixed SOCKET API and implementation to match Basis spec.
107       This required changing the internal representation of sockets to one
108       that remembers (for each socket file descriptor) whether it is currently
109       blocking or non-blocking.  This state is maintained lazily (i.e., a system
110       call is made only if the state actually needs to change).
111    
112    2. OS-specific details of sockets were moved into separate files, thus
113       making it possible to unify the bulk of the socket implementations
114       between Unix and Win32.
115    
116    3. CML's socket API changed accordingly.
117       (Note that we need to remove non-blocking functions from this API
118       since they are redundant in the case of CML!)
119    
120    4. CML's socket implementation now makes use of non-blocking functions
121       provided by Basis, thus removing all OS-dependent code from this part
122       of CML.
123    
124    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
125    
126    ----------------------------------------------------------------------
127    Name: Matthias Blume (blume (at) tti - c (dot) org)
128    Date: 2003/09/22 12:10:00 CDT
129    Tag: blume-20030922-110_43_2
130    Description:
131    
132    Made a new interim version and bootfiles for developer's bootstrapping
133    convenience.
134    
135       110.43.2 -- NEW BOOTFILES
136    
137    ----------------------------------------------------------------------
138    Name: Matthias Blume (blume (at) tti - c (dot) org)
139    Date: 2003/09/19 15:55:00 CDT
140    Tag: blume-20030919-cmdir
141    Description:
142    
143    1. new-install.sh -> install.sh
144    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
145    3. tweaked installer so that another name instead of .cm can be chosen
146       at install time (by setting the CM_DIR_ARC environment variable
147       during installation); once installation is complete, the name is
148       fixed
149    
150    ----------------------------------------------------------------------
151    Name: Matthias Blume (blume (at) tti - c (dot) org)
152    Date: 2003/09/18 16:00:00 CDT
153    Tag: blume-20030918-110_43_1
154    Description:
155    
156    Made a new interim version and bootfiles for developer's bootstrapping
157    convenience.
158    
159       110.43.1 -- NEW BOOTFILES
160    
161    ----------------------------------------------------------------------
162    Name: Matthias Blume (blume (at) tti - c (dot) org)
163    Date: 2003/09/18 15:20:00 CDT
164    Tag: blume-20030918-misc
165    Description:
166    
167    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
168       this seems to be controversial at the moment) and used those in
169       Posix.ProcEnv.times.
170    
171    2. Added Time.{from,to}Nanoseconds to Time.
172    
173    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
174       For example, fromLargeInt never needs to look at more than 3 "big
175       digits" to get its 53 bits of precision.
176    
177    ----------------------------------------------------------------------
178    Name: Matthias Blume (blume (at) tti - c (dot) org)
179    Date: 2003/09/17 16:30:00 CDT
180    Tag: blume-20030917-real32-slices
181    Description:
182    
183    Added an entry to the primitive environment
184    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
185    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
186    
187    Removed some of the "magic" constants in real64.sml and replaced them
188    with code that generates these values from their corresponding
189    integer counterparts.
190    
191    Made all(?) the slice-related changes to the Basis and made everything
192    compile again...
193    
194    ----------------------------------------------------------------------
195    Name: Matthias Blume (blume (at) tti - c (dot) org)
196    Date: 2003/09/15 17:45:00 CDT
197    Tag: blume-20030915-rbase
198    Description:
199    
200    Fixed bug in Real.fromLargeInt.
201    
202    ----------------------------------------------------------------------
203    Name: Matthias Blume (blume (at) tti - c (dot) org)
204    Date: 2003/09/13 18:11:00 CDT
205    Tag: blume-20030913-libinstall
206    Description:
207    
208    Minor bugfix in config/libinstall (set anchor with path to
209    standalone tool after installing it, otherwise libraries that
210    need ml-lex or ml-yacc won't compile the first time the installer
211    runs).
212    
213    ----------------------------------------------------------------------
214    Name: Matthias Blume (blume (at) tti - c (dot) org)
215    Date: 2003/09/12 11:45:00 CDT
216    Tag: blume-20030912-various
217    Description:
218    
219     - fixed bug in Real.toLargeInt
220     - fixed bug in Posix.ProcEnv.times
221     - changed inputLine functions to return an option
222     - minor installer improvements / bugfixes
223     - changed default @SMLalloc parameter for x86/celeron to 64k
224    
225    ----------------------------------------------------------------------
226    Name: Matthias Blume (blume (at) tti - c (dot) org)
227    Date: 2003/09/09 22:00:00 CDT
228    Tag: Release_110_43
229    Description:
230    
231        New working release 110.43.  New bootfiles.
232    
233    ----------------------------------------------------------------------
234    Name: Matthias Blume (blume (at) tti - c (dot) org)
235    Date: 2003/09/09 19:20:00 CDT
236    Tag: blume-20030909-installer
237    Description:
238    
239    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
240    Modified config/install.bat to take advantage of it.  Also modified
241    config/install.sh (and called it config/new-install.sh) to take advantage
242    of it on Unix systems.  (The SML code is (supposed to be) platform-
243    independent.)
244    
245    The installer can now install everything under Win32
246    as well as under *nix as long as it compiles.
247    
248    Other changes:
249    
250      - made CML compile again under Win32
251      - made eXene compile under Win32 (by providing a fake structure UnixSock
252        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
253      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
254        same as type int;  under Win32 it isn't
255      - fixed some slice-related problems in the win32-specific parts of CML
256      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
257        (under Win32, the volume name is case-insensitive, and the
258        OS.Path code compares volume names for equality)
259    
260    ----------------------------------------------------------------------
261    Name: Matthias Blume (blume (at) tti - c (dot) org)
262    Date: 2003/09/08 11:55:00 CDT
263    Tag: blume-20030908-fullpath
264    Description:
265    
266    Made Win32 version of OS.FileSys.fullPath return current directory
267    when given an empty string.  This is what the spec says, and incidentally,
268    CM depends on it.  (CM otherwise goes into an infinite loop in certain
269    cases when presented with the name of a non-existing .cm file.)
270    
271    ----------------------------------------------------------------------
272    Name: Matthias Blume (blume (at) tti - c (dot) org)
273    Date: 2003/09/04 16:30:00 CDT
274    Tag: blume-20030905-slices-etc
275    Description:
276    
277      1. Changed interface to vectors and arrays in Basis to match
278         (draft) Basis spec.
279      2. Added signatures and implementations of slices according to
280         Basis spec.
281      3. Edited source code throughout the system to make it compile again
282         under 1. and 2.  (In some cases code had to be added to have it
283         match the new signatures.)
284      4. MLRISC should be backward-compatible: the copies of the originals
285         of files that needed to change under 3. were retained, the .cm files
286         check the compiler version number and use old versions when
287         appropriate.
288      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
289         dirstream -> string option (in accordance with Basis spec).
290      6. When generating code that counts lines, ml-lex used function
291         CharVector.foldli, taking advantage of its old interface.
292         This has been replaced with the corresponding code from
293         CharVectorSlice.  (html-lex must be re-lexed!)
294      7. BitArray in smlnj-lib/Util has been extended/modified to match the
295         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
296      8. Removed temporary additions (fromInternal, toInternal) from the
297         (now obsolete) IntInf in smlnj-lib/Util.
298      9. Cleaned up structure Byte.
299     10. Added localOffset, scan, and fromString to Date (according to spec).
300         Cleaned/corrected implementation of Date.
301         (Still need to check for correctness; implement better canonicalizeDate.)
302     11. Added "scan" to signature IEEE_REAL.
303     12. Some improvements to IntInf [in particular: efficiency-hack for
304         mod and rem when second operand is 2 (for parity checks).]
305     13. Changed representation of type Time.time, using a single IntInf.int
306         value counting microseconds.  This considerably simplified the
307         implementation of structure Time.  We now support negative time
308         values; scan and fromString handle signs.
309     14. Functor PrimIO now takes two additional arguments (VectorSlice and
310         ArraySlice).
311    
312    ----------------------------------------------------------------------
313    Name: Matthias Blume (blume (at) tti - c (dot) org)
314    Date: 2003/08/28 17:00:00 CDT
315    Tag: blume-20030828-intinf
316    Description:
317    
318    This is a major update which comes with a version number bump
319    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
320    and an implementation of IntInf in the Basis.
