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 1003, Fri Dec 7 02:45:32 2001 UTC revision 1488, Tue May 11 19:20:17 2004 UTC
# Line 11  Line 11 
11  Date: yyyy/mm/dd  Date: yyyy/mm/dd
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14    
15    ----------------------------------------------------------------------
16    Name: Matthias Blume (blume (at) tti - c (dot) org)
17    Date: 2004/05/11 14:20:00 CDT
18    Tag: blume-20040511-installml
19    Description:
20    
21    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
22    the patch.)
23    
24    ----------------------------------------------------------------------
25    Name: Matthias Blume (blume (at) tti - c (dot) org)
26    Date: 2004/05/11 14:05:00 CDT
27    Tag: blume-20040511-nlffi-netbsd
28    Description:
29    
30    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
31    the patch.)
32    
33    ----------------------------------------------------------------------
34    Name: Matthias Blume (blume (at) tti - c (dot) org)
35    Date: 2004/05/11 12:05:00 CDT
36    Tag: blume-20040511-exports
37    Description:
38    
39    As per request by Adam Chlipala <adam@hcoop.net>, extended various
40    export lists in compiler-related .cm-files.
41    
42    ----------------------------------------------------------------------
43    Name: Matthias Blume (blume (at) tti - c (dot) org)
44    Date: 2004/05/11 11:35:00 CDT
45    Tag: blume-20040511-allsource
46    Description:
47    
48    The installer now honors the "src-smlnj" target again, although its meaning
49    has changed from "all sources required for the compiler" to "all sources
50    the installer knows about".  In other words, if you enable "src-smlnj"
51    in the "targets" file, then the installer will pull in sources for
52    everything.  (Notice that this refers to source code only.  Compiled
53    code is still only installed for modules that were requested explicitly
54    or which are required for other modules that were requested explicitly.)
55    
56    ----------------------------------------------------------------------
57    Name: Matthias Blume (blume (at) tti - c (dot) org)
58    Date: 2004/04/23 17:40:00 CDT
59    Tag: blume-20040423-ieee-scan
60    Description:
61    
62    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
63    in the exponent calculation we get INF or ZERO (depending on the mantissa
64    and the sign of the exponent).
65    
66    ----------------------------------------------------------------------
67    Name: Matthias Blume (blume (at) tti - c (dot) org)
68    Date: 2004/04/23 10:40:00 CDT
69    Tag: blume-20040423-ml-build
70    Description:
71    
72    The ml-build script now terminates with a non-0 status when something
73    goes wrong.
74    
75    ----------------------------------------------------------------------
76    Name: Matthias Blume (blume (at) tti - c (dot) org)
77    Date: 2004/04/22 16:35:00 CDT
78    Tag: blume-20040422-Option
79    Description:
80    
81    Made exception Option to be the same as exception Option.Option
82    (as it should be).
83    
84    ----------------------------------------------------------------------
85    Name: Allen Leung (leunga (at) reservoir (dot) com)
86    Date: 2004/03/19 14:40:00 EST
87    Tag: leunga-20040319-cygwin-nlffi
88    Description:
89    
90    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
91    of SML/NJ.  The problem is that
92    
93       lib = dlopen(NULL, ...)
94       f   = dlsym(lib, "malloc");
95    
96    does not work on Windows unless we explicitly export symbols
97    such as 'malloc' during linking.  We fixed this by explicitly
98    exporting the required symbols with the magic gcc incantation:
99    
100        -Wl,--export-all cygwin.def
101    
102    where cygwin.def is a file containing all the symbols that
103    we wish to export.
104    
105    I suspect this is a Windows problem and we'll have to
106    do the same (somehow with windows compilers) when
107    we build the native win32 version with the system
108    calls LoadLibrary/GetProcAddress.
109    
110    ----------------------------------------------------------------------
111    Name: Matthias Blume (blume (at) tti - c (dot) org)
112    Date: 2004/03/04 16:35:00 CST
113    Tag: blume-20040304-intinf-fmt
114    Description:
115    
116    Fixed problem with IntInf.fmt (sign would show up on the right instead
117    of on the left for BIN, OCT, and HEX).
118    
119    ----------------------------------------------------------------------
120    Name: Matthias Blume (blume (at) tti - c (dot) org)
121    Date: 2004/03/04 11:25:00 CST
122    Tag: blume-20040304-symlinks
123    Description:
124    
125    Fixed problem with installer script (unix only) where bin/ml-yacc and
126    friends pointed (via symlinks) to absolute locations instead of just
127    .run-sml.  This was reported by Vesa A Norrman.
128    
129    ----------------------------------------------------------------------
130    Name: Matthias Blume (blume (at) tti - c (dot) org)
131    Date: 2004/02/13 14:50:00 CST
132    Tag: Release_110_45
133    Description:
134    
135    New working version (110.45). New bootfiles.
136    
137    ----------------------------------------------------------------------
138    Name: Matthias Blume (blume (at) tti - c (dot) org)
139    Date: 2004/01/26 15:15:15 CST
140    Tag: blume-20040126-toplevel
141    Description:
142    
143    Improved handling of exceptions at the interactive toplevel.
144    
145    ----------------------------------------------------------------------
146    Name: Matthias Blume (blume (at) tti - c (dot) org)
147    Date: 2004/01/26 11:25:00
148    Tag: blume-20040126-app
149    Description:
150    
151    Type of top-level "app" corrected.
152    Added code for setting vp_limitPtrMask to Win32-specific runtime.
153    
154    ----------------------------------------------------------------------
155    Name: Matthias Blume (blume (at) tti - c (dot) org)
156    Date: 2003/11/18 17:10 CST
157    Tag: blume-20031118-basis-fiddle
158    Description:
159    
160    - changed Timer interface to what might become the spec
161    - POSIX_FLAGS -> BIT_FLAGS according to spec
162    - some other minor discrepancies wrt. spec eliminated
163    
164    ----------------------------------------------------------------------
165    Name: Matthias Blume (blume (at) tti - c (dot) org)
166    Date: 2003/11/06 12:00:00 CST
167    Tag: Release_110_44
168    Description:
169    
170    New working version (110.44). New bootfiles.
171    
172    ----------------------------------------------------------------------
173    Name: Matthias Blume (blume (at) tti - c (dot) org)
174    Date: 2003/11/04 11:50:00 CST
175    Tag: blume-20031104-move-libraries
176    Description:
177    
178    Eliminated the "dont_move_libraries" directive in config/targets.
179    (The mechanism was broken and could not be fixed easily.  Moreover,
180    there does not seem to be any reason not to move all libraries into
181    lib during installation.  I originally implemented this directive as a
182    backward-compatibility feature when I first introduced the new CM.
183    Now that things have been stable for a long time and going back to the
184    old CM is not an option, there is no reason to keep it around.)
185    
186    ----------------------------------------------------------------------
187    Name: Matthias Blume (blume (at) tti - c (dot) org)
188    Date: 2003/11/03 16:00:00 CST
189    Tag: blume-20031103-installdir
190    Description:
191    
192    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
193    Richards for pointing out the problem and providing the solution.)
194    
195    ----------------------------------------------------------------------
196    Name: Matthias Blume (blume (at) tti - c (dot) org)
197    Date: 2003/10/01 17:05:00 CDT
198    Tag: blume-20031001-lal-mlrisc
199    Description:
200    
201    MLRISC bug fix from Lal.
202    
203    ----------------------------------------------------------------------
204    Name: Matthias Blume (blume (at) tti - c (dot) org)
205    Date: 2003/09/30 16:10:00 CDT
206    Tag: blume-20030930-primio-bat
207    Description:
208    
209    1. Added openVector, nullRd, and nullWr to PRIM_IO.
210    2. Improved .bat files (for Win32 port) to make things work under Win95.
211       (thanks to Aaron S. Hawley for this one)
212    
213    ----------------------------------------------------------------------
214    Name: Matthias Blume (blume (at) tti - c (dot) org)
215    Date: 2003/09/26 16:05:00 CDT
216    Tag: blume-20030926-wrappriv
217    Description:
218    
219    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
220    
221    ----------------------------------------------------------------------
222    Name: Matthias Blume (blume (at) tti - c (dot) org)
223    Date: 2003/09/26 15:00:00 CDT
224    Tag: blume-20030926-110_43_3
225    Description:
226    
227    - additional cleanup
228    - version number bump, NEW BOOTFILES
229    
230    ----------------------------------------------------------------------
231    Name: Matthias Blume (blume (at) tti - c (dot) org)
232    Date: 2003/09/26 12:00:00 CDT
233    Tag: blume-20030926-ppautoload
234    Description:
235    
236    I modified the read-eval-print loop so that the autoloader gets
237    invoked whenever the prettyprinter tries to look up a symbol that
238    is not currently defined in the toplevel environment but which
239    appears in CM's autoload registry.  As a result, we see far fewer of
240    those ?.Foo.Bar.xxx names in the prettyprinter's output.
241    
242    In addition to this I tried to clean up some pieces of the Basis
243    implementation (e.g., Socket, Word8Array) in order to prevent other
244    instances of these ?.Foo.Bar.xxx names from being printed.
245    
246    The mechanism that picks names for types still needs some work, though.
247    (Right now it seems that if there is a type A.t which is defined to
248    be B.u, but B is unavailable at toplevel, then A.t gets printed as
249    "?.B.u" although the perhaps more sensible solution would be to use
250    "A.t" in this case.  In other words, the prettyprinter should follow
251    a chain of DEFtycs not farther than there are corresponding toplevel
252    names in the current environment.)
253    
254    ----------------------------------------------------------------------
255    Name: Matthias Blume (blume (at) tti - c (dot) org)
256    Date: 2003/09/24 16:31:00 CDT
257    Tag: blume-20030924-installer
258    Description:
259    
260    Another installer tweak:  All the ML code for the installer is now
261    compiled during CMB.make and put into a little library called
262    $smlnj/installer.cm.  The installation then simply invokes
263    
264       sml -m $smlnj/installer.cm
265    
266    and everything happens automagically.
267    
268    Win32:  ML code senses value of environment variable SMLNJ_HOME.
269    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
270            and BINDIR.
271    
272    The new scheme guarantees that the ML code responsible for the installation
273    is in sync with the APIs of the main system.  Also, the installer is
274    somewhat faster because the installer script is precompiled.
275    
276    ----------------------------------------------------------------------
277    Name: Matthias Blume (blume (at) tti - c (dot) org)
278    Date: 2003/09/24 15:35:00 CDT
279    Tag: blume-20030924-synsock
280    Description:
281    
282    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
283    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
284    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
285    implementations of non-blocking operations from CML's Socket
286    structure.
287    
288    ----------------------------------------------------------------------
289    Name: Matthias Blume (blume (at) tti - c (dot) org)
290    Date: 2003/09/24 15:10:05 CDT
291    Tag: blume-20030924-sockets
292    Description:
293    
294    1. Fixed SOCKET API and implementation to match Basis spec.
295       This required changing the internal representation of sockets to one
296       that remembers (for each socket file descriptor) whether it is currently
297       blocking or non-blocking.  This state is maintained lazily (i.e., a system
298       call is made only if the state actually needs to change).
299    
300    2. OS-specific details of sockets were moved into separate files, thus
301       making it possible to unify the bulk of the socket implementations
302       between Unix and Win32.
303    
304    3. CML's socket API changed accordingly.
305       (Note that we need to remove non-blocking functions from this API
306       since they are redundant in the case of CML!)
307    
308    4. CML's socket implementation now makes use of non-blocking functions
309       provided by Basis, thus removing all OS-dependent code from this part
310       of CML.
311    
312    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
313    
314    ----------------------------------------------------------------------
315    Name: Matthias Blume (blume (at) tti - c (dot) org)
316    Date: 2003/09/22 12:10:00 CDT
317    Tag: blume-20030922-110_43_2
318    Description:
319    
320    Made a new interim version and bootfiles for developer's bootstrapping
321    convenience.
