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 589, Fri Mar 31 16:14:36 2000 UTC revision 1489, Tue May 11 19:31:06 2004 UTC
# Line 8  Line 8 
8  The form of an entry should be:  The form of an entry should be:
9    
10  Name:  Name:
11  Date:  Date: yyyy/mm/dd
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14    
15  ----------------------------------------------------------------------  ----------------------------------------------------------------------
16    Name: Matthias Blume (blume (at) tti - c (dot) org)
17    Date: 2004/05/11 14:35:00 CDT
18    Tag: blume-20040511-win32sock
19    Description:
20    
21    Added previously missing support for many socket-related functions
22    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
23    for the voluminous patch!
24    
25    (I have not tested this patch under win32 yet.)
26    
27    ----------------------------------------------------------------------
28    Name: Matthias Blume (blume (at) tti - c (dot) org)
29    Date: 2004/05/11 14:20:00 CDT
30    Tag: blume-20040511-installml
31    Description:
32    
33    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
34    the patch.)
35    
36    ----------------------------------------------------------------------
37    Name: Matthias Blume (blume (at) tti - c (dot) org)
38    Date: 2004/05/11 14:05:00 CDT
39    Tag: blume-20040511-nlffi-netbsd
40    Description:
41    
42    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
43    the patch.)
44    
45    ----------------------------------------------------------------------
46    Name: Matthias Blume (blume (at) tti - c (dot) org)
47    Date: 2004/05/11 12:05:00 CDT
48    Tag: blume-20040511-exports
49    Description:
50    
51    As per request by Adam Chlipala <adam@hcoop.net>, extended various
52    export lists in compiler-related .cm-files.
53    
54    ----------------------------------------------------------------------
55    Name: Matthias Blume (blume (at) tti - c (dot) org)
56    Date: 2004/05/11 11:35:00 CDT
57    Tag: blume-20040511-allsource
58    Description:
59    
60    The installer now honors the "src-smlnj" target again, although its meaning
61    has changed from "all sources required for the compiler" to "all sources
62    the installer knows about".  In other words, if you enable "src-smlnj"
63    in the "targets" file, then the installer will pull in sources for
64    everything.  (Notice that this refers to source code only.  Compiled
65    code is still only installed for modules that were requested explicitly
66    or which are required for other modules that were requested explicitly.)
67    
68    ----------------------------------------------------------------------
69    Name: Matthias Blume (blume (at) tti - c (dot) org)
70    Date: 2004/04/23 17:40:00 CDT
71    Tag: blume-20040423-ieee-scan
72    Description:
73    
74    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
75    in the exponent calculation we get INF or ZERO (depending on the mantissa
76    and the sign of the exponent).
77    
78    ----------------------------------------------------------------------
79    Name: Matthias Blume (blume (at) tti - c (dot) org)
80    Date: 2004/04/23 10:40:00 CDT
81    Tag: blume-20040423-ml-build
82    Description:
83    
84    The ml-build script now terminates with a non-0 status when something
85    goes wrong.
86    
87    ----------------------------------------------------------------------
88    Name: Matthias Blume (blume (at) tti - c (dot) org)
89    Date: 2004/04/22 16:35:00 CDT
90    Tag: blume-20040422-Option
91    Description:
92    
93    Made exception Option to be the same as exception Option.Option
94    (as it should be).
95    
96    ----------------------------------------------------------------------
97    Name: Allen Leung (leunga (at) reservoir (dot) com)
98    Date: 2004/03/19 14:40:00 EST
99    Tag: leunga-20040319-cygwin-nlffi
100    Description:
101    
102    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
103    of SML/NJ.  The problem is that
104    
105       lib = dlopen(NULL, ...)
106       f   = dlsym(lib, "malloc");
107    
108    does not work on Windows unless we explicitly export symbols
109    such as 'malloc' during linking.  We fixed this by explicitly
110    exporting the required symbols with the magic gcc incantation:
111    
112        -Wl,--export-all cygwin.def
113    
114    where cygwin.def is a file containing all the symbols that
115    we wish to export.
116    
117    I suspect this is a Windows problem and we'll have to
118    do the same (somehow with windows compilers) when
119    we build the native win32 version with the system
120    calls LoadLibrary/GetProcAddress.
121    
122    ----------------------------------------------------------------------
123    Name: Matthias Blume (blume (at) tti - c (dot) org)
124    Date: 2004/03/04 16:35:00 CST
125    Tag: blume-20040304-intinf-fmt
126    Description:
127    
128    Fixed problem with IntInf.fmt (sign would show up on the right instead
129    of on the left for BIN, OCT, and HEX).