321    
322    There are a fairly large number of related changes and updates throughout
323    the system:
324    
325    Basis:
326            - Implemented IntInf.
327            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
328            - Added some missing Real64 operations, most notably Real.toLargeInt.
329            - Added FixedInt as a synonym for Int32.
330    
331    compiler:
332            * Added support for a built-in intinf type.
333              - literals
334              - pattern matching
335              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
336              - overloading on literals and operations
337    
338            This required adding a primitive type intinf, some additional
339            primops, and implementations for several non-trivial intinf
340            operations in Core.  (The intinf type is completely abstract
341            to the compiler; all operations get delegated back to the Core.)
342    
343            * Intinf equality is handled by polyequal.  However, the compiler
344            does not print its usual warning in this case (since polyequal
345            is the right thing to do there).
346    
347            * Improved the organization of structure InlineT.
348    
349            * A word about conversion primops:
350            If conversions involving intinf do not cancel out during
351            CPS contract, then the compiler must insert calls to Core functions.
352            Since all core access must be resolved already during the FLINT
353            translate phase, it would be too late a the time of CPS contract
354            to add new Core calls.  For this reason, conversion primops
355            for intinf carry two arguments:  1. the numeric argument that
356            they are supposed to convert, and 2. the Core function that
357            can help with this conversion if necessary.  If CPS contract
358            eliminates a primop, then the associated Core function becomes
359            dead and goes away.  Intinf conversion primops that do not get
360            eliminated by CPS contract get rewritten into calls of their
361            core functions by a separate, new phase.
362    
363    interactive system:
364            - Control.Print.intinfDepth controls max length of intinf constants
365            being printed. (Analogous to Control.Print.stringDepth.)
366            - Cleanup in printutil and pputil: got rid of unused stuff and
367            duplicates; replaced some of the code with code that makes better
368            use of library functionality.
369    
370    CM:
371            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
372            are no longer silent.
373    
374    CKIT:
375            Fixed mismatched uses of Int32 and LargeInt.  I always decided
376            in favor of LargeInt -- which is now the same as IntInf.
377            CKIT-knowledgable people should check whether this is what's
378            intended and otherwise change things back to using Int32 or
379            FixedInt.
380    
381    Throughout the code:
382            Started using IntInf.int literals and built-in operations
383            (e.g., comparison with 0) where this seems appropriate.
384    
385    
386    ----------------------------------------------------------------------
387    Name: Dave MacQueen (dbm@cs.uchicago.edu)
388    Date: 2003/08/13 11:36:00 CDT
389    Tag: dbm-20030813-mcz-merge1
390    Description:
391    
392    Merging changes from the mcz-branch development branch into trunk.
393    These changes involve replacement of the emulated old prettyprinter
394    interface with direct use of the SML/NJ Lib PP library, and fixing
395    of a couple of bugs (895, 1186) relating to error messages.  A new
396    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
397    has been added.
398    
399    ----------------------------------------------------------------------
400    Name: Matthias Blume (blume (at) tti - c (dot) org)
401    Date: 2003/08/11 15:45:00 CDT
402    Tag: blume-20030811-windows
403    Description:
404    
405    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
406    
407        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
408    
409    This patch restores SML/NJ's ability to run under win32.  There are a
410    number of changes, including fixes for several bugs that had gone
411    unnoticed until now:
412    
413      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
414        trying to run the win32 version from within cygwin.)
415    
416      - There are a number of simple .bat scripts that substitute for their
417        corresponding Unix shell-scripts.  (See below.)
418    
419      - The internals of ml-build have been modified slightly.  The main
420        difference is that instead of calling ".link-sml" (or link-sml.bat)
421        using OS.Process.system, the ML process delegates this task back
422        to the script.  Otherwise problems arise in mixed environments such
423        as Cygwin where scripts look and work like Unix scripts, but
424        where OS.Process.system cannot run them.
425    
426      - In CM, the srcpath pickler used native pathname syntax -- which
427        is incorrect in the case of cross-compilation.  The new pickle format
428        is independent of platform-specific naming conventions.
429    
430      - Path configuration files (such as lib/pathconfig) can now choose
431        between native and standard syntax.  Placing a line of the form
432    
433            standard!
434    
435        into the file causes all subsequent paths to be interpreted using
436        CM standard pathname syntax (= Unix conventions); a line
437    
438            native!
439    
440        switches back to native style.  This was needed so that
441        path config files can be written portably, see src/system/pathconfig.
442    
443      - Runtime system:
444    
445          - win32-filesys.c:  get_file_time and set_file_time now
446            access modification time, not creation time.
447    
448          - I/O code made aware of new array representation.
449    
450          - Bug fixes in X86.prim.masm.
451    
452       - src/system/makeml made aware of win32. (For use under cygwin
453            and other Unix-environments for windows.)
454    
455       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
456         which caused BinIO.inputAll to fail consistently.
457    
458    .bat scripts:
459    
460       Windows .bat scripts assume that SMLNJ_HOME is defined.
461    
462       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
463         applications (sml, ml-yacc, ml-lex).
464       - ml-build.bat: analogous to ml-build.
465       - config\install.bat:  Analogous to config/install.sh.  This requires
466         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
467         (nmake etc. must be on the path, and vcvars32 must have been run.)
468         Moreover, sources for ml-lex and ml-yacc need to exist under src,
469         and the bootfile hierarchy must have been unpacked under
470         sml.boot.x86-win32.
471         The script is very primitive and does a poor job at error checking.
472         It only installs the base system, ml-lex, and ml-yacc.  No other
473         libraries are being installed (i.e., you get only those that
474         are part of the compiler.)
475       - link-sml.bat:  analogous to .link-sml, but not currently used
476    
477    Unrelated bug fixes:
478    
479      - ml-nlffigen now exports structures ST_* corresponding to incomplete
480        types.
481      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
482        otherwise.)
483    
484    ----------------------------------------------------------------------
485    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
486    Date: 2003/06/17
487    Tag: macqueen-20030617-bug895
488    Description:
489    
490    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
491    Tag will be used for new development branch (mcz-branch) for
492    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
493    
494    ----------------------------------------------------------------------
495    Name: Matthias Blume (blume (at) tti - c (dot) org)
496    Date: 2003/05/27 16:55:00 CDT
497    Tag: blume-20030527-polyeq
498    Description:
499    
500    Tried to eliminated most cases of polymorphic equality.
501    
502    ----------------------------------------------------------------------
503    Name: Matthias Blume (blume (at) tti - c (dot) org)
504    Date: 2003/05/21 17:45:00 CDT
505    Tag: blume-20030517-complete
506    Description:
507    
508    Two changes:
509    
510      1. Added a flag for controlling whether non-exhaustive bindings will
511         be treated as errors (default is false).
512      2. Cleaned up the *entire* source tree so that CMB.make goes through
513         without a single non-exhaustive match- or bind warning.
514    
515    ----------------------------------------------------------------------
516    Name: Matthias Blume (blume (at) tti - c (dot) org)
517    Date: 2003/05/17 10:20:00 CDT
518    Tag: blume-20030517-absyn
519    Description:
520    
521    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
522    
523       This mainly affects the quality of error messages.  However, some
524       of the code is now more straightforward than before. (Treatment of
525       the above four constructs in translate.sml is much simpler than
526       the "macro-expansion" that was going on before.  Plus, the mach-
527       compiler no longer gets invoked just to be able to compile an
528       if-expression.)
529    
530    2. The ErrorMsg.Error exception is now caught and absorbed by the
531       interactive loop.
532    
533    ----------------------------------------------------------------------
534    Name: Allen Leung
535    Date: 2003/05/16 13:05:00 CDT
536    Tag: leunga-20030516-cygwin-runtime
537    Description:
538    
539        Ported the runtime system to cygwin, which uses the unix
540    x86-unix bin files.  Missing/buggy features:
541    
542       o getnetbyname, getnetbyaddr: these functions seem to be missing in
543         the Cygwin library.
544       o Ctrl-C handling may be flaky.
545       o Windows system calls and Windows I/O are not supported.