322    
323       110.43.2 -- NEW BOOTFILES
324    
325    ----------------------------------------------------------------------
326    Name: Matthias Blume (blume (at) tti - c (dot) org)
327    Date: 2003/09/19 15:55:00 CDT
328    Tag: blume-20030919-cmdir
329    Description:
330    
331    1. new-install.sh -> install.sh
332    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
333    3. tweaked installer so that another name instead of .cm can be chosen
334       at install time (by setting the CM_DIR_ARC environment variable
335       during installation); once installation is complete, the name is
336       fixed
337    
338    ----------------------------------------------------------------------
339    Name: Matthias Blume (blume (at) tti - c (dot) org)
340    Date: 2003/09/18 16:00:00 CDT
341    Tag: blume-20030918-110_43_1
342    Description:
343    
344    Made a new interim version and bootfiles for developer's bootstrapping
345    convenience.
346    
347       110.43.1 -- NEW BOOTFILES
348    
349    ----------------------------------------------------------------------
350    Name: Matthias Blume (blume (at) tti - c (dot) org)
351    Date: 2003/09/18 15:20:00 CDT
352    Tag: blume-20030918-misc
353    Description:
354    
355    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
356       this seems to be controversial at the moment) and used those in
357       Posix.ProcEnv.times.
358    
359    2. Added Time.{from,to}Nanoseconds to Time.
360    
361    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
362       For example, fromLargeInt never needs to look at more than 3 "big
363       digits" to get its 53 bits of precision.
364    
365    ----------------------------------------------------------------------
366    Name: Matthias Blume (blume (at) tti - c (dot) org)
367    Date: 2003/09/17 16:30:00 CDT
368    Tag: blume-20030917-real32-slices
369    Description:
370    
371    Added an entry to the primitive environment
372    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
373    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
374    
375    Removed some of the "magic" constants in real64.sml and replaced them
376    with code that generates these values from their corresponding
377    integer counterparts.
378    
379    Made all(?) the slice-related changes to the Basis and made everything
380    compile again...
381    
382    ----------------------------------------------------------------------
383    Name: Matthias Blume (blume (at) tti - c (dot) org)
384    Date: 2003/09/15 17:45:00 CDT
385    Tag: blume-20030915-rbase
386    Description:
387    
388    Fixed bug in Real.fromLargeInt.
389    
390    ----------------------------------------------------------------------
391    Name: Matthias Blume (blume (at) tti - c (dot) org)
392    Date: 2003/09/13 18:11:00 CDT
393    Tag: blume-20030913-libinstall
394    Description:
395    
396    Minor bugfix in config/libinstall (set anchor with path to
397    standalone tool after installing it, otherwise libraries that
398    need ml-lex or ml-yacc won't compile the first time the installer
399    runs).
400    
401    ----------------------------------------------------------------------
402    Name: Matthias Blume (blume (at) tti - c (dot) org)
403    Date: 2003/09/12 11:45:00 CDT
404    Tag: blume-20030912-various
405    Description:
406    
407     - fixed bug in Real.toLargeInt
408     - fixed bug in Posix.ProcEnv.times
409     - changed inputLine functions to return an option
410     - minor installer improvements / bugfixes
411     - changed default @SMLalloc parameter for x86/celeron to 64k
412    
413    ----------------------------------------------------------------------
414    Name: Matthias Blume (blume (at) tti - c (dot) org)
415    Date: 2003/09/09 22:00:00 CDT
416    Tag: Release_110_43
417    Description:
418    
419        New working release 110.43.  New bootfiles.
420    
421    ----------------------------------------------------------------------
422    Name: Matthias Blume (blume (at) tti - c (dot) org)
423    Date: 2003/09/09 19:20:00 CDT
424    Tag: blume-20030909-installer
425    Description:
426    
427    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
428    Modified config/install.bat to take advantage of it.  Also modified
429    config/install.sh (and called it config/new-install.sh) to take advantage
430    of it on Unix systems.  (The SML code is (supposed to be) platform-
431    independent.)
432    
433    The installer can now install everything under Win32
434    as well as under *nix as long as it compiles.
435    
436    Other changes:
437    
438      - made CML compile again under Win32
439      - made eXene compile under Win32 (by providing a fake structure UnixSock
440        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
441      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
442        same as type int;  under Win32 it isn't
443      - fixed some slice-related problems in the win32-specific parts of CML
444      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
445        (under Win32, the volume name is case-insensitive, and the
446        OS.Path code compares volume names for equality)
447    
448    ----------------------------------------------------------------------
449    Name: Matthias Blume (blume (at) tti - c (dot) org)
450    Date: 2003/09/08 11:55:00 CDT
451    Tag: blume-20030908-fullpath
452    Description:
453    
454    Made Win32 version of OS.FileSys.fullPath return current directory
455    when given an empty string.  This is what the spec says, and incidentally,
456    CM depends on it.  (CM otherwise goes into an infinite loop in certain
457    cases when presented with the name of a non-existing .cm file.)
458    
459    ----------------------------------------------------------------------
460    Name: Matthias Blume (blume (at) tti - c (dot) org)
461    Date: 2003/09/04 16:30:00 CDT
462    Tag: blume-20030905-slices-etc
463    Description:
464    
465      1. Changed interface to vectors and arrays in Basis to match
466         (draft) Basis spec.
467      2. Added signatures and implementations of slices according to
468         Basis spec.
469      3. Edited source code throughout the system to make it compile again
470         under 1. and 2.  (In some cases code had to be added to have it
471         match the new signatures.)
472      4. MLRISC should be backward-compatible: the copies of the originals
473         of files that needed to change under 3. were retained, the .cm files
474         check the compiler version number and use old versions when
475         appropriate.
476      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
477         dirstream -> string option (in accordance with Basis spec).
478      6. When generating code that counts lines, ml-lex used function
479         CharVector.foldli, taking advantage of its old interface.
480         This has been replaced with the corresponding code from
481         CharVectorSlice.  (html-lex must be re-lexed!)
482      7. BitArray in smlnj-lib/Util has been extended/modified to match the
483         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
484      8. Removed temporary additions (fromInternal, toInternal) from the
485         (now obsolete) IntInf in smlnj-lib/Util.
486      9. Cleaned up structure Byte.
487     10. Added localOffset, scan, and fromString to Date (according to spec).
488         Cleaned/corrected implementation of Date.
489         (Still need to check for correctness; implement better canonicalizeDate.)
490     11. Added "scan" to signature IEEE_REAL.
491     12. Some improvements to IntInf [in particular: efficiency-hack for
492         mod and rem when second operand is 2 (for parity checks).]
493     13. Changed representation of type Time.time, using a single IntInf.int
494         value counting microseconds.  This considerably simplified the
495         implementation of structure Time.  We now support negative time
496         values; scan and fromString handle signs.
497     14. Functor PrimIO now takes two additional arguments (VectorSlice and
498         ArraySlice).
499    
500    ----------------------------------------------------------------------
501    Name: Matthias Blume (blume (at) tti - c (dot) org)
502    Date: 2003/08/28 17:00:00 CDT
503    Tag: blume-20030828-intinf
504    Description:
505    
506    This is a major update which comes with a version number bump
507    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
508    and an implementation of IntInf in the Basis.
509    
510    There are a fairly large number of related changes and updates throughout
511    the system:
512    
513    Basis:
514            - Implemented IntInf.
515            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
516            - Added some missing Real64 operations, most notably Real.toLargeInt.
517            - Added FixedInt as a synonym for Int32.
518    
519    compiler:
520            * Added support for a built-in intinf type.
521              - literals
522              - pattern matching
523              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
524              - overloading on literals and operations
525    
526            This required adding a primitive type intinf, some additional
527            primops, and implementations for several non-trivial intinf
528            operations in Core.  (The intinf type is completely abstract
529            to the compiler; all operations get delegated back to the Core.)
530    
531            * Intinf equality is handled by polyequal.  However, the compiler
532            does not print its usual warning in this case (since polyequal
533            is the right thing to do there).
534    
535            * Improved the organization of structure InlineT.
536    
537            * A word about conversion primops:
538            If conversions involving intinf do not cancel out during
539            CPS contract, then the compiler must insert calls to Core functions.
540            Since all core access must be resolved already during the FLINT
541            translate phase, it would be too late a the time of CPS contract
542            to add new Core calls.  For this reason, conversion primops
543            for intinf carry two arguments:  1. the numeric argument that
544            they are supposed to convert, and 2. the Core function that
545            can help with this conversion if necessary.  If CPS contract
546            eliminates a primop, then the associated Core function becomes
547            dead and goes away.  Intinf conversion primops that do not get
548            eliminated by CPS contract get rewritten into calls of their
549            core functions by a separate, new phase.
550    
551    interactive system:
552            - Control.Print.intinfDepth controls max length of intinf constants
553            being printed. (Analogous to Control.Print.stringDepth.)
554            - Cleanup in printutil and pputil: got rid of unused stuff and
555            duplicates; replaced some of the code with code that makes better
556            use of library functionality.
557    
558    CM:
559            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
560            are no longer silent.
561    
562    CKIT:
563            Fixed mismatched uses of Int32 and LargeInt.  I always decided
564            in favor of LargeInt -- which is now the same as IntInf.
565            CKIT-knowledgable people should check whether this is what's
566            intended and otherwise change things back to using Int32 or
567            FixedInt.
568    
569    Throughout the code:
570            Started using IntInf.int literals and built-in operations
571            (e.g., comparison with 0) where this seems appropriate.
572    
573    
574    ----------------------------------------------------------------------
575    Name: Dave MacQueen (dbm@cs.uchicago.edu)
576    Date: 2003/08/13 11:36:00 CDT
577    Tag: dbm-20030813-mcz-merge1
578    Description:
579    
580    Merging changes from the mcz-branch development branch into trunk.
581    These changes involve replacement of the emulated old prettyprinter
582    interface with direct use of the SML/NJ Lib PP library, and fixing
583    of a couple of bugs (895, 1186) relating to error messages.  A new
584    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
585    has been added.
586    
587    ----------------------------------------------------------------------
588    Name: Matthias Blume (blume (at) tti - c (dot) org)
589    Date: 2003/08/11 15:45:00 CDT
590    Tag: blume-20030811-windows
591    Description:
592    
593    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
594    
595        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
596    
597    This patch restores SML/NJ's ability to run under win32.  There are a
598    number of changes, including fixes for several bugs that had gone
599    unnoticed until now:
600    
601      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
602        trying to run the win32 version from within cygwin.)
603    
604      - There are a number of simple .bat scripts that substitute for their
605        corresponding Unix shell-scripts.  (See below.)
606    
607      - The internals of ml-build have been modified slightly.  The main
608        difference is that instead of calling ".link-sml" (or link-sml.bat)
609        using OS.Process.system, the ML process delegates this task back
610        to the script.  Otherwise problems arise in mixed environments such
611        as Cygwin where scripts look and work like Unix scripts, but
612        where OS.Process.system cannot run them.
613    
614      - In CM, the srcpath pickler used native pathname syntax -- which
615        is incorrect in the case of cross-compilation.  The new pickle format
616        is independent of platform-specific naming conventions.
617    
618      - Path configuration files (such as lib/pathconfig) can now choose
619        between native and standard syntax.  Placing a line of the form
620    
621            standard!
622    
623        into the file causes all subsequent paths to be interpreted using
624        CM standard pathname syntax (= Unix conventions); a line
625    
626            native!
627    
628        switches back to native style.  This was needed so that
629        path config files can be written portably, see src/system/pathconfig.
630    
631      - Runtime system:
632    
633          - win32-filesys.c:  get_file_time and set_file_time now
634            access modification time, not creation time.
635    
636          - I/O code made aware of new array representation.
637    
638          - Bug fixes in X86.prim.masm.
639    
640       - src/system/makeml made aware of win32. (For use under cygwin
641            and other Unix-environments for windows.)
642    
643       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
644         which caused BinIO.inputAll to fail consistently.
645    
646    .bat scripts:
647    
648       Windows .bat scripts assume that SMLNJ_HOME is defined.
649    
650       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
651         applications (sml, ml-yacc, ml-lex).
652       - ml-build.bat: analogous to ml-build.
653       - config\install.bat:  Analogous to config/install.sh.  This requires
654         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
655         (nmake etc. must be on the path, and vcvars32 must have been run.)