130    
131    ----------------------------------------------------------------------
132    Name: Matthias Blume (blume (at) tti - c (dot) org)
133    Date: 2004/03/04 11:25:00 CST
134    Tag: blume-20040304-symlinks
135    Description:
136    
137    Fixed problem with installer script (unix only) where bin/ml-yacc and
138    friends pointed (via symlinks) to absolute locations instead of just
139    .run-sml.  This was reported by Vesa A Norrman.
140    
141    ----------------------------------------------------------------------
142    Name: Matthias Blume (blume (at) tti - c (dot) org)
143    Date: 2004/02/13 14:50:00 CST
144    Tag: Release_110_45
145    Description:
146    
147    New working version (110.45). New bootfiles.
148    
149    ----------------------------------------------------------------------
150    Name: Matthias Blume (blume (at) tti - c (dot) org)
151    Date: 2004/01/26 15:15:15 CST
152    Tag: blume-20040126-toplevel
153    Description:
154    
155    Improved handling of exceptions at the interactive toplevel.
156    
157    ----------------------------------------------------------------------
158    Name: Matthias Blume (blume (at) tti - c (dot) org)
159    Date: 2004/01/26 11:25:00
160    Tag: blume-20040126-app
161    Description:
162    
163    Type of top-level "app" corrected.
164    Added code for setting vp_limitPtrMask to Win32-specific runtime.
165    
166    ----------------------------------------------------------------------
167    Name: Matthias Blume (blume (at) tti - c (dot) org)
168    Date: 2003/11/18 17:10 CST
169    Tag: blume-20031118-basis-fiddle
170    Description:
171    
172    - changed Timer interface to what might become the spec
173    - POSIX_FLAGS -> BIT_FLAGS according to spec
174    - some other minor discrepancies wrt. spec eliminated
175    
176    ----------------------------------------------------------------------
177    Name: Matthias Blume (blume (at) tti - c (dot) org)
178    Date: 2003/11/06 12:00:00 CST
179    Tag: Release_110_44
180    Description:
181    
182    New working version (110.44). New bootfiles.
183    
184    ----------------------------------------------------------------------
185    Name: Matthias Blume (blume (at) tti - c (dot) org)
186    Date: 2003/11/04 11:50:00 CST
187    Tag: blume-20031104-move-libraries
188    Description:
189    
190    Eliminated the "dont_move_libraries" directive in config/targets.
191    (The mechanism was broken and could not be fixed easily.  Moreover,
192    there does not seem to be any reason not to move all libraries into
193    lib during installation.  I originally implemented this directive as a
194    backward-compatibility feature when I first introduced the new CM.
195    Now that things have been stable for a long time and going back to the
196    old CM is not an option, there is no reason to keep it around.)
197    
198    ----------------------------------------------------------------------
199    Name: Matthias Blume (blume (at) tti - c (dot) org)
200    Date: 2003/11/03 16:00:00 CST
201    Tag: blume-20031103-installdir
202    Description:
203    
204    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
205    Richards for pointing out the problem and providing the solution.)
206    
207    ----------------------------------------------------------------------
208    Name: Matthias Blume (blume (at) tti - c (dot) org)
209    Date: 2003/10/01 17:05:00 CDT
210    Tag: blume-20031001-lal-mlrisc
211    Description:
212    
213    MLRISC bug fix from Lal.
214    
215    ----------------------------------------------------------------------
216    Name: Matthias Blume (blume (at) tti - c (dot) org)
217    Date: 2003/09/30 16:10:00 CDT
218    Tag: blume-20030930-primio-bat
219    Description:
220    
221    1. Added openVector, nullRd, and nullWr to PRIM_IO.
222    2. Improved .bat files (for Win32 port) to make things work under Win95.
223       (thanks to Aaron S. Hawley for this one)
224    
225    ----------------------------------------------------------------------
226    Name: Matthias Blume (blume (at) tti - c (dot) org)
227    Date: 2003/09/26 16:05:00 CDT
228    Tag: blume-20030926-wrappriv
229    Description:
230    
231    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
232    
233    ----------------------------------------------------------------------
234    Name: Matthias Blume (blume (at) tti - c (dot) org)
235    Date: 2003/09/26 15:00:00 CDT
236    Tag: blume-20030926-110_43_3
237    Description:
238    
239    - additional cleanup
240    - version number bump, NEW BOOTFILES
241    
242    ----------------------------------------------------------------------
243    Name: Matthias Blume (blume (at) tti - c (dot) org)
244    Date: 2003/09/26 12:00:00 CDT
245    Tag: blume-20030926-ppautoload
246    Description:
247    
248    I modified the read-eval-print loop so that the autoloader gets
249    invoked whenever the prettyprinter tries to look up a symbol that
250    is not currently defined in the toplevel environment but which
251    appears in CM's autoload registry.  As a result, we see far fewer of
252    those ?.Foo.Bar.xxx names in the prettyprinter's output.