546    
547       A new set of binfiles is located at:
548    
549         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
550    
551       This is only needed for bootstrapping the cygwin version of smlnj.
552       Other x86 versions can use the existing binfiles.
553    
554    ----------------------------------------------------------------------
555    Name: Matthias Blume
556    Date: 2003/04/08 15:42:00 CDT
557    Tag: blume-20030408-listpair
558    Description:
559    
560    1. Added a target 'mlrisc' to installer.
561    
562    2. Added missing elements to structure ListPair.
563    
564    ----------------------------------------------------------------------
565    Name: Allen Leung
566    Date: 2003/01/07 10:40:00 EST
567    Tag: leunga-20030107-int-rem
568    Description:
569    
570    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
571    The arguments to the SUBL instruction were swapped.
572    
573    ----------------------------------------------------------------------
574    Name: Matthias Blume
575    Date: 2002/12/12 16:25:00 EST
576    Tag: blume-20021212-risc-ra
577    Description:
578    
579    Fixed a serious bug in the rewrite code for FP spilling/reloading that
580    sent the RA into an infinite loop when floating point registers get
581    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
582    110.37 and 110.38.)
583    There was another set of potential problems related to the handling of
584    MLRISC annotations (but those did not yet cause real problems, apparently).
585    
586    ----------------------------------------------------------------------
587    Name: Matthias Blume
588    Date: 2002/12/06 22:40:00 EST
589    Tag: blume-20021206-cm-fileid
590    Description:
591    
592    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
593    This fixes the problem of CM getting confused by files that suddenly
594    change their identity (e.g., by getting unlinked and recreated by some
595    text editor such as vi).  There might be a better/cheaper/cleaner way
596    of doing this, but for now this will have to do.
597    
598    ----------------------------------------------------------------------
599    Name: Matthias Blume
600    Date: 2002/10/28 09:50:00 EST
601    Tag: blume-20021028-typecheck
602    Description:
603    
604    Exported structure Typecheck from $smlnj/viscomp/core.cm.
605    
606    ----------------------------------------------------------------------
607    Name: Matthias Blume
608    Date: 2002/10/17 09:10:00 EDT
609    Tag: Release_110_42
610    Description:
611    
612    In good old tradition, there has been a slight hiccup so that we have
613    to patch 110.42 after the fact.   The old release tag has been replaced
614    (see below).
615    
616    The change solves a problem with two competing approaches the
617    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
618    each other's way.
619    
620    This change only affects the runtime system code and the installer script.
621    (No new bootfiles.)
622    
623    ----------------------------------------------------------------------
624    Name: Matthias Blume
625    Date: 2002/10/16 12:00:00 EDT
626    Tag: Release_110_42_removed
627    Description:
628    
629    New working release.  New bootfiles.
630    
631    ----------------------------------------------------------------------
632    Name: Matthias Blume
633    Date: 2002/10/10 13:10:00 EDT
634    Tag: blume-20021010-ppc-divs
635    Description:
636    
637    The mltree operator DIVS must be implemented with an overflow check on
638    the PPC because the hardware indicates divide-by-zero using "overflow" as
639    well.
640    
641    ----------------------------------------------------------------------
642    Name: Matthias Blume
643    Date: 2002/07/23 11:45:00
644    Tag: blume-20020723-smlnj-home
645    Description:
646    
647    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
648    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
649    CM_PATHCONFIG is also set), the path configuration file is assumed
650    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
651    the entire tree to some other place and everything will "just work".
652    
653    (Companion commands such as ml-build and ml-makedepend also sense this
654    variable.)
655    
656    ----------------------------------------------------------------------
657    Name: Matthias Blume
658    Date: 2002/07/12 21:19:00 EDT
659    Tag: blume-20020712-liveness
660    Description:
661    
662    Exported two useful "step" functions from liveness module (MLRISC).
663    
664    ----------------------------------------------------------------------
665    Name: Matthias Blume
666    Date: 2002/07/05 16:00 EDT
667    Tag: Release_110_41
668    Description:
669    
670    New working release.  New bootfiles.
671    
672    ----------------------------------------------------------------------
673    Name: Matthias Blume
674    Date: 2002/07/05 10:25:00 EDT
675    Tag: blume-20020705-btimp
676    Description:
677    
678    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
679    other clients can set up backtracing support.
680    
681    ----------------------------------------------------------------------
682    Name: Matthias Blume
683    Date: 2002/06/25 17:23:00 EDT
684    Tag: blume-20020625-fpmax
685    Description:
686    
687    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
688    case.  (The sense of the isNaN test was reversed -- which made min and
689    max always return their first argument.)
690    
691    ----------------------------------------------------------------------
692    Name: Matthias Blume
693    Date: 2002/06/11
694    Tag: blume-20020611-unixpath
695    Description:
696    
697    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
698    
699    ----------------------------------------------------------------------
700    Name: Matthias Blume
701    Date: 2002/06/10 16:35:00 EDT
702    Tag: blume-20020610-ieeereal
703    Description:
704    
705    I back-ported my implementation of IEEEReal.fromString from the
706    idlbasis-devel branch so that we can test it.
707    
708    Another small change is that ppDec tries to give more information
709    than just "<sig>" in the case of functors.  However, this code is
710    broken in some mysterious way if the functor's body's signature
711    has not been declared by ascription but gets inferred from the
712    implementation.  This needs fixing...
713    
714    ----------------------------------------------------------------------
715    Name: Matthias Blume
716    Date: 2002/05/31
717    Tag: blume-20020531-btrace-mode
718    Description:
719    
720    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
721    the wayside when I switched over to using Controls everywhere.)
722    
723    ----------------------------------------------------------------------
724    Name: Lal George
725    Date: 2002/05/23 12:21:40 EDT
726    Tag: george-20020523-visual-labels
727    Description:
728    
729    Labels are now displayed in the graphical output to make
730    the fall-through and target blocks obvious.
731    
732    ----------------------------------------------------------------------
733    Name: Matthias Blume
734    Date: 2002/05/22 11:03:00 EDT
735    Tag: blume-20020522-shrink
736    Description:
737    
738    John tweaked yesterday's fix for 1131 to handle an out-of-memory
739    situation that comes up when allocating huge arrays.
740    
741    ----------------------------------------------------------------------
742    Name: Matthias Blume
743    Date: 2002/05/21 16:00:00 EDT
744    Tag: Release_110_40
745    Description:
746    
747    New working release (110.40).  New bootfiles.
748    
749    [Also: John Reppy fixed GC bug 1131.]
750    
751    ----------------------------------------------------------------------
752    Name: Matthias Blume
753    Date: 2002/05/21 12:35:00 EDT
754    Tag: blume-20020521-cmdoc
755    Description:
756    
757    CM documentation update.
758    
759    ----------------------------------------------------------------------
760    Name: Matthias Blume
761    Date: 2002/05/21 10:55:00 EDT
762    Tag: blume-20020521-misc
763    Description:
764    
765    - John tweaked runtime to be silent on heap export (except when
766      GC messages are on).
767    - I added a few more things (cross-compiling versions of CMB) to
768      config/preloads (as suggestions).
769    
770    ----------------------------------------------------------------------
771    Name: Matthias Blume
772    Date: 2002/05/20 22:25:00 EDT
773    Tag: blume-20020520-controls
774    Description:
775    
776    - Added ControlUtil structure to control-lib.cm.
777    - Use it throughout.
778    - Used Controls facility to define MLRISC controls (as opposed to
779      registering MLRISC control ref cells with Controls after the fact)
780    - Fixed messed-up controls priorities.
781    
782    * Removed again all the stuff from config/preloads that one wouldn't
783      be able to preload at the time the initial heap image is built.
784      (Many libraries, e.g., CML, do not exist yet at this time.  The
785       only libraries that can be preloaded via config/preloads are those
786       that come bundled with the bootfiles.)
787    
788    ----------------------------------------------------------------------
789    Name: Matthias Blume
790    Date: 2002/05/20 10:59:00 EDT
791    Tag: blume-20020520-preloads
792    Description:
793    
794    Added a lot of commented-out suggestions for things to be included
795    in config/preloads.