656         Moreover, sources for ml-lex and ml-yacc need to exist under src,
657         and the bootfile hierarchy must have been unpacked under
658         sml.boot.x86-win32.
659         The script is very primitive and does a poor job at error checking.
660         It only installs the base system, ml-lex, and ml-yacc.  No other
661         libraries are being installed (i.e., you get only those that
662         are part of the compiler.)
663       - link-sml.bat:  analogous to .link-sml, but not currently used
664    
665    Unrelated bug fixes:
666    
667      - ml-nlffigen now exports structures ST_* corresponding to incomplete
668        types.
669      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
670        otherwise.)
671    
672    ----------------------------------------------------------------------
673    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
674    Date: 2003/06/17
675    Tag: macqueen-20030617-bug895
676    Description:
677    
678    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
679    Tag will be used for new development branch (mcz-branch) for
680    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
681    
682    ----------------------------------------------------------------------
683    Name: Matthias Blume (blume (at) tti - c (dot) org)
684    Date: 2003/05/27 16:55:00 CDT
685    Tag: blume-20030527-polyeq
686    Description:
687    
688    Tried to eliminated most cases of polymorphic equality.
689    
690    ----------------------------------------------------------------------
691    Name: Matthias Blume (blume (at) tti - c (dot) org)
692    Date: 2003/05/21 17:45:00 CDT
693    Tag: blume-20030517-complete
694    Description:
695    
696    Two changes:
697    
698      1. Added a flag for controlling whether non-exhaustive bindings will
699         be treated as errors (default is false).
700      2. Cleaned up the *entire* source tree so that CMB.make goes through
701         without a single non-exhaustive match- or bind warning.
702    
703    ----------------------------------------------------------------------
704    Name: Matthias Blume (blume (at) tti - c (dot) org)
705    Date: 2003/05/17 10:20:00 CDT
706    Tag: blume-20030517-absyn
707    Description:
708    
709    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
710    
711       This mainly affects the quality of error messages.  However, some
712       of the code is now more straightforward than before. (Treatment of
713       the above four constructs in translate.sml is much simpler than
714       the "macro-expansion" that was going on before.  Plus, the mach-
715       compiler no longer gets invoked just to be able to compile an
716       if-expression.)
717    
718    2. The ErrorMsg.Error exception is now caught and absorbed by the
719       interactive loop.
720    
721    ----------------------------------------------------------------------
722    Name: Allen Leung
723    Date: 2003/05/16 13:05:00 CDT
724    Tag: leunga-20030516-cygwin-runtime
725    Description:
726    
727        Ported the runtime system to cygwin, which uses the unix
728    x86-unix bin files.  Missing/buggy features:
729    
730       o getnetbyname, getnetbyaddr: these functions seem to be missing in
731         the Cygwin library.
732       o Ctrl-C handling may be flaky.
733       o Windows system calls and Windows I/O are not supported.
734    
735       A new set of binfiles is located at:
736    
737         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
738    
739       This is only needed for bootstrapping the cygwin version of smlnj.
740       Other x86 versions can use the existing binfiles.
741    
742    ----------------------------------------------------------------------
743    Name: Matthias Blume
744    Date: 2003/04/08 15:42:00 CDT
745    Tag: blume-20030408-listpair
746    Description:
747    
748    1. Added a target 'mlrisc' to installer.
749    
750    2. Added missing elements to structure ListPair.
751    
752    ----------------------------------------------------------------------
753    Name: Allen Leung
754    Date: 2003/01/07 10:40:00 EST
755    Tag: leunga-20030107-int-rem
756    Description:
757    
758    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
759    The arguments to the SUBL instruction were swapped.
760    
761    ----------------------------------------------------------------------
762    Name: Matthias Blume
763    Date: 2002/12/12 16:25:00 EST
764    Tag: blume-20021212-risc-ra
765    Description:
766    
767    Fixed a serious bug in the rewrite code for FP spilling/reloading that
768    sent the RA into an infinite loop when floating point registers get
769    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
770    110.37 and 110.38.)
771    There was another set of potential problems related to the handling of
772    MLRISC annotations (but those did not yet cause real problems, apparently).
773    
774    ----------------------------------------------------------------------
775    Name: Matthias Blume
776    Date: 2002/12/06 22:40:00 EST
777    Tag: blume-20021206-cm-fileid
778    Description:
779    
780    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
781    This fixes the problem of CM getting confused by files that suddenly
782    change their identity (e.g., by getting unlinked and recreated by some
783    text editor such as vi).  There might be a better/cheaper/cleaner way
784    of doing this, but for now this will have to do.
785    
786    ----------------------------------------------------------------------
787    Name: Matthias Blume
788    Date: 2002/10/28 09:50:00 EST
789    Tag: blume-20021028-typecheck
790    Description:
791    
792    Exported structure Typecheck from $smlnj/viscomp/core.cm.
793    
794    ----------------------------------------------------------------------
795    Name: Matthias Blume
796    Date: 2002/10/17 09:10:00 EDT
797    Tag: Release_110_42
798    Description:
799    
800    In good old tradition, there has been a slight hiccup so that we have
801    to patch 110.42 after the fact.   The old release tag has been replaced
802    (see below).
803    
804    The change solves a problem with two competing approaches the
805    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
806    each other's way.
807    
808    This change only affects the runtime system code and the installer script.
809    (No new bootfiles.)
810    
811    ----------------------------------------------------------------------
812    Name: Matthias Blume
813    Date: 2002/10/16 12:00:00 EDT
814    Tag: Release_110_42_removed
815    Description:
816    
817    New working release.  New bootfiles.
818    
819    ----------------------------------------------------------------------
820    Name: Matthias Blume
821    Date: 2002/10/10 13:10:00 EDT
822    Tag: blume-20021010-ppc-divs
823    Description:
824    
825    The mltree operator DIVS must be implemented with an overflow check on
826    the PPC because the hardware indicates divide-by-zero using "overflow" as
827    well.
828    
829    ----------------------------------------------------------------------
830    Name: Matthias Blume
831    Date: 2002/07/23 11:45:00
832    Tag: blume-20020723-smlnj-home
833    Description:
834    
835    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
836    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
837    CM_PATHCONFIG is also set), the path configuration file is assumed
838    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
839    the entire tree to some other place and everything will "just work".
840    
841    (Companion commands such as ml-build and ml-makedepend also sense this
842    variable.)
843    
844    ----------------------------------------------------------------------
845    Name: Matthias Blume
846    Date: 2002/07/12 21:19:00 EDT
847    Tag: blume-20020712-liveness
848    Description:
849    
850    Exported two useful "step" functions from liveness module (MLRISC).
851    
852    ----------------------------------------------------------------------
853    Name: Matthias Blume
854    Date: 2002/07/05 16:00 EDT
855    Tag: Release_110_41
856    Description:
857    
858    New working release.  New bootfiles.
859    
860    ----------------------------------------------------------------------
861    Name: Matthias Blume
862    Date: 2002/07/05 10:25:00 EDT
863    Tag: blume-20020705-btimp
864    Description:
865    
866    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
867    other clients can set up backtracing support.
868    
869    ----------------------------------------------------------------------
870    Name: Matthias Blume
871    Date: 2002/06/25 17:23:00 EDT
872    Tag: blume-20020625-fpmax
873    Description:
874    
875    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
876    case.  (The sense of the isNaN test was reversed -- which made min and
877    max always return their first argument.)
878    
879    ----------------------------------------------------------------------
880    Name: Matthias Blume
881    Date: 2002/06/11
882    Tag: blume-20020611-unixpath
883    Description:
884    
885    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
886    
887    ----------------------------------------------------------------------
888    Name: Matthias Blume
889    Date: 2002/06/10 16:35:00 EDT
890    Tag: blume-20020610-ieeereal
891    Description:
892    
893    I back-ported my implementation of IEEEReal.fromString from the
894    idlbasis-devel branch so that we can test it.
895    
896    Another small change is that ppDec tries to give more information
897    than just "<sig>" in the case of functors.  However, this code is
898    broken in some mysterious way if the functor's body's signature
899    has not been declared by ascription but gets inferred from the
900    implementation.  This needs fixing...
901    
902    ----------------------------------------------------------------------
903    Name: Matthias Blume
904    Date: 2002/05/31
905    Tag: blume-20020531-btrace-mode
906    Description:
907    
908    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
909    the wayside when I switched over to using Controls everywhere.)
910    
911    ----------------------------------------------------------------------
912    Name: Lal George
913    Date: 2002/05/23 12:21:40 EDT
914    Tag: george-20020523-visual-labels
915    Description:
916    
917    Labels are now displayed in the graphical output to make
918    the fall-through and target blocks obvious.
919    
920    ----------------------------------------------------------------------
921    Name: Matthias Blume
922    Date: 2002/05/22 11:03:00 EDT
923    Tag: blume-20020522-shrink
924    Description:
925    
926    John tweaked yesterday's fix for 1131 to handle an out-of-memory
927    situation that comes up when allocating huge arrays.
928    
929    ----------------------------------------------------------------------
930    Name: Matthias Blume
931    Date: 2002/05/21 16:00:00 EDT
932    Tag: Release_110_40
933    Description:
934    
935    New working release (110.40).  New bootfiles.
936    
937    [Also: John Reppy fixed GC bug 1131.]
938    
939    ----------------------------------------------------------------------
940    Name: Matthias Blume
941    Date: 2002/05/21 12:35:00 EDT
942    Tag: blume-20020521-cmdoc
943    Description:
944    
945    CM documentation update.
946    
947    ----------------------------------------------------------------------
948    Name: Matthias Blume
949    Date: 2002/05/21 10:55:00 EDT
950    Tag: blume-20020521-misc
951    Description:
952    
953    - John tweaked runtime to be silent on heap export (except when
954      GC messages are on).
955    - I added a few more things (cross-compiling versions of CMB) to
956      config/preloads (as suggestions).
957    
958    ----------------------------------------------------------------------
959    Name: Matthias Blume
960    Date: 2002/05/20 22:25:00 EDT
961    Tag: blume-20020520-controls
962    Description:
963    
964    - Added ControlUtil structure to control-lib.cm.
965    - Use it throughout.
966    - Used Controls facility to define MLRISC controls (as opposed to
967      registering MLRISC control ref cells with Controls after the fact)
968    - Fixed messed-up controls priorities.
969    
970    * Removed again all the stuff from config/preloads that one wouldn't
971      be able to preload at the time the initial heap image is built.
972      (Many libraries, e.g., CML, do not exist yet at this time.  The
973       only libraries that can be preloaded via config/preloads are those
974       that come bundled with the bootfiles.)
975    
976    ----------------------------------------------------------------------
977    Name: Matthias Blume
978    Date: 2002/05/20 10:59:00 EDT
979    Tag: blume-20020520-preloads
980    Description:
981    
982    Added a lot of commented-out suggestions for things to be included
983    in config/preloads.
984    
985    ----------------------------------------------------------------------
986    Name: Allen Leung
987    Date: 2002/05/18 14:20:00 EDT
988    Tag: leunga-20020518-mdl
989    Description:
990    
991    o Made the mdl tool stuff compile and run again.
992    o I've disabled all the stuff that depends on RTL specifications; they
993      are all badly broken anyway.
994    
995    ----------------------------------------------------------------------
996    Name: Matthias Blume
997    Date: 2002/05/17 16:49:00 EDT
998    Tag: blume-20020517-controls
999    Description:
1000    
1001    0. John Reppy made several modifications to the SML/NJ library.
1002       In particular, there is a shiny new controls-lib.cm.
1003    
1004    1. Pushed new controls interface through compiler so that everything
1005       compiles again.
1006    
1007    2. Added FormatComb and FORMAT_COMB to the CML version of the
1008       SML/NJ library (so that CML compiles again).
1009    
1010    3. Modified init scripts because XXX_DEFAULT environment variables
1011       are no longer with us.  (Boot-time initialization is now done
1012       using the same environment variables that are also used for
1013       startup-time initialization of controls.)