253    
254    In addition to this I tried to clean up some pieces of the Basis
255    implementation (e.g., Socket, Word8Array) in order to prevent other
256    instances of these ?.Foo.Bar.xxx names from being printed.
257    
258    The mechanism that picks names for types still needs some work, though.
259    (Right now it seems that if there is a type A.t which is defined to
260    be B.u, but B is unavailable at toplevel, then A.t gets printed as
261    "?.B.u" although the perhaps more sensible solution would be to use
262    "A.t" in this case.  In other words, the prettyprinter should follow
263    a chain of DEFtycs not farther than there are corresponding toplevel
264    names in the current environment.)
265    
266    ----------------------------------------------------------------------
267    Name: Matthias Blume (blume (at) tti - c (dot) org)
268    Date: 2003/09/24 16:31:00 CDT
269    Tag: blume-20030924-installer
270    Description:
271    
272    Another installer tweak:  All the ML code for the installer is now
273    compiled during CMB.make and put into a little library called
274    $smlnj/installer.cm.  The installation then simply invokes
275    
276       sml -m $smlnj/installer.cm
277    
278    and everything happens automagically.
279    
280    Win32:  ML code senses value of environment variable SMLNJ_HOME.
281    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
282            and BINDIR.
283    
284    The new scheme guarantees that the ML code responsible for the installation
285    is in sync with the APIs of the main system.  Also, the installer is
286    somewhat faster because the installer script is precompiled.
287    
288    ----------------------------------------------------------------------
289    Name: Matthias Blume (blume (at) tti - c (dot) org)
290    Date: 2003/09/24 15:35:00 CDT
291    Tag: blume-20030924-synsock
292    Description:
293    
294    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
295    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
296    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
297    implementations of non-blocking operations from CML's Socket
298    structure.
299    
300    ----------------------------------------------------------------------
301    Name: Matthias Blume (blume (at) tti - c (dot) org)
302    Date: 2003/09/24 15:10:05 CDT
303    Tag: blume-20030924-sockets
304    Description:
305    
306    1. Fixed SOCKET API and implementation to match Basis spec.
307       This required changing the internal representation of sockets to one
308       that remembers (for each socket file descriptor) whether it is currently
309       blocking or non-blocking.  This state is maintained lazily (i.e., a system
310       call is made only if the state actually needs to change).
311    
312    2. OS-specific details of sockets were moved into separate files, thus
313       making it possible to unify the bulk of the socket implementations
314       between Unix and Win32.
315    
316    3. CML's socket API changed accordingly.
317       (Note that we need to remove non-blocking functions from this API
318       since they are redundant in the case of CML!)
319    
320    4. CML's socket implementation now makes use of non-blocking functions
321       provided by Basis, thus removing all OS-dependent code from this part
322       of CML.
323    
324    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
325    
326    ----------------------------------------------------------------------
327    Name: Matthias Blume (blume (at) tti - c (dot) org)
328    Date: 2003/09/22 12:10:00 CDT
329    Tag: blume-20030922-110_43_2
330    Description:
331    
332    Made a new interim version and bootfiles for developer's bootstrapping
333    convenience.
334    
335       110.43.2 -- NEW BOOTFILES
336    
337    ----------------------------------------------------------------------
338    Name: Matthias Blume (blume (at) tti - c (dot) org)
339    Date: 2003/09/19 15:55:00 CDT
340    Tag: blume-20030919-cmdir
341    Description:
342    
343    1. new-install.sh -> install.sh
344    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
345    3. tweaked installer so that another name instead of .cm can be chosen
346       at install time (by setting the CM_DIR_ARC environment variable
347       during installation); once installation is complete, the name is
348       fixed
349    
350    ----------------------------------------------------------------------
351    Name: Matthias Blume (blume (at) tti - c (dot) org)
352    Date: 2003/09/18 16:00:00 CDT
353    Tag: blume-20030918-110_43_1
354    Description:
355    
356    Made a new interim version and bootfiles for developer's bootstrapping
357    convenience.
358    
359       110.43.1 -- NEW BOOTFILES
360    
361    ----------------------------------------------------------------------
362    Name: Matthias Blume (blume (at) tti - c (dot) org)
363    Date: 2003/09/18 15:20:00 CDT
364    Tag: blume-20030918-misc
365    Description:
366    
367    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
368       this seems to be controversial at the moment) and used those in
369       Posix.ProcEnv.times.
370    
371    2. Added Time.{from,to}Nanoseconds to Time.