796    
797    ----------------------------------------------------------------------
798    Name: Allen Leung
799    Date: 2002/05/18 14:20:00 EDT
800    Tag: leunga-20020518-mdl
801    Description:
802    
803    o Made the mdl tool stuff compile and run again.
804    o I've disabled all the stuff that depends on RTL specifications; they
805      are all badly broken anyway.
806    
807    ----------------------------------------------------------------------
808    Name: Matthias Blume
809    Date: 2002/05/17 16:49:00 EDT
810    Tag: blume-20020517-controls
811    Description:
812    
813    0. John Reppy made several modifications to the SML/NJ library.
814       In particular, there is a shiny new controls-lib.cm.
815    
816    1. Pushed new controls interface through compiler so that everything
817       compiles again.
818    
819    2. Added FormatComb and FORMAT_COMB to the CML version of the
820       SML/NJ library (so that CML compiles again).
821    
822    3. Modified init scripts because XXX_DEFAULT environment variables
823       are no longer with us.  (Boot-time initialization is now done
824       using the same environment variables that are also used for
825       startup-time initialization of controls.)
826    
827    ----------------------------------------------------------------------
828    Name: Lal George
829    Date: 2002/05/15 09:20:10 EDT
830    Tag: george-20020515-pseudo-op-decls
831    Description:
832    
833            All pseudo-ops emitted before the first segment declaration
834    such as TEXT, DATA, and BSS directives are assumed to be global
835    declarations and are emitted first in the assembly file. This is
836    useful in a number of situations where one has pseudo-ops that are not
837    specific to any segment, and also works around the constraint that one
838    cannot have client pseudo-ops in the TEXT segment.
839    
840            Because no segment is associated with these declarations it is
841    an error to allocate any space or objects before the first segment
842    directive and an exception will be raised. However, we cannot make
843    this check for client pseudo-ops.
844    
845            These top level declarations are a field in the CFG graph_info.
846    In theory you can continue to add to this field after the CFG has been
847    built -- provided you know what you are doing;-)
848    
849    ----------------------------------------------------------------------
850    Name: Matthias Blume
851    Date: 2002/05/13 16:40:00 EDT
852    Tag: blume-20020513-pp-etc
853    Description:
854    
855    A few minor bugfixes:
856    
857      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
858        (Bogus printouts for redefined bindings still occur.  Compiler
859        bug should no longer occur now.  We need to redo the prettyprinter
860        from scratch.)
861    
862      - CM pathname printer now also adds escape sequences for ( and )
863    
864      - commend and docu fixes for ml-nlffi
865    
866    ----------------------------------------------------------------------
867    Name: Matthias Blume
868    Date: 2002/05/10 16:40:00 EDT
869    Tag: blume-20020510-erg-textio
870    Description:
871    
872    Applied the following bugfix provided by Emden Gansner:
873    
874        Output is corrupted when outputSubstr is used rather than output.
875    
876        The problem occurs when a substring
877    
878            ss = (s, dataStart, dataLen)
879    
880        where dataStart > 0, fills a stream buffer with avail bytes left.
881        avail bytes of s, starting at index dataStart, are copied into the
882        buffer, the buffer is flushed, and then the remaining dataLen-avail
883        bytes of ss are copied into the beginning of the buffer. Instead of
884        starting this copy at index dataStart+avail in s, the current code
885        starts the copy at index avail.
886    
887        Fix:
888        In text-io-fn.sml, change line 695 from
889             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
890        to
891             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
892    
893    ----------------------------------------------------------------------
894    Name: Matthias Blume
895    Date: 2002/04/12 13:55:00 EDT
896    Tag: blume-20020412-assyntax
897    Description:
898    
899    1. Grabbed newer assyntax.h from the XFree86 project.
900    2. Fiddled with how to compile X86.prim.asm without warnings.
901    3. (Very) Minor cleanup in CM.
902    
903    ----------------------------------------------------------------------
904    Name: Matthias Blume
905    Date: 2002/04/01 (no joke!) 17:07:00 EST
906    Tag: blume-20020401-x86div
907    Description:
908    
909    Added full support for div/mod/rem/quot on the x86, using the machine
910    instruction's two results (without clumsily recomputing the remainder)
911    directly where appropriate.
912    
913    Some more extensive power-of-two support was added to the x86 instruction
914    selector (avoiding expensive divs, mods, and muls where they can be
915    replaced with cheaper shifts and masks).  However, this sort of thing
916    ought to be done earlier, e.g., within the CPS optimizer so that
917    all architectures benefit from it.
918    
919    The compiler compiles to a fixed point, but changes might be somewhat
920    fragile nevertheless.  Please, report any strange things that you might
921    see wrt. div/mod/quot/rem...
922    
923    ----------------------------------------------------------------------
924    Name: Matthias Blume
925    Date: 2002/03/29 17:22:00
926    Tag: blume-20020329-div
927    Description:
928    
929    Fixed my broken div/mod logic.  Unfortunately, this means that the
930    inline code for div/mod now has one more comparison than before.
931    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
932    The problem was with quotient = 0, because that alone does not tell
933    us which way the rounding went.  One then has to look at whether
934    remainder and divisor have the same sign...  :(
935    
936    Anyway, I replaced the bootfiles with fresh ones...
937    
938    ----------------------------------------------------------------------
939    Name: Matthias Blume
940    Date: 2002/03/29 14:10:00 EST
941    Tag: blume-20020329-inlprims
942    Description:
943    
944    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
945    
946    Primops have changed. This means that the bin/boot-file formats have
947    changed as well.
948    
949    To make sure that there is no confusion, I made a new version.
950    
951    
952    CHANGES:
953    
954    * removed REMT from mltree (remainder should never overflow).
955    
956    * added primops to deal with divisions of all flavors to the frontend
957    
958    * handled these primops all the way through so they map to their respective
959      MLRISC support
960    
961    * used these primops in the implementation of Int, Int32, Word, Word32
962    
963    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
964    
965    * parameterized INLMIN, INLMAX, and INLABS by a numkind
966    
967    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
968      floating point
969    
970    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
971      and Real (but Real.abs maps to a separate floating-point-only primop)
972    
973    
974    TODO items:
975    
976    * Hacked Alpha32 instruction selection, disabling the selection of REMx
977      instructions because the machine instruction encoder cannot handle
978      them.  (Hppa, PPC, and Sparc instruction selection did not handle
979      REM in the first place, and REM is supported by the x86 machine coder.)
980    
981    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
982      selection phase.  (The two can be streamlined because the hardware
983      delivers both quotient and remainder at the same time anyway.)
984    
985    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
986      (Currently the behavior is inconsistent both across architectures and
987      wrt. the draft Basis spec.)
988    
989    * Word8 should eventually be handled natively, too.
990    
991    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
992      as if there currently is no execution path that could trigger it in
993      SML/NJ.  (The assumptions underlying functions arith and promotable do not
994      hold for things like multiplication and division.)
995    
996    ----------------------------------------------------------------------
997    Name: Matthias Blume
998    Date: 2002/03/27 16:27:00 EST
999    Tag: blume-20020327-mlrisc-divisions
1000    Description:
1001    
1002    Added support for all four division operations (ML's div, mod, quot,
1003    and rem) to MLRISC.  In the course of doing so, I also rationalized
1004    the naming (no more annoying switch-around of DIV and QUOT), by
1005    parameterizing the operation by div_rounding_mode (which can be either
1006    DIV_TO_ZERO or DIV_TO_NEGINF).
1007    
1008    The generic MLTreeGen functor takes care of compiling all four
1009    operations down to only round-to-zero div.
1010    
1011    Missing pieces:
1012    
1013      * Doing something smarter than relying on MLTreeGen on architectures
1014        like, e.g., the x86 where hardware division delivers both quotient and
1015        remainder at the same time.  With this, the implementation of the
1016        round-to-neginf operations could be further streamlined.
1017    
1018      * Remove inlining support for div/mod/rem from the frontend and replace it
1019        with primops that get carried through to the backend.  Do this for all
1020        int and word types.