1014    
1015    ----------------------------------------------------------------------
1016    Name: Lal George
1017    Date: 2002/05/15 09:20:10 EDT
1018    Tag: george-20020515-pseudo-op-decls
1019    Description:
1020    
1021            All pseudo-ops emitted before the first segment declaration
1022    such as TEXT, DATA, and BSS directives are assumed to be global
1023    declarations and are emitted first in the assembly file. This is
1024    useful in a number of situations where one has pseudo-ops that are not
1025    specific to any segment, and also works around the constraint that one
1026    cannot have client pseudo-ops in the TEXT segment.
1027    
1028            Because no segment is associated with these declarations it is
1029    an error to allocate any space or objects before the first segment
1030    directive and an exception will be raised. However, we cannot make
1031    this check for client pseudo-ops.
1032    
1033            These top level declarations are a field in the CFG graph_info.
1034    In theory you can continue to add to this field after the CFG has been
1035    built -- provided you know what you are doing;-)
1036    
1037    ----------------------------------------------------------------------
1038    Name: Matthias Blume
1039    Date: 2002/05/13 16:40:00 EDT
1040    Tag: blume-20020513-pp-etc
1041    Description:
1042    
1043    A few minor bugfixes:
1044    
1045      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1046        (Bogus printouts for redefined bindings still occur.  Compiler
1047        bug should no longer occur now.  We need to redo the prettyprinter
1048        from scratch.)
1049    
1050      - CM pathname printer now also adds escape sequences for ( and )
1051    
1052      - commend and docu fixes for ml-nlffi
1053    
1054    ----------------------------------------------------------------------
1055    Name: Matthias Blume
1056    Date: 2002/05/10 16:40:00 EDT
1057    Tag: blume-20020510-erg-textio
1058    Description:
1059    
1060    Applied the following bugfix provided by Emden Gansner:
1061    
1062        Output is corrupted when outputSubstr is used rather than output.
1063    
1064        The problem occurs when a substring
1065    
1066            ss = (s, dataStart, dataLen)
1067    
1068        where dataStart > 0, fills a stream buffer with avail bytes left.
1069        avail bytes of s, starting at index dataStart, are copied into the
1070        buffer, the buffer is flushed, and then the remaining dataLen-avail
1071        bytes of ss are copied into the beginning of the buffer. Instead of
1072        starting this copy at index dataStart+avail in s, the current code
1073        starts the copy at index avail.
1074    
1075        Fix:
1076        In text-io-fn.sml, change line 695 from
1077             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1078        to
1079             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1080    
1081    ----------------------------------------------------------------------
1082    Name: Matthias Blume
1083    Date: 2002/04/12 13:55:00 EDT
1084    Tag: blume-20020412-assyntax
1085    Description:
1086    
1087    1. Grabbed newer assyntax.h from the XFree86 project.
1088    2. Fiddled with how to compile X86.prim.asm without warnings.
1089    3. (Very) Minor cleanup in CM.
1090    
1091    ----------------------------------------------------------------------
1092    Name: Matthias Blume
1093    Date: 2002/04/01 (no joke!) 17:07:00 EST
1094    Tag: blume-20020401-x86div
1095    Description:
1096    
1097    Added full support for div/mod/rem/quot on the x86, using the machine
1098    instruction's two results (without clumsily recomputing the remainder)
1099    directly where appropriate.
1100    
1101    Some more extensive power-of-two support was added to the x86 instruction
1102    selector (avoiding expensive divs, mods, and muls where they can be
1103    replaced with cheaper shifts and masks).  However, this sort of thing
1104    ought to be done earlier, e.g., within the CPS optimizer so that
1105    all architectures benefit from it.
1106    
1107    The compiler compiles to a fixed point, but changes might be somewhat
1108    fragile nevertheless.  Please, report any strange things that you might
1109    see wrt. div/mod/quot/rem...
1110    
1111    ----------------------------------------------------------------------
1112    Name: Matthias Blume
1113    Date: 2002/03/29 17:22:00
1114    Tag: blume-20020329-div
1115    Description:
1116    
1117    Fixed my broken div/mod logic.  Unfortunately, this means that the
1118    inline code for div/mod now has one more comparison than before.
1119    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1120    The problem was with quotient = 0, because that alone does not tell
1121    us which way the rounding went.  One then has to look at whether
1122    remainder and divisor have the same sign...  :(
1123    
1124    Anyway, I replaced the bootfiles with fresh ones...
1125    
1126    ----------------------------------------------------------------------
1127    Name: Matthias Blume
1128    Date: 2002/03/29 14:10:00 EST
1129    Tag: blume-20020329-inlprims
1130    Description:
1131    
1132    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1133    
1134    Primops have changed. This means that the bin/boot-file formats have
1135    changed as well.
1136    
1137    To make sure that there is no confusion, I made a new version.
1138    
1139    
1140    CHANGES:
1141    
1142    * removed REMT from mltree (remainder should never overflow).
1143    
1144    * added primops to deal with divisions of all flavors to the frontend
1145    
1146    * handled these primops all the way through so they map to their respective
1147      MLRISC support
1148    
1149    * used these primops in the implementation of Int, Int32, Word, Word32
1150    
1151    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1152    
1153    * parameterized INLMIN, INLMAX, and INLABS by a numkind
1154    
1155    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1156      floating point
1157    
1158    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1159      and Real (but Real.abs maps to a separate floating-point-only primop)
1160    
1161    
1162    TODO items:
1163    
1164    * Hacked Alpha32 instruction selection, disabling the selection of REMx
1165      instructions because the machine instruction encoder cannot handle
1166      them.  (Hppa, PPC, and Sparc instruction selection did not handle
1167      REM in the first place, and REM is supported by the x86 machine coder.)
1168    
1169    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1170      selection phase.  (The two can be streamlined because the hardware
1171      delivers both quotient and remainder at the same time anyway.)
1172    
1173    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1174      (Currently the behavior is inconsistent both across architectures and
1175      wrt. the draft Basis spec.)
1176    
1177    * Word8 should eventually be handled natively, too.
1178    
1179    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1180      as if there currently is no execution path that could trigger it in
1181      SML/NJ.  (The assumptions underlying functions arith and promotable do not
1182      hold for things like multiplication and division.)
1183    
1184    ----------------------------------------------------------------------
1185    Name: Matthias Blume
1186    Date: 2002/03/27 16:27:00 EST
1187    Tag: blume-20020327-mlrisc-divisions
1188    Description:
1189    
1190    Added support for all four division operations (ML's div, mod, quot,
1191    and rem) to MLRISC.  In the course of doing so, I also rationalized
1192    the naming (no more annoying switch-around of DIV and QUOT), by
1193    parameterizing the operation by div_rounding_mode (which can be either
1194    DIV_TO_ZERO or DIV_TO_NEGINF).
1195    
1196    The generic MLTreeGen functor takes care of compiling all four
1197    operations down to only round-to-zero div.
1198    
1199    Missing pieces:
1200    
1201      * Doing something smarter than relying on MLTreeGen on architectures
1202        like, e.g., the x86 where hardware division delivers both quotient and
1203        remainder at the same time.  With this, the implementation of the
1204        round-to-neginf operations could be further streamlined.
1205    
1206      * Remove inlining support for div/mod/rem from the frontend and replace it
1207        with primops that get carried through to the backend.  Do this for all
1208        int and word types.
1209    
1210    ----------------------------------------------------------------------
1211    Name: Matthias Blume
1212    Date: 2002/03/25 17:25:00 EST
1213    Tag: blume-20020325-divmod
1214    Description:
1215    
1216    I improved (hopefully without breaking them) the implementation of Int.div,
1217    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1218    advantage of the following observations:
1219    
1220      Let  q = x quot y      r = x rem y
1221           d = x div  y      m = x mod y
1222    
1223    where "quot" is the round-to-zero version of integer division that
1224    hardware usually provides.  Then we have:
1225    
1226         r = x - q * y        where neither the * nor the - will overflow
1227         d = if q >= 0 orelse x = q * y then q else q - 1
1228                              where neither the * nor the - will overflow
1229         m = if q >= 0 orelse r = 0 then r else r + y
1230                              where the + will not overflow
1231    
1232    This results in substantial simplification of the generated code.
1233    The following table shows the number of CFG nodes and edges generated
1234    for
1235            fun f (x, y) = x OPER y
1236            (* with OPER \in div, mod, quot, rem *)
1237    
1238    
1239        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1240        --------------------------------------------------------
1241         div |         24 |         39 |         12 |         16
1242         mod |         41 |         71 |         12 |         16
1243        quot |          8 |         10 |          8 |         10
1244         rem |         10 |         14 |          8 |         10
1245    
1246    
1247    ----------------------------------------------------------------------
1248    Name: Matthias Blume
1249    Date: 2002/03/25 22:06:00 EST
1250    Tag: blume-20020325-cprotobug
1251    Description:
1252    
1253    Fixed a bug in cproto (c prototype decoder).
1254    
1255    ----------------------------------------------------------------------
1256    Name: Matthias Blume
1257    Date: 2002/03/25 16:00:00 EST
1258    Tag: blume-20020325-raw-primops
1259    Description:
1260    
1261    I did some cleanup to Allen's new primop code and
1262    replaced yesterday's bootfiles with new ones.
1263    (But they are stored in the same place.)
1264    
1265    ----------------------------------------------------------------------
1266    Name: Matthias Blume
1267    Date: 2002/03/24 22:40:00 EST
1268    Tag: blume-20020324-bootfiles
1269    Description:
1270    
1271    Made the bootfiles that Allen asked for.
1272    
1273    ----------------------------------------------------------------------
1274    Name: Allen Leung
1275    Date: 2002/03/23 15:50:00 EST
1276    Tag: leunga-20020323-flint-cps-rcc-primops
1277    Description:
1278    
1279      1. Changes to FLINT primops:
1280    
1281        (* make a call to a C-function;
1282         * The primop carries C function prototype information and specifies
1283         * which of its (ML-) arguments are floating point. C prototype
1284         * information is for use by the backend, ML information is for
1285         * use by the CPS converter. *)
1286      | RAW_CCALL of { c_proto: CTypes.c_proto,
1287                       ml_args: ccall_type list,
1288                       ml_res_opt: ccall_type option,
1289                       reentrant : bool
1290                     } option
1291       (* Allocate uninitialized storage on the heap.
1292        * The record is meant to hold short-lived C objects, i.e., they
1293        * are not ML pointers.  With the tag, the representation is
1294        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1295        *)
1296      | RAW_RECORD of {tag:bool,sz:int}
1297      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1298    
1299      2.  These CPS primops are now overloaded:
1300    
1301           rawload of {kind:numkind}
1302           rawstore of {kind:numkind}
1303    
1304          The one argument form is:
1305    
1306             rawload {kind} address
1307    
1308          The two argument form is:
1309    
1310             rawload {kind} [ml object, byte-offset]
1311    
1312      3. RAW_CCALL/RCC now takes two extra arguments:
1313    
1314         a. The first is whether the C call is reentrant, i.e., whether
1315            ML state should be saved and restored.
1316         b. The second argument is a string argument specifying the name of
1317            library and the C function.
1318    
1319         These things are currently not handled in the code generator, yet.
1320    
1321      4. In CProto,
1322    
1323         An encoding type of "bool" means "ml object" and is mapped into
1324         C prototype of PTR.  Note that "bool" is different than "string",
1325         even though "string" is also mapped into PTR, because "bool"
1326         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1327    
1328      5. Pickler/unpicker
1329    
1330         Changed to handle RAW_RECORD and newest RAW_CCALL
1331    
1332      6. MLRiscGen,
1333    
1334         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1335         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1336            in the file CodeGen/cpscompile/cps-c-calls.sml
1337    
1338      7. Added the conditional move operator
1339    
1340             condmove of branch
1341    
1342         to cps.  Generation of this is still buggy so it is currently
1343         disabled.
1344    
1345    ----------------------------------------------------------------------
1346    Name: Lal George
1347    Date: 2002/03/22 14:18:25 EST
1348    Tag: george-20020322-cps-branch-prob
1349    Description:
1350    
1351    Implemented the Ball-Larus branch prediction-heuristics, and
1352    incorporated graphical viewers for control flow graphs.