1021    
1022    ----------------------------------------------------------------------
1023    Name: Matthias Blume
1024    Date: 2002/03/25 17:25:00 EST
1025    Tag: blume-20020325-divmod
1026    Description:
1027    
1028    I improved (hopefully without breaking them) the implementation of Int.div,
1029    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1030    advantage of the following observations:
1031    
1032      Let  q = x quot y      r = x rem y
1033           d = x div  y      m = x mod y
1034    
1035    where "quot" is the round-to-zero version of integer division that
1036    hardware usually provides.  Then we have:
1037    
1038         r = x - q * y        where neither the * nor the - will overflow
1039         d = if q >= 0 orelse x = q * y then q else q - 1
1040                              where neither the * nor the - will overflow
1041         m = if q >= 0 orelse r = 0 then r else r + y
1042                              where the + will not overflow
1043    
1044    This results in substantial simplification of the generated code.
1045    The following table shows the number of CFG nodes and edges generated
1046    for
1047            fun f (x, y) = x OPER y
1048            (* with OPER \in div, mod, quot, rem *)
1049    
1050    
1051        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1052        --------------------------------------------------------
1053         div |         24 |         39 |         12 |         16
1054         mod |         41 |         71 |         12 |         16
1055        quot |          8 |         10 |          8 |         10
1056         rem |         10 |         14 |          8 |         10
1057    
1058    
1059    ----------------------------------------------------------------------
1060    Name: Matthias Blume
1061    Date: 2002/03/25 22:06:00 EST
1062    Tag: blume-20020325-cprotobug
1063    Description:
1064    
1065    Fixed a bug in cproto (c prototype decoder).
1066    
1067    ----------------------------------------------------------------------
1068    Name: Matthias Blume
1069    Date: 2002/03/25 16:00:00 EST
1070    Tag: blume-20020325-raw-primops
1071    Description:
1072    
1073    I did some cleanup to Allen's new primop code and
1074    replaced yesterday's bootfiles with new ones.
1075    (But they are stored in the same place.)
1076    
1077    ----------------------------------------------------------------------
1078    Name: Matthias Blume
1079    Date: 2002/03/24 22:40:00 EST
1080    Tag: blume-20020324-bootfiles
1081    Description:
1082    
1083    Made the bootfiles that Allen asked for.
1084    
1085    ----------------------------------------------------------------------
1086    Name: Allen Leung
1087    Date: 2002/03/23 15:50:00 EST
1088    Tag: leunga-20020323-flint-cps-rcc-primops
1089    Description:
1090    
1091      1. Changes to FLINT primops:
1092    
1093        (* make a call to a C-function;
1094         * The primop carries C function prototype information and specifies
1095         * which of its (ML-) arguments are floating point. C prototype
1096         * information is for use by the backend, ML information is for
1097         * use by the CPS converter. *)
1098      | RAW_CCALL of { c_proto: CTypes.c_proto,
1099                       ml_args: ccall_type list,
1100                       ml_res_opt: ccall_type option,
1101                       reentrant : bool
1102                     } option
1103       (* Allocate uninitialized storage on the heap.
1104        * The record is meant to hold short-lived C objects, i.e., they
1105        * are not ML pointers.  With the tag, the representation is
1106        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1107        *)
1108      | RAW_RECORD of {tag:bool,sz:int}
1109      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1110    
1111      2.  These CPS primops are now overloaded:
1112    
1113           rawload of {kind:numkind}
1114           rawstore of {kind:numkind}
1115    
1116          The one argument form is:
1117    
1118             rawload {kind} address
1119    
1120          The two argument form is:
1121    
1122             rawload {kind} [ml object, byte-offset]
1123    
1124      3. RAW_CCALL/RCC now takes two extra arguments:
1125    
1126         a. The first is whether the C call is reentrant, i.e., whether
1127            ML state should be saved and restored.
1128         b. The second argument is a string argument specifying the name of
1129            library and the C function.
1130    
1131         These things are currently not handled in the code generator, yet.
1132    
1133      4. In CProto,
1134    
1135         An encoding type of "bool" means "ml object" and is mapped into
1136         C prototype of PTR.  Note that "bool" is different than "string",
1137         even though "string" is also mapped into PTR, because "bool"
1138         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1139    
1140      5. Pickler/unpicker
1141    
1142         Changed to handle RAW_RECORD and newest RAW_CCALL
1143    
1144      6. MLRiscGen,
1145    
1146         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1147         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1148            in the file CodeGen/cpscompile/cps-c-calls.sml
1149    
1150      7. Added the conditional move operator
1151    
1152             condmove of branch
1153    
1154         to cps.  Generation of this is still buggy so it is currently
1155         disabled.
1156    
1157    ----------------------------------------------------------------------
1158    Name: Lal George
1159    Date: 2002/03/22 14:18:25 EST
1160    Tag: george-20020322-cps-branch-prob
1161    Description:
1162    
1163    Implemented the Ball-Larus branch prediction-heuristics, and
1164    incorporated graphical viewers for control flow graphs.
1165    
1166    Ball-Larus Heuristics:
1167    ---------------------
1168    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1169    
1170    By design it uses the Dempster-Shafer theory for combining
1171    probabilities.  For example, in the function:
1172    
1173        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1174    
1175    the ball-larus heuristics predicts that the n=0 is unlikely
1176    (OH-heuristic), and the 'then' branch is unlikely because of the
1177    RH-heuristic -- giving the 'then' branch an even lower combined
1178    probability using the Dempster-Shafer theory.
1179    
1180    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1181    probability of the 'then' branch because of the loop in the else
1182    branch.
1183    
1184    
1185    Graphical Viewing:
1186    ------------------
1187    I merely plugged in Allen's graphical viewers into the compiler. The
1188    additional code is not much. At the top level, saying:
1189    
1190            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1191    
1192    will display the graphical view of the control flow graph just before
1193    back-patching.  daVinci must be in your path for this to work. If
1194    daVinci is not available, then the default viewer can be changed
1195    using:
1196    
1197            Control.MLRISC.getString "viewer"
1198    
1199    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1200    course, these viewers must be in your path.
1201    
1202    The above will display the compilation unit at the level of clusters,
1203    many of which are small, boring, and un-interesting. Also setting:
1204    
1205            Control.MLRISC.getInt "cfg-graphical-view_size"
1206    
1207    will display clusters that are larger than the value set by the above.
1208    
1209    
1210    ----------------------------------------------------------------------
1211    Name: Matthias Blume
1212    Date: 2002/03/21 22:20:00 EST
1213    Tag: blume-20020321-kmp-bugfix
1214    Description:
1215    
1216    Changed the interface to the KMP routine in PreString and fixed
1217    a minor bug in one place where it was used.
1218    
1219    ----------------------------------------------------------------------
1220    Name: Allen Leung
1221    Date: 2002/03/21 20:30:00 EST
1222    Tag: leunga-20020321-cfg
1223    Description:
1224    
1225      Fixed a potential problem in cfg edge splitting.
1226    
1227    ----------------------------------------------------------------------
1228    Name: Allen Leung
1229    Date: 2002/03/21 17:15:00 EST
1230    Tag: leunga-20020321-x86-fp-cfg
1231    Description:
1232    
1233      1. Recoded the buggy parts of x86-fp.
1234    
1235         a. All the block reordering code has been removed.
1236            We now depend on the block placement phases to do this work.
1237    
1238         b. Critical edge splitting code has been simplified and moved into the
1239            CFG modules, as where they belong.
1240    
1241         Both of these were quite buggy and complex.  The code is now much, much
1242         simpler.
1243    
1244      2. X86 backend.
1245    
1246         a. Added instructions for 64-bit support.  Instruction selection for
1247            64-bit has not been committed, however, since that
1248            requires changes to MLTREE which haven't been approved by
1249            Lal and John.
1250    
1251         b. Added support for FUCOMI and FUCOMIP when generating code for
1252            PentiumPro and above.  We only generate these instructions in
1253            the fast-fp mode.
1254    
1255         c. Added cases for JP and JNP in X86FreqProps.
1256    
1257      3. CFG
1258    
1259         CFG now has a bunch of methods for edge splitting and merging.
1260    
1261      4. Machine description.
1262    
1263         John's simplification of MLTREE_BASIS.fcond broke a few machine
1264         description things:
1265    
1266         rtl-build.{sig,sml} and hppa.mdl fixed.