1353    
1354    Ball-Larus Heuristics:
1355    ---------------------
1356    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1357    
1358    By design it uses the Dempster-Shafer theory for combining
1359    probabilities.  For example, in the function:
1360    
1361        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1362    
1363    the ball-larus heuristics predicts that the n=0 is unlikely
1364    (OH-heuristic), and the 'then' branch is unlikely because of the
1365    RH-heuristic -- giving the 'then' branch an even lower combined
1366    probability using the Dempster-Shafer theory.
1367    
1368    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1369    probability of the 'then' branch because of the loop in the else
1370    branch.
1371    
1372    
1373    Graphical Viewing:
1374    ------------------
1375    I merely plugged in Allen's graphical viewers into the compiler. The
1376    additional code is not much. At the top level, saying:
1377    
1378            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1379    
1380    will display the graphical view of the control flow graph just before
1381    back-patching.  daVinci must be in your path for this to work. If
1382    daVinci is not available, then the default viewer can be changed
1383    using:
1384    
1385            Control.MLRISC.getString "viewer"
1386    
1387    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1388    course, these viewers must be in your path.
1389    
1390    The above will display the compilation unit at the level of clusters,
1391    many of which are small, boring, and un-interesting. Also setting:
1392    
1393            Control.MLRISC.getInt "cfg-graphical-view_size"
1394    
1395    will display clusters that are larger than the value set by the above.
1396    
1397    
1398    ----------------------------------------------------------------------
1399    Name: Matthias Blume
1400    Date: 2002/03/21 22:20:00 EST
1401    Tag: blume-20020321-kmp-bugfix
1402    Description:
1403    
1404    Changed the interface to the KMP routine in PreString and fixed
1405    a minor bug in one place where it was used.
1406    
1407    ----------------------------------------------------------------------
1408    Name: Allen Leung
1409    Date: 2002/03/21 20:30:00 EST
1410    Tag: leunga-20020321-cfg
1411    Description:
1412    
1413      Fixed a potential problem in cfg edge splitting.
1414    
1415    ----------------------------------------------------------------------
1416    Name: Allen Leung
1417    Date: 2002/03/21 17:15:00 EST
1418    Tag: leunga-20020321-x86-fp-cfg
1419    Description:
1420    
1421      1. Recoded the buggy parts of x86-fp.
1422    
1423         a. All the block reordering code has been removed.
1424            We now depend on the block placement phases to do this work.
1425    
1426         b. Critical edge splitting code has been simplified and moved into the
1427            CFG modules, as where they belong.
1428    
1429         Both of these were quite buggy and complex.  The code is now much, much
1430         simpler.
1431    
1432      2. X86 backend.
1433    
1434         a. Added instructions for 64-bit support.  Instruction selection for
1435            64-bit has not been committed, however, since that
1436            requires changes to MLTREE which haven't been approved by
1437            Lal and John.
1438    
1439         b. Added support for FUCOMI and FUCOMIP when generating code for
1440            PentiumPro and above.  We only generate these instructions in
1441            the fast-fp mode.
1442    
1443         c. Added cases for JP and JNP in X86FreqProps.
1444    
1445      3. CFG
1446    
1447         CFG now has a bunch of methods for edge splitting and merging.
1448    
1449      4. Machine description.
1450    
1451         John's simplification of MLTREE_BASIS.fcond broke a few machine
1452         description things:
1453    
1454         rtl-build.{sig,sml} and hppa.mdl fixed.
1455    
1456         NOTE: the machine description stuff in the repository is still broken.
1457               Again, I can't put my fixes in because that involves
1458               changes to MLTREE.
1459    
1460    ----------------------------------------------------------------------
1461    Name: Matthias Blume
1462    Date: 2002/03/20 15:55:00 EST
1463    Tag: blume-20020320-kmp
1464    Description:
1465    
1466    Implemented Knuth-Morris-Pratt string matching in PreString and used
1467    it for String.isSubstring, Substring.isSubstring, and
1468    Substring.position.
1469    
1470    (Might need some stress-testing.  Simple examples worked fine.)
1471    
1472    ----------------------------------------------------------------------
1473    Name: Matthias Blume
1474    Date: 2002/03/19 16:37:00 EST
1475    Tag: blume-20020319-witnesses
1476    Description:
1477    
1478    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1479    
1480    This implements a generic mechanism for changing constness qualifiers
1481    anywhere within big C types without resorting to outright "casts".
1482    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1483    modify the constness at the outermost level.)
1484    The implementation of "convert" is based on the idea of "witness"
1485    values -- values that are not used by the operation but whose types
1486    "testify" to their applicability.  On the implementation side, "convert"
1487    is simply a projection (returning its second curried argument).  With
1488    cross-module inlining, it should not result in any machine code being
1489    generated.
1490    
1491    ----------------------------------------------------------------------
1492    Name: Matthias Blume
1493    Date: 2002/03/15 16:40:00 EST
1494    Tag: blume-20020315-basis
1495    Description:
1496    
1497    Provided (preliminary?) implementations for
1498    
1499      {String,Substring}.{concatWith,isSuffix,isSubstring}
1500    
1501    and
1502    
1503      Substring.full
1504    
1505    Those are in the Basis spec but they were missing in SML/NJ.
1506    
1507    ----------------------------------------------------------------------
1508    Name: Matthias Blume
1509    Date: 2002/03/14 21:30:00 EST
1510    Tag: blume-20020314-controls
1511    Description:
1512    
1513    Controls:
1514    ---------
1515    
1516    1. Factored out the recently-added Controls : CONTROLS stuff and put
1517       it into its own library $/controls-lib.cm.  The source tree for
1518       this is under src/smlnj-lib/Controls.
1519    
1520    2. Changed the names of types and functions in this interface, so they
1521       make a bit more "sense":
1522    
1523          module -> registry
1524          'a registry -> 'a group
1525    
1526    3. The interface now deals in ref cells only.  The getter/setter interface
1527       is (mostly) gone.
1528    
1529    4. Added a function that lets one register an already-existing ref cell.
1530    
1531    5. Made the corresponding modifications to the rest of the code so that
1532       everything compiles again.
1533    
1534    6. Changed the implementation of Controls.MLRISC back to something closer
1535       to the original.  In particular, this module (and therefore MLRISC)
1536       does not depend on Controls.  There now is some link-time code in
1537       int-sys.sml that registers the MLRISC controls with the Controls
1538       module.
1539    
1540    CM:
1541    ---
1542    
1543      * One can now specify the lambda-split aggressiveness in init.cmi.
1544    
1545    ----------------------------------------------------------------------
1546    Name: Allen Leung
1547    Date: 2002/03/13 17:30:00 EST
1548    Tag: leunga-20020313-x86-fp-unary
1549    Description:
1550    
1551    Bug fix for:
1552    
1553    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1554    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1555    > - fun f(x,(y,z)) = Real.~ y;
1556    > [autoloading]
1557    > [autoloading done]
1558    >       fchsl   (%eax), 184(%esp)
1559    > Error: MLRisc bug: X86MCEmitter.emitInstr
1560    >
1561    > uncaught exception Error
1562    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1563    
1564    The problem was that the code generator did not generate any fp registers
1565    in this case, and the ra didn't know that it needed to run the X86FP phase to
1566    translate the pseudo fp instruction.   This only happened with unary fp
1567    operators in certain situations.
1568    
1569    ----------------------------------------------------------------------
1570    Name: Matthias Blume
1571    Date: 2002/03/13 14:00:00 EST
1572    Tag: blume-20020313-overload-etc
1573    Description:
1574    
1575    1. Added _overload as a synonym for overload for backward compatibility.
1576       (Control.overloadKW must be true for either version to be accepted.)
1577    
1578    2. Fixed bug in install script that caused more things to be installed
1579       than what was requested in config/targets.
1580    
1581    3. Made CM aware of the (_)overload construct so that autoloading
1582       works.
1583    
1584    ----------------------------------------------------------------------
1585    Name: Matthias Blume
1586    Date: 2002/03/12 22:03:00 EST
1587    Tag: blume-20020312-url
1588    Description:
1589    
1590    Forgot to update BOOT and srcarchiveurl.
1591    
1592    ----------------------------------------------------------------------
1593    Name: Matthias Blume
1594    Date: 2002/03/12 17:30:00 EST
1595    Tag: blume-20020312-version110392
1596    Description:
1597    
1598    Yet another version number bump (because of small changes to the
1599    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1600    
1601    Changes:
1602    
1603      The new pid generation scheme described a few weeks ago was overly
1604      complicated.  I implemented a new mechanism that is simpler and
1605      provides a bit more "stability":  Once CM has seen a compilation
1606      unit, it keeps its identity constant (as long as you do not delete
1607      those crucial CM/GUID/* files).  This means that when you change
1608      an interface, compile, then go back to the old interface, and
1609      compile again, you arrive at the original pid.
1610    
1611      There now also is a mechanism that instructs CM to use the plain
1612      environment hash as a module's pid (effectively making its GUID
1613      the empty string).  For this, "noguid" must be specified as an
1614      option to the .sml file in question within its .cm file.
1615      This is most useful for code that is being generated by tools such
1616      as ml-nlffigen (because during development programmers tend to
1617      erase the tool's entire output directory tree including CM's cached
1618      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1619      revert to the old, broken behavior of SML/NJ, but in specific cases
1620      where there is no danger of interface confusion, its use is ok
1621      (I think).
1622    
1623      ml-nlffigen by default generates "noguid" annotations.  They can be
1624      turned off by specifying -guid in its command line.
1625    
1626    ----------------------------------------------------------------------
1627    Name: Lal George
1628    Date: 2002/03/12 12 14:42:36 EST
1629    Tag: george-20020312-frequency-computation
1630    Description:
1631    
1632    Integrated jump chaining and static block frequency into the
1633    compiler. More details and numbers later.
1634    
1635    ----------------------------------------------------------------------
1636    Name: Lal George
1637    Date: 2002/03/11 11 22:38:53 EST
1638    Tag: george-20020311-jump-chain-elim
1639    Description:
1640    
1641    Tested the jump chain elimination on all architectures (except the
1642    hppa).  This is on by default right now and is profitable for the
1643    alpha and x86, however, it may not be profitable for the sparc and ppc
1644    when compiling the compiler.
1645    
1646    The gc test will typically jump to a label at the end of the cluster,
1647    where there is another jump to an external cluster containing the actual
1648    code to invoke gc. This is to allow factoring of common gc invocation
1649    sequences. That is to say, we generate:
1650    
1651            f:
1652               testgc
1653               ja   L1      % jump if above to L1
1654    
1655            L1:
1656               jmp L2
1657    
1658    
1659    After jump chain elimination the 'ja L1' instructions is converted to
1660    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1661    up being implemented in their long form (if L2 is far away) using:
1662    
1663            jbe     L3      % jump if below or equal to L3
1664            jmp     L2
1665         L3:
1666            ...
1667    
1668    
1669    For large compilation units L2  may be far away.
1670    
1671    
1672    ----------------------------------------------------------------------
1673    Name: Matthias Blume
1674    Date: 2002/03/11 13:30:00 EST
1675    Tag: blume-20020311-mltreeeval
1676    Description:
1677    
1678    A functor parameter was missing.
1679    
1680    ----------------------------------------------------------------------
1681    Name: Allen Leung
1682    Date: 2002/03/11 10:30:00 EST
1683    Tag: leunga-20020311-runtime-string0
1684    Description:
1685    
1686       The representation of the empty string now points to a
1687    legal null terminated C string instead of unit.  It is now possible
1688    to convert an ML string into C string with InlineT.CharVector.getData.
1689    This compiles into one single machine instruction.
1690    
1691    ----------------------------------------------------------------------
1692    Name: Allen Leung
1693    Date: 2002/03/10 23:55:00 EST
1694    Tag: leunga-20020310-x86-call
1695    Description:
1696    
1697       Added machine generation for CALL instruction (relative displacement mode)
1698    
1699    ----------------------------------------------------------------------
1700    Name: Matthias Blume
1701    Date: 2002/03/08 16:05:00
1702    Tag: blume-20020308-entrypoints
1703    Description:
1704    
1705    Version number bumped to 110.39.1.  NEW BOOTFILES!
1706    
1707    Entrypoints: non-zero offset into a code object where execution should begin.
1708    
1709    - Added the notion of an entrypoint to CodeObj.