1267    
1268         NOTE: the machine description stuff in the repository is still broken.
1269               Again, I can't put my fixes in because that involves
1270               changes to MLTREE.
1271    
1272    ----------------------------------------------------------------------
1273    Name: Matthias Blume
1274    Date: 2002/03/20 15:55:00 EST
1275    Tag: blume-20020320-kmp
1276    Description:
1277    
1278    Implemented Knuth-Morris-Pratt string matching in PreString and used
1279    it for String.isSubstring, Substring.isSubstring, and
1280    Substring.position.
1281    
1282    (Might need some stress-testing.  Simple examples worked fine.)
1283    
1284    ----------------------------------------------------------------------
1285    Name: Matthias Blume
1286    Date: 2002/03/19 16:37:00 EST
1287    Tag: blume-20020319-witnesses
1288    Description:
1289    
1290    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1291    
1292    This implements a generic mechanism for changing constness qualifiers
1293    anywhere within big C types without resorting to outright "casts".
1294    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1295    modify the constness at the outermost level.)
1296    The implementation of "convert" is based on the idea of "witness"
1297    values -- values that are not used by the operation but whose types
1298    "testify" to their applicability.  On the implementation side, "convert"
1299    is simply a projection (returning its second curried argument).  With
1300    cross-module inlining, it should not result in any machine code being
1301    generated.
1302    
1303    ----------------------------------------------------------------------
1304    Name: Matthias Blume
1305    Date: 2002/03/15 16:40:00 EST
1306    Tag: blume-20020315-basis
1307    Description:
1308    
1309    Provided (preliminary?) implementations for
1310    
1311      {String,Substring}.{concatWith,isSuffix,isSubstring}
1312    
1313    and
1314    
1315      Substring.full
1316    
1317    Those are in the Basis spec but they were missing in SML/NJ.
1318    
1319    ----------------------------------------------------------------------
1320    Name: Matthias Blume
1321    Date: 2002/03/14 21:30:00 EST
1322    Tag: blume-20020314-controls
1323    Description:
1324    
1325    Controls:
1326    ---------
1327    
1328    1. Factored out the recently-added Controls : CONTROLS stuff and put
1329       it into its own library $/controls-lib.cm.  The source tree for
1330       this is under src/smlnj-lib/Controls.
1331    
1332    2. Changed the names of types and functions in this interface, so they
1333       make a bit more "sense":
1334    
1335          module -> registry
1336          'a registry -> 'a group
1337    
1338    3. The interface now deals in ref cells only.  The getter/setter interface
1339       is (mostly) gone.
1340    
1341    4. Added a function that lets one register an already-existing ref cell.
1342    
1343    5. Made the corresponding modifications to the rest of the code so that
1344       everything compiles again.
1345    
1346    6. Changed the implementation of Controls.MLRISC back to something closer
1347       to the original.  In particular, this module (and therefore MLRISC)
1348       does not depend on Controls.  There now is some link-time code in
1349       int-sys.sml that registers the MLRISC controls with the Controls
1350       module.
1351    
1352    CM:
1353    ---
1354    
1355      * One can now specify the lambda-split aggressiveness in init.cmi.
1356    
1357    ----------------------------------------------------------------------
1358    Name: Allen Leung
1359    Date: 2002/03/13 17:30:00 EST
1360    Tag: leunga-20020313-x86-fp-unary
1361    Description:
1362    
1363    Bug fix for:
1364    
1365    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1366    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1367    > - fun f(x,(y,z)) = Real.~ y;
1368    > [autoloading]
1369    > [autoloading done]
1370    >       fchsl   (%eax), 184(%esp)
1371    > Error: MLRisc bug: X86MCEmitter.emitInstr
1372    >
1373    > uncaught exception Error
1374    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1375    
1376    The problem was that the code generator did not generate any fp registers
1377    in this case, and the ra didn't know that it needed to run the X86FP phase to
1378    translate the pseudo fp instruction.   This only happened with unary fp
1379    operators in certain situations.
1380    
1381    ----------------------------------------------------------------------
1382    Name: Matthias Blume
1383    Date: 2002/03/13 14:00:00 EST
1384    Tag: blume-20020313-overload-etc
1385    Description:
1386    
1387    1. Added _overload as a synonym for overload for backward compatibility.
1388       (Control.overloadKW must be true for either version to be accepted.)
1389    
1390    2. Fixed bug in install script that caused more things to be installed
1391       than what was requested in config/targets.
1392    
1393    3. Made CM aware of the (_)overload construct so that autoloading
1394       works.
1395    
1396    ----------------------------------------------------------------------
1397    Name: Matthias Blume
1398    Date: 2002/03/12 22:03:00 EST
1399    Tag: blume-20020312-url
1400    Description:
1401    
1402    Forgot to update BOOT and srcarchiveurl.
1403    
1404    ----------------------------------------------------------------------
1405    Name: Matthias Blume
1406    Date: 2002/03/12 17:30:00 EST
1407    Tag: blume-20020312-version110392
1408    Description:
1409    
1410    Yet another version number bump (because of small changes to the
1411    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1412    
1413    Changes:
1414    
1415      The new pid generation scheme described a few weeks ago was overly
1416      complicated.  I implemented a new mechanism that is simpler and
1417      provides a bit more "stability":  Once CM has seen a compilation
1418      unit, it keeps its identity constant (as long as you do not delete
1419      those crucial CM/GUID/* files).  This means that when you change
1420      an interface, compile, then go back to the old interface, and
1421      compile again, you arrive at the original pid.
1422    
1423      There now also is a mechanism that instructs CM to use the plain
1424      environment hash as a module's pid (effectively making its GUID
1425      the empty string).  For this, "noguid" must be specified as an
1426      option to the .sml file in question within its .cm file.
1427      This is most useful for code that is being generated by tools such
1428      as ml-nlffigen (because during development programmers tend to
1429      erase the tool's entire output directory tree including CM's cached
1430      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1431      revert to the old, broken behavior of SML/NJ, but in specific cases
1432      where there is no danger of interface confusion, its use is ok
1433      (I think).
1434    
1435      ml-nlffigen by default generates "noguid" annotations.  They can be
1436      turned off by specifying -guid in its command line.
1437    
1438    ----------------------------------------------------------------------
1439    Name: Lal George
1440    Date: 2002/03/12 12 14:42:36 EST
1441    Tag: george-20020312-frequency-computation
1442    Description:
1443    
1444    Integrated jump chaining and static block frequency into the
1445    compiler. More details and numbers later.
1446    
1447    ----------------------------------------------------------------------
1448    Name: Lal George
1449    Date: 2002/03/11 11 22:38:53 EST
1450    Tag: george-20020311-jump-chain-elim
1451    Description:
1452    
1453    Tested the jump chain elimination on all architectures (except the
1454    hppa).  This is on by default right now and is profitable for the
1455    alpha and x86, however, it may not be profitable for the sparc and ppc
1456    when compiling the compiler.
1457    
1458    The gc test will typically jump to a label at the end of the cluster,
1459    where there is another jump to an external cluster containing the actual
1460    code to invoke gc. This is to allow factoring of common gc invocation
1461    sequences. That is to say, we generate:
1462    
1463            f:
1464               testgc
1465               ja   L1      % jump if above to L1
1466    
1467            L1:
1468               jmp L2
1469    
1470    
1471    After jump chain elimination the 'ja L1' instructions is converted to
1472    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1473    up being implemented in their long form (if L2 is far away) using:
1474    
1475            jbe     L3      % jump if below or equal to L3
1476            jmp     L2
1477         L3:
1478            ...
1479    
1480    
1481    For large compilation units L2  may be far away.
1482    
1483    
1484    ----------------------------------------------------------------------
1485    Name: Matthias Blume
1486    Date: 2002/03/11 13:30:00 EST
1487    Tag: blume-20020311-mltreeeval
1488    Description:
1489    
1490    A functor parameter was missing.
1491    
1492    ----------------------------------------------------------------------
1493    Name: Allen Leung
1494    Date: 2002/03/11 10:30:00 EST
1495    Tag: leunga-20020311-runtime-string0
1496    Description:
1497    
1498       The representation of the empty string now points to a
1499    legal null terminated C string instead of unit.  It is now possible
1500    to convert an ML string into C string with InlineT.CharVector.getData.