1710    - Added reading/writing of entrypoint info to Binfile.
1711    - Made runtime system bootloader aware of entrypoints.
1712    - Use the address of the label of the first function given to mlriscGen
1713      as the entrypoint.  This address is currently always 0, but it will
1714      not be 0 once we turn on block placement.
1715    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
1716      with entry points) from mlriscGen.
1717    
1718    ----------------------------------------------------------------------
1719    Name: Allen Leung
1720    Date: 2002/03/07 20:45:00 EST
1721    Tag: leunga-20020307-x86-cmov
1722    Description:
1723    
1724       Bug fixes for CMOVcc on x86.
1725    
1726       1. Added machine code generation for CMOVcc
1727       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1728       3. CMOVcc cannot have an immediate operand as argument.
1729    
1730    ----------------------------------------------------------------------
1731    Name: Matthias Blume
1732    Date: 2002/03/07 16:15:00 EST
1733    Tag: blume-20020307-controls
1734    Description:
1735    
1736    This is a very large but mostly boring patch which makes (almost)
1737    every tuneable compiler knob (i.e., pretty much everything under
1738    Control.* plus a few other things) configurable via both the command
1739    line and environment variables in the style CM did its configuration
1740    until now.
1741    
1742    Try starting sml with '-h' (or, if you are brave, '-H')
1743    
1744    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1745    implements the underlying generic mechanism.
1746    
1747    The interface to some of the existing such facilities has changed somewhat.
1748    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1749    (The getFoo interface is still there for backward-compatibility, but its
1750    use is deprecated.)
1751    
1752    The ml-build script passes -Cxxx=yyy command-line arguments through so
1753    that one can now twiddle the compiler settings when using this "batch"
1754    compiler.
1755    
1756    TODO items:
1757    
1758    We should go through and throw out all controls that are no longer
1759    connected to anything.  Moreover, we should go through and provide
1760    meaningful (and correct!) documentation strings for those controls
1761    that still are connected.
1762    
1763    Currently, multiple calls to Controls.new are accepted (only the first
1764    has any effect).  Eventually we should make sure that every control
1765    is being made (via Controls.new) exactly once.  Future access can then
1766    be done using Controls.acc.
1767    
1768    Finally, it would probably be a good idea to use the getter-setter
1769    interface to controls rather than ref cells.  For the time being, both
1770    styles are provided by the Controls module, but getter-setter pairs are
1771    better if thread-safety is of any concern because they can be wrapped.
1772    
1773    *****************************************
1774    
1775    One bug fix: The function blockPlacement in three of the MLRISC
1776    backpatch files used to be hard-wired to one of two possibilities at
1777    link time (according to the value of the placementFlag).  But (I
1778    think) it should rather sense the flag every time.
1779    
1780    *****************************************
1781    
1782    Other assorted changes (by other people who did not supply a HISTORY entry):
1783    
1784    1. the cross-module inliner now works much better (Monnier)
1785    2. representation of weights, frequencies, and probabilities in MLRISC
1786       changed in preparation of using those for weighted block placement
1787       (Reppy, George)
1788    
1789    ----------------------------------------------------------------------
1790    Name: Lal George
1791    Date: 2002/03/07 14:44:24 EST 2002
1792    Tag: george-20020307-weighted-block-placement
1793    
1794    Tested the weighted block placement optimization on all architectures
1795    (except the hppa) using AMPL to generate the block and edge frequencies.
1796    Changes were required in the machine properties to correctly
1797    categorize trap instructions. There is an MLRISC flag
1798    "weighted-block-placement" that can be used to enable weighted block
1799    placement, but this will be ineffective without block/edge
1800    frequencies (coming soon).
1801    
1802    
1803    ----------------------------------------------------------------------
1804    Name: Lal George
1805    Date: 2002/03/05 17:24:48 EST
1806    Tag: george-20020305-linkage-cluster
1807    
1808    In order to support the block placement optimization, a new cluster
1809    is generated as the very first cluster (called the linkage cluster).
1810    It contains a single jump to the 'real' entry point for the compilation
1811    unit. Block placement has no effect on the linkage cluster itself, but
1812    all the other clusters  have full freedom in the manner in which they
1813    reorder blocks or functions.
1814    
1815    On the x86 the typical linkage code that is generated is:
1816       ----------------------
1817            .align 2
1818       L0:
1819            addl    $L1-L0, 72(%esp)
1820            jmp     L1
1821    
1822    
1823            .align  2
1824       L1:
1825       ----------------------
1826    
1827    72(%esp) is the memory location for the stdlink register. This
1828    must contain the address of the CPS function being called. In the
1829    above example, it contains the address of  L0; before
1830    calling L1 (the real entry point for the compilation unit), it
1831    must contain the address for L1, and hence
1832    
1833            addl $L1-L0, 72(%esp)
1834    
1835    I have tested this on all architectures except the hppa.The increase
1836    in code size is of course negligible
1837    
1838    ----------------------------------------------------------------------
1839    Name: Allen Leung
1840    Date: 2002/03/03 13:20:00 EST
1841    Tag: leunga-20020303-mlrisc-tools
1842    
1843      Added #[ ... ] expressions to mlrisc tools
1844    
1845    ----------------------------------------------------------------------
1846    Name: Matthias Blume
1847    Date: 2002/02/27 12:29:00 EST
1848    Tag: blume-20020227-cdebug
1849    Description:
1850    
1851    - made types in structure C and C_Debug to be equal
1852    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1853    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1854    
1855    ----------------------------------------------------------------------
1856    Name: Matthias Blume
1857    Date: 2002/02/26 12:00:00 EST
1858    Tag: blume-20020226-ffi
1859    Description:
1860    
1861    1. Fixed a minor bug in CM's "noweb" tool:
1862       If numbering is turned off, then truly don't number (i.e., do not
1863       supply the -L option to noweb).  The previous behavior was to supply
1864       -L'' -- which caused noweb to use the "default" line numbering scheme.
1865       Thanks to Chris Richards for pointing this out (and supplying the fix).
1866    
1867    2. Once again, I reworked some aspects of the FFI:
1868    
1869       A. The incomplete/complete type business:
1870    
1871       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
1872         gone!
1873       - ML types representing an incomplete type are now *equal* to
1874         ML types representing their corresponding complete types (just like
1875         in C).  This is still safe because ml-nlffigen will not generate
1876         RTTI for incomplete types, nor will it generate functions that
1877         require access to such RTTI.   But when ML code generated from both
1878         incomplete and complete versions of the C type meet, the ML types
1879         are trivially interoperable.
1880    
1881         NOTE:  These changes restore the full generality of the translation
1882         (which was previously lost when I eliminated functorization)!
1883    
1884       B. Enum types:
1885    
1886       - Structure C now has a type constructor "enum" that is similar to
1887         how the "su" constructor works.  However, "enum" is not a phantom
1888         type because each "T enum" has values (and is isomorphic to
1889         MLRep.Signed.int).
1890       - There are generic access operations for enum objects (using
1891         MLRep.Signed.int).
1892       - ml-nlffigen will generate a structure E_foo for each "enum foo".
1893         * The structure contains the definition of type "mlrep" (the ML-side
1894         representation type of the enum).  Normally, mlrep is the same
1895         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
1896         then mlrep will be defined as a datatype -- thus facilitating
1897         pattern matching on mlrep values.
1898         ("-ec" will be suppressed if there are duplicate values in an
1899          enumeration.)
1900         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
1901         will be generated for each C enum constant xxx.
1902         * Conversion functions m2i and i2m convert between mlrep and
1903         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
1904         * Coversion functions c and ml convert between mlrep and "tag enum".
1905         * Access functions (get/set) fetch and store mlrep values.
1906       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
1907         enumerations are merged into one single enumeration represented by
1908         structure E_'.
1909    
1910    ----------------------------------------------------------------------
1911    Name: Allen Leung
1912    Date: 2002/02/25 04:45:00 EST
1913    Tag: leunga-20020225-cps-spill
1914    
1915    This is a new implementation of the CPS spill phase.
1916    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
1917    In case of problems, replace it with the old file spill.sml
1918    
1919    The current compiler runs into some serious performance problems when
1920    constructing a large record.  This can happen when we try to compile a
1921    structure with many items.  Even a very simple structure like the following
1922    makes the compiler slow down.
1923    
1924        structure Foo = struct
1925           val x_1 = 0w1 : Word32.int
1926           val x_2 = 0w2 : Word32.int
1927           val x_3 = 0w3 : Word32.int
1928           ...
1929           val x_N = 0wN : Word32.int
1930        end
1931    
1932    The following table shows the compile time, from N=1000 to N=4000,
1933    with the old compiler:
1934    
1935    N
1936    1000   CPS 100 spill                           0.04u  0.00s  0.00g
1937           MLRISC ra                               0.06u  0.00s  0.05g
1938              (spills = 0 reloads = 0)
1939           TOTAL                                   0.63u  0.07s  0.21g
1940    
1941    1100   CPS 100 spill                           8.25u  0.32s  0.64g
1942           MLRISC ra                               5.68u  0.59s  3.93g
1943              (spills = 0 reloads = 0)
1944           TOTAL                                   14.71u  0.99s  4.81g
1945    
1946    1500   CPS 100 spill                           58.55u  2.34s  1.74g
1947           MLRISC ra                               5.54u  0.65s  3.91g
1948              (spills = 543 reloads = 1082)
1949           TOTAL                                   65.40u  3.13s  6.00g
1950    
1951    2000   CPS 100 spill                           126.69u  4.84s  3.08g
1952           MLRISC ra                               0.80u  0.10s  0.55g
1953              (spills = 42 reloads = 84)
1954           TOTAL                                   129.42u  5.10s  4.13g
1955    
1956    3000   CPS 100 spill                           675.59u  19.03s  11.64g
1957           MLRISC ra                               2.69u  0.27s  1.38g
1958              (spills = 62 reloads = 124)
1959           TOTAL                                   682.48u  19.61s  13.99g
1960    
1961    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
1962           MLRISC ra                               4.96u  0.27s  2.72g
1963              (spills = 85 reloads = 170)
1964           TOTAL                                   2375.26u  57.21s  48.00g
1965    
1966    As you can see the old cps spill module suffers from some serious
1967    performance problem.  But since I cannot decipher the old code fully,
1968    instead of patching the problems up, I'm reimplementing it
1969    with a different algorithm.  The new code is more modular,
1970    smaller when compiled, and substantially faster
1971    (O(n log n) time and O(n) space).  Timing of the new spill module:
1972    
1973    4000  CPS 100 spill                           0.02u  0.00s  0.00g
1974          MLRISC ra                               0.25u  0.02s  0.15g
1975             (spills=1 reloads=3)
1976          TOTAL                                   7.74u  0.34s  1.62g
1977    
1978    Implementation details:
1979    
1980    As far as I can tell, the purpose of the CPS spill module is to make sure the
1981    number of live variables at any program point (the bandwidth)
1982    does not exceed a certain limit, which is determined by the
1983    size of the spill area.
1984    
1985    When the bandwidth is too large, we decrease the register pressure by
1986    packing live variables into spill records.  How we achieve this is
1987    completely different than what we did in the old code.
1988    
1989    First, there is something about the MLRiscGen code generator
1990    that we should be aware of:
1991    
1992    o MLRiscGen performs code motion!
1993    
1994       In particular, it will move floating point computations and
1995       address computations involving only the heap pointer to
1996       their use sites (if there is only a single use).
1997       What this means is that if we have a CPS record construction
1998       statement
1999    
2000           RECORD(k,vl,w,e)
2001    
2002       we should never count the new record address w as live if w
2003       has only one use (which is often the case).
2004    
2005       We should do something similar to floating point, but the transformation
2006       there is much more complex, so I won't deal with that.
2007    
2008    Secondly, there are now two new cps primops at our disposal:
2009    
2010     1. rawrecord of record_kind option
2011        This pure operator allocates some uninitialized storage from the heap.
2012        There are two forms:
2013    
2014         rawrecord NONE [INT n]  allocates a tagless record of length n
2015         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2016                                     and initializes the tag.
2017    
2018     2. rawupdate of cty
2019          rawupdate cty (v,i,x)
2020          Assigns to x to the ith component of record v.
2021          The storelist is not updated.
2022    
2023    We use these new primops for both spilling and increment record construction.
2024    
2025     1. Spilling.
2026    
2027        This is implemented with a linear scan algorithm (but generalized
2028        to trees).  The algorithm will create a single spill record at the
2029        beginning of the cps function and use rawupdate to spill to it,
2030        and SELECT or SELp to reload from it.  So both spills and reloads
2031        are fine-grain operations.  In contrast, in the old algorithm
2032        "spills" have to be bundled together in records.
2033    
2034        Ideally, we should sink the spill record construction to where
2035        it is needed.  We can even split the spill record into multiple ones
2036        at the places where they are needed.  But CPS is not a good
2037        representation for global code motion, so I'll keep it simple and
2038        am not attempting this.
2039    
2040     2. Incremental record construction (aka record splitting).
2041    
2042        Long records with many component values which are simulatenously live
2043        (recall that single use record addresses are not considered to
2044         be live) are constructed with rawrecord and rawupdate.
2045        We allocate space on the heap with rawrecord first, then gradually
2046        fill it in with rawupdate.  This is the technique suggested to me
2047        by Matthias.
2048    
2049        Some restrictions on when this is applicable:
2050        1. It is not a VECTOR record.  The code generator currently does not handle
2051           this case. VECTOR record uses double indirection like arrays.
2052        2. All the record component values are defined in the same "basic block"
2053           as the record constructor.  This is to prevent speculative
2054           record construction.
2055    
2056    ----------------------------------------------------------------------
2057    Name: Allen Leung
2058    Date: 2002/02/22 01:02:00 EST
2059    Tag: leunga-20020222-mlrisc-tools
2060    
2061    Minor bug fixes in the parser and rewriter
2062    
2063    ----------------------------------------------------------------------
2064    Name: Allen Leung
2065    Date: 2002/02/21 20:20:00 EST
2066    Tag: leunga-20020221-peephole
2067    
2068    Regenerated the peephole files.  Some contained typos in the specification
2069    and some didn't compile because of pretty printing bugs in the old version
2070    of 'nowhere'.
2071    
2072    ----------------------------------------------------------------------
2073    Name: Allen Leung
2074    Date: 2002/02/19 20:20:00 EST
2075    Tag: leunga-20020219-mlrisc-tools
2076    Description:
2077    
2078       Minor bug fixes to the mlrisc-tools library:
2079    
2080       1.  Fixed up parsing colon suffixed keywords
2081       2.  Added the ability to shut the error messages up
2082       3.  Reimplemented the pretty printer and fixed up/improved
2083           the pretty printing of handle and -> types.
2084       4.  Fixed up generation of literal symbols in the nowhere tool.
2085       5.  Added some SML keywords to to sml.sty
2086    
2087    ----------------------------------------------------------------------
2088    Name: Matthias Blume
2089    Date: 2002/02/19 16:20:00 EST
2090    Tag: blume-20020219-cmffi
2091    Description:
2092    
2093    A wild mix of changes, some minor, some major:
2094    
2095    * All C FFI-related libraries are now anchored under $c:
2096        $/c.cm      --> $c/c.cm
2097        $/c-int.cm  --> $c/internals/c-int.cm
2098        $/memory.cm --> $c/memory/memory.cm
2099    
2100    * "make" tool (in CM) now treats its argument pathname slightly
2101      differently:
2102        1. If the native expansion is an absolute name, then before invoking
2103           the "make" command on it, CM will apply OS.Path.mkRelative
2104           (with relativeTo = OS.FileSys.getDir()) to it.
2105        2. The argument will be passed through to subsequent phases of CM
2106           processing without "going native".  In particular, if the argument
2107           was an anchored path, then "make" will not lose track of that anchor.
2108    
2109    * Compiler backends now "know" their respective C calling conventions
2110      instead of having to be told about it by ml-nlffigen.  This relieves
2111      ml-nlffigen from one of its burdens.
2112    
2113    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
2114    
2115    * Export C_DEBUG and C_Debug from $c/c.cm.
2116    
2117    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
2118      subtyping relationship between incomplete pointers and their complete
2119      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
2120      with the convention of instantiating 'o with ('t, 'c) obj whenever
2121      the pointer target type is complete.  In the incomplete case, 'o
2122      will be instantiated with some "'c iobj" -- a type obtained by
2123      using one of the functors PointerToIncompleteType or PointerToCompleteType.
2124    
2125      Operations that work on both incomplete and complete pointer types are
2126      typed as taking an 'o ptr while operations that require the target to
2127      be known are typed as taking some ('t, 'c) obj ptr.
2128    
2129      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
2130      where void is an eqtype without any values.  This makes it possible
2131      to work on voidptr values using functions meant to operate on light
2132      incomplete pointers.
2133    
2134    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
2135      been vastly simplified.
2136    
2137    ----------------------------------------------------------------------
2138    Name: Matthias Blume
2139    Date: 2002/02/19 10:48:00 EST
2140    Tag: blume-20020219-pqfix
2141    Description:
2142    
2143    Applied Chris Okasaki's bug fix for priority queues.
2144    
2145    ----------------------------------------------------------------------
2146    Name: Matthias Blume
2147    Date: 2002/02/15 17:05:00
2148    Tag: Release_110_39
2149    Description:
2150    
2151    Last-minute retagging is becoming a tradition... :-(
2152    
2153    This is the working release 110.39.
2154    
2155    ----------------------------------------------------------------------
2156    Name: Matthias Blume
2157    Date: 2002/02/15 16:00:00 EST
2158    Tag: Release_110_39-orig
2159    Description:
2160    
2161    Working release 110.39.  New bootfiles.
2162    
2163    (Update: There was a small bug in the installer so it wouldn't work
2164    with all shells.  So I retagged. -Matthias)
2165    
2166    ----------------------------------------------------------------------
2167    Name: Matthias Blume
2168    Date: 2002/02/15 14:17:00 EST
2169    Tag: blume-20020215-showbindings
2170    Description:
2171    
2172    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
2173    the latter can be useful for exploring what bindings are available at
2174    the interactive prompt.  (The first function returns only the list
2175    of symbols that are really bound, the second prints those but also the
2176    ones that CM's autoloading mechanism knows about.)
2177    
2178    ----------------------------------------------------------------------
2179    Name: Matthias Blume
2180    Date: 2002/02/15 12:08:00 EST
2181    Tag: blume-20020215-iptrs
2182    Description:
2183    
2184    Two improvements to ml-nlffigen:
2185    
2186      1. Write files only if they do not exist or if their current contents
2187         do not coincide with what's being written.  (That is, avoid messing
2188         with the time stamps unless absolutely necessary.)
2189    
2190      2. Implement a "repository" mechanism for generated files related
2191         to "incomplete pointer types".   See the README file for details.
2192    
2193    ----------------------------------------------------------------------
2194    Name: Matthias Blume
2195    Date: 2002/02/14 11:50:00 EST
2196    Tag: blume-20020214-quote
2197    Description:
2198    
2199    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
2200    because of the new and improved tag generation scheme.  (Thanks to Allen
2201    Leung for pointing it out.)
2202    
2203    ----------------------------------------------------------------------
2204    Name: Lal George
2205    Date: 2002/02/14 09:55:27 EST 2002
2206    Tag: george-20020214-isabelle-bug
2207    Description:
2208    
2209    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2210    of Isabelle on the x86.
2211    
2212    From Allen:
2213    -----------
2214     I've found the problem:
2215    
2216         in ra-core.sml, I use the counter "blocked" to keep track of the
2217         true number of elements in the freeze queue.  When the counter goes
2218         to zero, I skip examining the queue.  But I've messed up the
2219         bookkeeping in combine():
2220    
2221             else ();
2222             case !ucol of
2223               PSEUDO => (if !cntv > 0 then
2224                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2225                                        ^^^^^^^^^^^^^^^^^^^^^^^
2226                      moveu := mergeMoveList(!movev, !moveu)
2227                     )
2228                  else ();
2229    
2230         combine() is called to coalesce two nodes u and v.
2231         I think I was thinking that if the move counts of u and v are both
2232         greater than zero then after they are coalesced then one node is
2233         removed from the freeze queue.  Apparently I was thinking that
2234         both u and v are of low degree, but that's clearly not necessarily true.
2235    
2236    
2237    02/12/2002:
2238        Here's the patch.  HOL now compiles.
2239    
2240        I don't know how this impact on performance (compile
2241        time or runtime).  This bug caused the RA (especially on the x86)
2242        to go thru the potential spill phase when there are still nodes on the
2243        freeze queue.
2244    
2245    
2246    
2247    
2248    ----------------------------------------------------------------------
2249    Name: Matthias Blume
2250    Date: 2002/02/13 22:40:00 EST
2251    Tag: blume-20020213-fptr-rtti
2252    Description:
2253    
2254    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2255    updates.
2256    
2257    ----------------------------------------------------------------------
2258    Name: Matthias Blume
2259    Date: 2002/02/13 16:41:00 EST
2260    Tag: blume-20020213-cmlpq
2261    Description:
2262    
2263    Added new priority queue export symbols (which have just been added to
2264    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2265    not compile and the installer would choke.)
2266    
2267    ----------------------------------------------------------------------
2268    Name: Matthias Blume
2269    Date: 2002/02/13 16:15:00 EST
2270    Tag: blume-20020213-various
2271    Description:
2272    
2273    1. More tweaks to ml-nlffigen:
2274    
2275       - better internal datastructures (resulting in slight speedup)
2276       - "-match" option requires exact match
2277       - "localized" gensym counters (untagged structs/unions nested within
2278         other structs/unions or within typedefs get a fresh counter; their
2279         tag will be prefixed by a concatenation of their parents' tags)
2280       - bug fixes (related to calculation of transitive closure of types
2281         to be included in the output)
2282    
2283    2. Minor Basis updates:
2284    
2285       - added implementations for List.collate and Option.app
2286    
2287    ----------------------------------------------------------------------
2288    Name: Matthias Blume
2289    Date: 2002/02/11 15:55:00 EST
2290    Tag: blume-20020211-gensym
2291    Description:
2292    
2293    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2294    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2295    names (ML structure names that correspond to unnamed C structs, unions,
2296    and enums), so that separate runs of ml-nlffigen do not clash.
2297    
2298    ----------------------------------------------------------------------
2299    Name: Matthias Blume
2300    Date: 2002/02/11 12:05:00 EST
2301    Tag: blume-20020211-gensml
2302    Description:
2303    
2304    A quick fix for a problem with GenSML (in the pgraph-util library):
2305    Make generation of toplevel "local" optional.  (Strictly speaking,
2306    signature definitions within "local" are not legal SML.)
2307    
2308    Other than that: updates to INSTALL and cm/TODO.
2309    
2310    ----------------------------------------------------------------------
2311    Name: Matthias Blume
2312    Date: 2002/02/08 15:00:00 EST
2313    Tag: blume-20020208-uniquepid
2314    Description:
2315    
2316    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2317    
2318    1. The installer (config/install.sh) has gotten smarter:
2319    
2320         - Configuration options are a bit easier to specify now
2321           (in config/targets).
2322         - Bug in recognizing .tar.bz2 files fixed.
2323         - Installer automatically resolves dependencies between
2324           configuration options (e.g., if you ask for eXene, you will
2325           also get cml -- regardless whether you asked for it or not).
2326         - Installer can run in "quieter mode" by setting the environment
2327           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2328           "completely silent", though.
2329         - Build HashCons library as part of smlnj-lib.
2330    
2331    2. A new scheme for assigning persistent identifiers to compilation
2332       units (and, by extension, to types etc.) has been put into place.
2333       This fixes a long-standing bug where types and even dynamic values
2334       can get internally confused, thereby compromising type safety
2335       (abstraction) and dynamic correctness.  See
2336    
2337         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2338    
2339       for an example of how things could go wrong until now.