1501    This compiles into one single machine instruction.
1502    
1503    ----------------------------------------------------------------------
1504    Name: Allen Leung
1505    Date: 2002/03/10 23:55:00 EST
1506    Tag: leunga-20020310-x86-call
1507    Description:
1508    
1509       Added machine generation for CALL instruction (relative displacement mode)
1510    
1511    ----------------------------------------------------------------------
1512    Name: Matthias Blume
1513    Date: 2002/03/08 16:05:00
1514    Tag: blume-20020308-entrypoints
1515    Description:
1516    
1517    Version number bumped to 110.39.1.  NEW BOOTFILES!
1518    
1519    Entrypoints: non-zero offset into a code object where execution should begin.
1520    
1521    - Added the notion of an entrypoint to CodeObj.
1522    - Added reading/writing of entrypoint info to Binfile.
1523    - Made runtime system bootloader aware of entrypoints.
1524    - Use the address of the label of the first function given to mlriscGen
1525      as the entrypoint.  This address is currently always 0, but it will
1526      not be 0 once we turn on block placement.
1527    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
1528      with entry points) from mlriscGen.
1529    
1530    ----------------------------------------------------------------------
1531    Name: Allen Leung
1532    Date: 2002/03/07 20:45:00 EST
1533    Tag: leunga-20020307-x86-cmov
1534    Description:
1535    
1536       Bug fixes for CMOVcc on x86.
1537    
1538       1. Added machine code generation for CMOVcc
1539       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1540       3. CMOVcc cannot have an immediate operand as argument.
1541    
1542    ----------------------------------------------------------------------
1543    Name: Matthias Blume
1544    Date: 2002/03/07 16:15:00 EST
1545    Tag: blume-20020307-controls
1546    Description:
1547    
1548    This is a very large but mostly boring patch which makes (almost)
1549    every tuneable compiler knob (i.e., pretty much everything under
1550    Control.* plus a few other things) configurable via both the command
1551    line and environment variables in the style CM did its configuration
1552    until now.
1553    
1554    Try starting sml with '-h' (or, if you are brave, '-H')
1555    
1556    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1557    implements the underlying generic mechanism.
1558    
1559    The interface to some of the existing such facilities has changed somewhat.
1560    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1561    (The getFoo interface is still there for backward-compatibility, but its
1562    use is deprecated.)
1563    
1564    The ml-build script passes -Cxxx=yyy command-line arguments through so
1565    that one can now twiddle the compiler settings when using this "batch"
1566    compiler.
1567    
1568    TODO items:
1569    
1570    We should go through and throw out all controls that are no longer
1571    connected to anything.  Moreover, we should go through and provide
1572    meaningful (and correct!) documentation strings for those controls
1573    that still are connected.
1574    
1575    Currently, multiple calls to Controls.new are accepted (only the first
1576    has any effect).  Eventually we should make sure that every control
1577    is being made (via Controls.new) exactly once.  Future access can then
1578    be done using Controls.acc.
1579    
1580    Finally, it would probably be a good idea to use the getter-setter
1581    interface to controls rather than ref cells.  For the time being, both
1582    styles are provided by the Controls module, but getter-setter pairs are
1583    better if thread-safety is of any concern because they can be wrapped.
1584    
1585    *****************************************
1586    
1587    One bug fix: The function blockPlacement in three of the MLRISC
1588    backpatch files used to be hard-wired to one of two possibilities at
1589    link time (according to the value of the placementFlag).  But (I
1590    think) it should rather sense the flag every time.
1591    
1592    *****************************************
1593    
1594    Other assorted changes (by other people who did not supply a HISTORY entry):
1595    
1596    1. the cross-module inliner now works much better (Monnier)
1597    2. representation of weights, frequencies, and probabilities in MLRISC
1598       changed in preparation of using those for weighted block placement
1599       (Reppy, George)
1600    
1601    ----------------------------------------------------------------------
1602    Name: Lal George
1603    Date: 2002/03/07 14:44:24 EST 2002
1604    Tag: george-20020307-weighted-block-placement
1605    
1606    Tested the weighted block placement optimization on all architectures
1607    (except the hppa) using AMPL to generate the block and edge frequencies.
1608    Changes were required in the machine properties to correctly
1609    categorize trap instructions. There is an MLRISC flag
1610    "weighted-block-placement" that can be used to enable weighted block
1611    placement, but this will be ineffective without block/edge
1612    frequencies (coming soon).
1613    
1614    
1615    ----------------------------------------------------------------------
1616    Name: Lal George
1617    Date: 2002/03/05 17:24:48 EST
1618    Tag: george-20020305-linkage-cluster
1619    
1620    In order to support the block placement optimization, a new cluster
1621    is generated as the very first cluster (called the linkage cluster).
1622    It contains a single jump to the 'real' entry point for the compilation
1623    unit. Block placement has no effect on the linkage cluster itself, but
1624    all the other clusters  have full freedom in the manner in which they
1625    reorder blocks or functions.
1626    
1627    On the x86 the typical linkage code that is generated is:
1628       ----------------------
1629            .align 2
1630       L0:
1631            addl    $L1-L0, 72(%esp)
1632            jmp     L1
1633    
1634    
1635            .align  2
1636       L1:
1637       ----------------------
1638    
1639    72(%esp) is the memory location for the stdlink register. This
1640    must contain the address of the CPS function being called. In the
1641    above example, it contains the address of  L0; before
1642    calling L1 (the real entry point for the compilation unit), it
1643    must contain the address for L1, and hence
1644    
1645            addl $L1-L0, 72(%esp)
1646    
1647    I have tested this on all architectures except the hppa.The increase
1648    in code size is of course negligible
1649    
1650    ----------------------------------------------------------------------
1651    Name: Allen Leung
1652    Date: 2002/03/03 13:20:00 EST
1653    Tag: leunga-20020303-mlrisc-tools
1654    
1655      Added #[ ... ] expressions to mlrisc tools
1656    
1657    ----------------------------------------------------------------------
1658    Name: Matthias Blume
1659    Date: 2002/02/27 12:29:00 EST
1660    Tag: blume-20020227-cdebug
1661    Description:
1662    
1663    - made types in structure C and C_Debug to be equal
1664    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1665    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1666    
1667    ----------------------------------------------------------------------
1668    Name: Matthias Blume
1669    Date: 2002/02/26 12:00:00 EST
1670    Tag: blume-20020226-ffi
1671    Description:
1672    
1673    1. Fixed a minor bug in CM's "noweb" tool:
1674       If numbering is turned off, then truly don't number (i.e., do not
1675       supply the -L option to noweb).  The previous behavior was to supply
1676       -L'' -- which caused noweb to use the "default" line numbering scheme.
1677       Thanks to Chris Richards for pointing this out (and supplying the fix).
1678    
1679    2. Once again, I reworked some aspects of the FFI:
1680    
1681       A. The incomplete/complete type business:
1682    
1683       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
1684         gone!
1685       - ML types representing an incomplete type are now *equal* to
1686         ML types representing their corresponding complete types (just like
1687         in C).  This is still safe because ml-nlffigen will not generate
1688         RTTI for incomplete types, nor will it generate functions that
1689         require access to such RTTI.   But when ML code generated from both
1690         incomplete and complete versions of the C type meet, the ML types
1691         are trivially interoperable.
1692    
1693         NOTE:  These changes restore the full generality of the translation
1694         (which was previously lost when I eliminated functorization)!
1695    
1696       B. Enum types:
1697    
1698       - Structure C now has a type constructor "enum" that is similar to
1699         how the "su" constructor works.  However, "enum" is not a phantom
1700         type because each "T enum" has values (and is isomorphic to
1701         MLRep.Signed.int).
1702       - There are generic access operations for enum objects (using
1703         MLRep.Signed.int).
1704       - ml-nlffigen will generate a structure E_foo for each "enum foo".
1705         * The structure contains the definition of type "mlrep" (the ML-side
1706         representation type of the enum).  Normally, mlrep is the same
1707         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
1708         then mlrep will be defined as a datatype -- thus facilitating
1709         pattern matching on mlrep values.