2340    
2341       The downside of the new scheme is that pids are not quite as
2342       persistent as they used to be: CM will generate a fresh pid
2343       for every compilation unit that it thinks it sees for the first
2344       time.  That means that if you compile starting from a clean, fresh
2345       source tree at two different times, you end up with different
2346       binaries.
2347    
2348       Cutoff recompilation, however, has not been compromised because
2349       CM keeps pid information in special caches between runs.
2350    
2351    ----------------------------------------------------------------------
2352    Name: Lal George
2353    Date: 2002/02/07 15:34:13 EST 2002
2354    Tag: <none>
2355    Description:
2356    
2357    Compilers that generate assembly code may produce  global labels
2358    whose value is resolved at link time. The various peephole optimization
2359    modules did not take this in account.
2360    
2361    TODO. The Labels.addrOf function should really return an option
2362    type so that clients are forced to deal with this issue, rather
2363    than an exception being raised.
2364    
2365    ----------------------------------------------------------------------
2366    Name: Lal George
2367    Date: 2002/02/06 13:55:02 EST
2368    Tag: george-20020206-ra-breakup
2369    Description:
2370    
2371    1. A bug fix from Allen.
2372    
2373        A typo causes extra fstp %st(0)'s to be generated at compensation
2374        edges, which might cause stack underflow traps at runtime.  This
2375        occurs in fft where there are extraneous fstps right before the 'into'
2376        trap instruction (in this case they are harmless since none of the
2377        integers overflow.)
2378    
2379    2. Pulled out various utility modules that were embedded in the modules
2380       of the register allocator. I need these modules for other purposes, but
2381       they are not complete enough to put into a library (just yet).
2382    ----------------------------------------------------------------------
2383    Name: Matthias Blume
2384    Date: 2002/01/31 16:05:00 EST
2385    Tag: blume-20020131-sparc-ccalls
2386    Description:
2387    
2388    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2389       of extra stack space by mistake.  Fixed.
2390    
2391    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2392    
2393    ----------------------------------------------------------------------
2394    Name: Allen Leung
2395    Date: 2002/01/30
2396    Tag: leunga-20020130-nowhere-bug-fix
2397    Description:
2398    
2399       MLRISC bug fixes:
2400       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2401       2. MachineInt.fromString was negating its value.
2402    
2403    ----------------------------------------------------------------------
2404    Name: Matthias Blume
2405    Date: 2002/01/29
2406    Tag: blume-20020129-INSTALL
2407    Description:
2408    
2409    - Added somewhat detailed installation instructions (file INSTALL).
2410    - Fixed curl-detection bug in config/install.sh.
2411    - It is now possible to select the URL getter using the URLGETTER
2412      environment variable:
2413    
2414          not set / "unknown"      --> automatic detection (script tries wget,
2415                                       curl, and lynx)
2416          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2417                                       how to properly invoke them)
2418          other                    --> use $URLGETTER directly, it must take
2419                                       precisely two command-line arguments
2420                                       (source URL and destination file name)
2421    
2422    ----------------------------------------------------------------------
2423    Name: Matthias Blume
2424    Date: 2002/01/28
2425    Tag: blume-20020128-sparc-ccalls
2426    Description:
2427    
2428    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2429    - Make use of the allocParam argument in sparc-c-calls.
2430    
2431    ----------------------------------------------------------------------
2432    Name: Matthias Blume
2433    Date: 2002/01/28
2434    Tag: blume-20020128-allocParam
2435    Description:
2436    
2437    John Reppy:  Changes c-calls API to accept client-callback for
2438    allocating extra stack space.
2439    me: Corresponding changes to mlriscGen (using a dummy argument that
2440        does not change the current behavior).
2441    
2442    ----------------------------------------------------------------------
2443    Name: Matthias Blume
2444    Date: 2002/01/28 12:00:00
2445    Tag: Release_110_38
2446    Description:
2447    
2448    This time for real!!!
2449    
2450    ----------------------------------------------------------------------
2451    Name: Matthias Blume
2452    Date: 2002/01/28 10:56:00 EST
2453    Tag: blume-20020128-retraction
2454    Description:
2455    
2456    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
2457       with blume-Release_110_38-retracted.)
2458    
2459    1. Fixed a problem with incorrect rounding modes in real64.sml.
2460       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2461    
2462    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2463       and enums fixed.  The naming of corresponding ML identifiers should
2464       now be consistent again.
2465    
2466    ----------------------------------------------------------------------
2467    Name: Allen Leung
2468    Date: 2002/01/27
2469    Tag: leunga-20020127-nowhere
2470    Description:
2471    
2472       Added a target called nowhere in the configuration scripts.
2473       Enabling this will build the MLRISC 'nowhere' tool (for translating
2474       programs with where-clauses into legal SML code) during installation.
2475    
2476    ----------------------------------------------------------------------
2477    Name: Matthias Blume
2478    Date: 2002/01/25 21:27:00 EST
2479    Tag: blume-Release_110_38-retracted
2480    Description:
2481    
2482    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
2483    
2484    README will be added later.
2485    
2486    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
2487    (Release_110_38) removed.  Reason: Last-minute bug fixes.
2488    
2489    ----------------------------------------------------------------------
2490    Name: Matthias Blume
2491    Date: 2002/01/25
2492    Tag: blume-20020125-ffi
2493    Description:
2494    
2495    A large number of tweaks and improvements to ml-nlffi-lib and
2496    ml-nlffigen:
2497    
2498       - ML represenation types have been streamlined
2499       - getter and setter functions work with concrete values, not abstract
2500         ones where possible
2501       - ml-nlffigen command line more flexible (see README file there)
2502       - some bugs have been fixed (hopefully)
2503    
2504    ----------------------------------------------------------------------
2505    Name: Lal George
2506    Date: 2002/01/24
2507    Tag: george-20020124-risc-ra-interface
2508    Description:
2509    
2510       There is a dramatic simplification in the interface to the
2511       register allocator for RISC architectures as a result of making
2512       parallel copy instructions explicit.
2513    
2514    ----------------------------------------------------------------------
2515    Name: Matthias Blume
2516    Date: 2002/01/22
2517    Tag: blume-20020122-x86-ccalls
2518    Description:
2519    
2520    Bug fix for c-calls on x86 (having to do with how char- and
2521    short-arguments are being handled).
2522    
2523    ----------------------------------------------------------------------
2524    Name: Matthias Blume
2525    Date: 2002/01/21
2526    Tag: blume-20020121-ff
2527    Description:
2528    
2529    Another day of fiddling with the FFI...
2530    
2531    1. Bug fix/workaround:  CKIT does not complain about negative array
2532       dimensions, so ml-nlffigen has to guard itself against this possibility.
2533       (Otherwise a negative dimension would send it into an infinite loop.)
2534    
2535    2. Some of the abstract types (light objects, light pointers, most "base"
2536       types) in structure C are now eqtypes.
2537    
2538    3. Added constructors and test functions for NULL function pointers.
2539    
2540    ----------------------------------------------------------------------
2541    Name: Matthias Blume
2542    Date: 2002/01/18
2543    Tag: blume-20020118-ready-for-new-release
2544    Description:
2545    
2546    Made config/srcarchiveurl point to a new place.  (Will provide boot
2547    files shortly.)
2548    
2549    Maybe we christen this to be 110.38?
2550    
2551    ----------------------------------------------------------------------
2552    Name: Matthias Blume
2553    Date: 2002/01/18
2554    Tag: blume-20020118-more-ffifiddle
2555    Description:
2556    
2557    Today's FFI fiddling:
2558    
2559      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
2560        C.Set.* that use concrete (MLRep.*) arguments and results instead
2561        of abstract ones.
2562    
2563      - Provided word-style bit operations etc. for "int" representation
2564        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
2565        and Long.
2566    
2567    ----------------------------------------------------------------------
2568    Name: Matthias Blume
2569    Date: 2002/01/18
2570    Tag: blume-20020118-use-x86-fp
2571    Description:
2572    
2573    Now that x86-fast-fp seems to be working, I turned it back on again
2574    by default.  (Seems to work fine now, even with the FFI.)
2575    
2576    Other than that, I added some documentation about the FFI to
2577    src/ml-nlffigen/README and updated the FFI test examples in
2578    src/ml-nlffi-lib/Tests/*.
2579    
2580    ----------------------------------------------------------------------
2581    Name: Allen Leung
2582    Date: 2002/01/17
2583    Tag: leunga-20020117-x86-fast-fp-call
2584    Description:
2585    
2586       1. Fixed a problem with handling return fp values when x86's fast fp
2587          mode is turned on.
2588    
2589       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
2590          of %f32.
2591    
2592       3. Added a constructor INT32lit to the ast of MLRISC tools.
2593    
2594    ----------------------------------------------------------------------
2595    Name: Matthias Blume
2596    Date: 2002/01/16
2597    Tag: blume-20020116-ffifiddle
2598    Description:
2599    
2600    More fiddling with the FFI interface:
2601    
2602     - Make constness 'c instead of rw wherever possible.  This eliminates
2603       the need for certain explicit coercions.  (However, due to ML's
2604       value polymorphism, there will still be many cases where explicit
2605       coercions are necessary.  Phantom types are not the whole answer
2606       to modeling a subtyping relationship in ML.)
2607    
2608     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
2609    
2610     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
2611       clumsy workaround for varargs problem.)
2612    
2613    ----------------------------------------------------------------------
2614    Name: Lal George
2615    Date: 2002/01/15
2616    Tag: <none>
2617    Description:
2618    
2619    1. Since COPY instructions are no longer native to the architecture,
2620       a generic functor can be used to implement the expandCopies function.
2621    
2622    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
2623       TEXT segment.
2624    
2625    ----------------------------------------------------------------------
2626    Name: Matthias Blume
2627    Date: 2002/01/15
2628    Tag: blume-20020115-ffiupdates
2629    Description:
2630    
2631    1. Fix for bug resulting in single-precision float values being returned
2632       incorrectly from FFI calls.
2633    
2634    2. Small modifications to C FFI API:
2635    
2636        - memory-allocation routines return straight objects (no options)
2637          and raise an exception in out-of-memory situations
2638        - unsafe extensions to cast between function pointers and pointers
2639          from/to ints
2640        - added structure C_Debug as an alternative to structure C where
2641          pointer-dereferencing (|*| and |*!) always check for null-pointers
2642        - added open_lib' to DynLinkage;  open_lib' works like open_lib
2643          but also takes a (possibly empty) list of existing library handles
2644          that the current library depends on
2645    
2646    ----------------------------------------------------------------------
2647    Name: Matthias Blume
2648    Date: 2002/01/10
2649    Tag: blume-20020110-newffigen
2650    Description:
2651    
2652    1. Updates to portable graph code.
2653    
2654    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
2655       more scalable now so that even huge interfaces such as the one
2656       for GTK compile in finite time and space. :-)
2657       See src/ml-nlffigen/README for details on what's new.
2658    
2659    ----------------------------------------------------------------------
2660    Name: Lal George
2661    Date: 2001/01/09 14:31:35 EST 2002
2662    Tag: george-20011206-rm-native-copy
2663    Description:
2664    
2665            Removed the native COPY and FCOPY instructions
2666            from all the architectures and replaced it with the
2667            explicit COPY instruction from the previous commit.
2668    
2669            It is now possible to simplify many of the optimizations
2670            modules that manipulate copies. This has not been
2671            done in this change.
2672    
2673  ----------------------------------------------------------------------  ----------------------------------------------------------------------
2674  Name: Lal George  Name: Lal George
2675  Date: 2001/12/06 16:50:13 EST 2001  Date: 2001/12/06 16:50:13 EST 2001
# Line 341  Line 3000 
3000    
3001  ----------------------------------------------------------------------  ----------------------------------------------------------------------
3002  Name: Matthias Blume  Name: Matthias Blume
 >>>>>>> 1.169  
3003  Date: 2001/09/18 15:35:00 EDT  Date: 2001/09/18 15:35:00 EDT
3004  Tag: blume-20010918-readme11036  Tag: blume-20010918-readme11036
3005  Description:  Description:

Legend:
Removed from v.1003  
changed lines
  Added in v.1488

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