1710         ("-ec" will be suppressed if there are duplicate values in an
1711          enumeration.)
1712         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
1713         will be generated for each C enum constant xxx.
1714         * Conversion functions m2i and i2m convert between mlrep and
1715         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
1716         * Coversion functions c and ml convert between mlrep and "tag enum".
1717         * Access functions (get/set) fetch and store mlrep values.
1718       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
1719         enumerations are merged into one single enumeration represented by
1720         structure E_'.
1721    
1722    ----------------------------------------------------------------------
1723    Name: Allen Leung
1724    Date: 2002/02/25 04:45:00 EST
1725    Tag: leunga-20020225-cps-spill
1726    
1727    This is a new implementation of the CPS spill phase.
1728    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
1729    In case of problems, replace it with the old file spill.sml
1730    
1731    The current compiler runs into some serious performance problems when
1732    constructing a large record.  This can happen when we try to compile a
1733    structure with many items.  Even a very simple structure like the following
1734    makes the compiler slow down.
1735    
1736        structure Foo = struct
1737           val x_1 = 0w1 : Word32.int
1738           val x_2 = 0w2 : Word32.int
1739           val x_3 = 0w3 : Word32.int
1740           ...
1741           val x_N = 0wN : Word32.int
1742        end
1743    
1744    The following table shows the compile time, from N=1000 to N=4000,
1745    with the old compiler:
1746    
1747    N
1748    1000   CPS 100 spill                           0.04u  0.00s  0.00g
1749           MLRISC ra                               0.06u  0.00s  0.05g
1750              (spills = 0 reloads = 0)
1751           TOTAL                                   0.63u  0.07s  0.21g
1752    
1753    1100   CPS 100 spill                           8.25u  0.32s  0.64g
1754           MLRISC ra                               5.68u  0.59s  3.93g
1755              (spills = 0 reloads = 0)
1756           TOTAL                                   14.71u  0.99s  4.81g
1757    
1758    1500   CPS 100 spill                           58.55u  2.34s  1.74g
1759           MLRISC ra                               5.54u  0.65s  3.91g
1760              (spills = 543 reloads = 1082)
1761           TOTAL                                   65.40u  3.13s  6.00g
1762    
1763    2000   CPS 100 spill                           126.69u  4.84s  3.08g
1764           MLRISC ra                               0.80u  0.10s  0.55g
1765              (spills = 42 reloads = 84)
1766           TOTAL                                   129.42u  5.10s  4.13g
1767    
1768    3000   CPS 100 spill                           675.59u  19.03s  11.64g
1769           MLRISC ra                               2.69u  0.27s  1.38g
1770              (spills = 62 reloads = 124)
1771           TOTAL                                   682.48u  19.61s  13.99g
1772    
1773    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
1774           MLRISC ra                               4.96u  0.27s  2.72g
1775              (spills = 85 reloads = 170)
1776           TOTAL                                   2375.26u  57.21s  48.00g
1777    
1778    As you can see the old cps spill module suffers from some serious
1779    performance problem.  But since I cannot decipher the old code fully,
1780    instead of patching the problems up, I'm reimplementing it
1781    with a different algorithm.  The new code is more modular,
1782    smaller when compiled, and substantially faster
1783    (O(n log n) time and O(n) space).  Timing of the new spill module:
1784    
1785    4000  CPS 100 spill                           0.02u  0.00s  0.00g
1786          MLRISC ra                               0.25u  0.02s  0.15g
1787             (spills=1 reloads=3)
1788          TOTAL                                   7.74u  0.34s  1.62g
1789    
1790    Implementation details:
1791    
1792    As far as I can tell, the purpose of the CPS spill module is to make sure the
1793    number of live variables at any program point (the bandwidth)
1794    does not exceed a certain limit, which is determined by the
1795    size of the spill area.
1796    
1797    When the bandwidth is too large, we decrease the register pressure by
1798    packing live variables into spill records.  How we achieve this is
1799    completely different than what we did in the old code.
1800    
1801    First, there is something about the MLRiscGen code generator
1802    that we should be aware of:
1803    
1804    o MLRiscGen performs code motion!
1805    
1806       In particular, it will move floating point computations and
1807       address computations involving only the heap pointer to
1808       their use sites (if there is only a single use).
1809       What this means is that if we have a CPS record construction
1810       statement
1811    
1812           RECORD(k,vl,w,e)
1813    
1814       we should never count the new record address w as live if w
1815       has only one use (which is often the case).
1816    
1817       We should do something similar to floating point, but the transformation
1818       there is much more complex, so I won't deal with that.
1819    
1820    Secondly, there are now two new cps primops at our disposal:
1821    
1822     1. rawrecord of record_kind option
1823        This pure operator allocates some uninitialized storage from the heap.
1824        There are two forms:
1825    
1826         rawrecord NONE [INT n]  allocates a tagless record of length n
1827         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
1828                                     and initializes the tag.
1829    
1830     2. rawupdate of cty
1831          rawupdate cty (v,i,x)
1832          Assigns to x to the ith component of record v.
1833          The storelist is not updated.
1834    
1835    We use these new primops for both spilling and increment record construction.
1836    
1837     1. Spilling.
1838    
1839        This is implemented with a linear scan algorithm (but generalized
1840        to trees).  The algorithm will create a single spill record at the
1841        beginning of the cps function and use rawupdate to spill to it,
1842        and SELECT or SELp to reload from it.  So both spills and reloads
1843        are fine-grain operations.  In contrast, in the old algorithm
1844        "spills" have to be bundled together in records.
1845    
1846        Ideally, we should sink the spill record construction to where
1847        it is needed.  We can even split the spill record into multiple ones
1848        at the places where they are needed.  But CPS is not a good
1849        representation for global code motion, so I'll keep it simple and
1850        am not attempting this.
1851    
1852     2. Incremental record construction (aka record splitting).
1853    
1854        Long records with many component values which are simulatenously live
1855        (recall that single use record addresses are not considered to
1856         be live) are constructed with rawrecord and rawupdate.
1857        We allocate space on the heap with rawrecord first, then gradually
1858        fill it in with rawupdate.  This is the technique suggested to me
1859        by Matthias.
1860    
1861        Some restrictions on when this is applicable:
1862        1. It is not a VECTOR record.  The code generator currently does not handle
1863           this case. VECTOR record uses double indirection like arrays.
1864        2. All the record component values are defined in the same "basic block"
1865           as the record constructor.  This is to prevent speculative
1866           record construction.
1867    
1868    ----------------------------------------------------------------------
1869    Name: Allen Leung
1870    Date: 2002/02/22 01:02:00 EST
1871    Tag: leunga-20020222-mlrisc-tools
1872    
1873    Minor bug fixes in the parser and rewriter
1874    
1875    ----------------------------------------------------------------------
1876    Name: Allen Leung
1877    Date: 2002/02/21 20:20:00 EST
1878    Tag: leunga-20020221-peephole
1879    
1880    Regenerated the peephole files.  Some contained typos in the specification
1881    and some didn't compile because of pretty printing bugs in the old version
1882    of 'nowhere'.
1883    
1884    ----------------------------------------------------------------------
1885    Name: Allen Leung
1886    Date: 2002/02/19 20:20:00 EST
1887    Tag: leunga-20020219-mlrisc-tools
1888    Description:
1889    
1890       Minor bug fixes to the mlrisc-tools library:
1891    
1892       1.  Fixed up parsing colon suffixed keywords
1893       2.  Added the ability to shut the error messages up
1894       3.  Reimplemented the pretty printer and fixed up/improved
1895           the pretty printing of handle and -> types.
1896       4.  Fixed up generation of literal symbols in the nowhere tool.
1897       5.  Added some SML keywords to to sml.sty
1898    
1899    ----------------------------------------------------------------------
1900  Name: Matthias Blume  Name: Matthias Blume
1901  Date: 2002/02/19 16:20:00 EST  Date: 2002/02/19 16:20:00 EST
1902  Tag: blume-20020219-cmffi  Tag: blume-20020219-cmffi

Legend:
Removed from v.1078  
changed lines
  Added in v.1399

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