372    
373    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
374       For example, fromLargeInt never needs to look at more than 3 "big
375       digits" to get its 53 bits of precision.
376    
377    ----------------------------------------------------------------------
378    Name: Matthias Blume (blume (at) tti - c (dot) org)
379    Date: 2003/09/17 16:30:00 CDT
380    Tag: blume-20030917-real32-slices
381    Description:
382    
383    Added an entry to the primitive environment
384    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
385    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
386    
387    Removed some of the "magic" constants in real64.sml and replaced them
388    with code that generates these values from their corresponding
389    integer counterparts.
390    
391    Made all(?) the slice-related changes to the Basis and made everything
392    compile again...
393    
394    ----------------------------------------------------------------------
395    Name: Matthias Blume (blume (at) tti - c (dot) org)
396    Date: 2003/09/15 17:45:00 CDT
397    Tag: blume-20030915-rbase
398    Description:
399    
400    Fixed bug in Real.fromLargeInt.
401    
402    ----------------------------------------------------------------------
403    Name: Matthias Blume (blume (at) tti - c (dot) org)
404    Date: 2003/09/13 18:11:00 CDT
405    Tag: blume-20030913-libinstall
406    Description:
407    
408    Minor bugfix in config/libinstall (set anchor with path to
409    standalone tool after installing it, otherwise libraries that
410    need ml-lex or ml-yacc won't compile the first time the installer
411    runs).
412    
413    ----------------------------------------------------------------------
414    Name: Matthias Blume (blume (at) tti - c (dot) org)
415    Date: 2003/09/12 11:45:00 CDT
416    Tag: blume-20030912-various
417    Description:
418    
419     - fixed bug in Real.toLargeInt
420     - fixed bug in Posix.ProcEnv.times
421     - changed inputLine functions to return an option
422     - minor installer improvements / bugfixes
423     - changed default @SMLalloc parameter for x86/celeron to 64k
424    
425    ----------------------------------------------------------------------
426    Name: Matthias Blume (blume (at) tti - c (dot) org)
427    Date: 2003/09/09 22:00:00 CDT
428    Tag: Release_110_43
429    Description:
430    
431        New working release 110.43.  New bootfiles.
432    
433    ----------------------------------------------------------------------
434    Name: Matthias Blume (blume (at) tti - c (dot) org)
435    Date: 2003/09/09 19:20:00 CDT
436    Tag: blume-20030909-installer
437    Description:
438    
439    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
440    Modified config/install.bat to take advantage of it.  Also modified
441    config/install.sh (and called it config/new-install.sh) to take advantage
442    of it on Unix systems.  (The SML code is (supposed to be) platform-
443    independent.)
444    
445    The installer can now install everything under Win32
446    as well as under *nix as long as it compiles.
447    
448    Other changes:
449    
450      - made CML compile again under Win32
451      - made eXene compile under Win32 (by providing a fake structure UnixSock
452        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
453      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
454        same as type int;  under Win32 it isn't
455      - fixed some slice-related problems in the win32-specific parts of CML
456      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
457        (under Win32, the volume name is case-insensitive, and the
458        OS.Path code compares volume names for equality)
459    
460    ----------------------------------------------------------------------
461    Name: Matthias Blume (blume (at) tti - c (dot) org)
462    Date: 2003/09/08 11:55:00 CDT
463    Tag: blume-20030908-fullpath
464    Description:
465    
466    Made Win32 version of OS.FileSys.fullPath return current directory
467    when given an empty string.  This is what the spec says, and incidentally,
468    CM depends on it.  (CM otherwise goes into an infinite loop in certain
469    cases when presented with the name of a non-existing .cm file.)
470    
471    ----------------------------------------------------------------------
472    Name: Matthias Blume (blume (at) tti - c (dot) org)
473    Date: 2003/09/04 16:30:00 CDT
474    Tag: blume-20030905-slices-etc
475    Description:
476    
477      1. Changed interface to vectors and arrays in Basis to match
478         (draft) Basis spec.
479      2. Added signatures and implementations of slices according to
480         Basis spec.
481      3. Edited source code throughout the system to make it compile again
482         under 1. and 2.  (In some cases code had to be added to have it
483         match the new signatures.)
484      4. MLRISC should be backward-compatible: the copies of the originals
485         of files that needed to change under 3. were retained, the .cm files
486         check the compiler version number and use old versions when
487         appropriate.
488      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
489         dirstream -> string option (in accordance with Basis spec).
490      6. When generating code that counts lines, ml-lex used function
491         CharVector.foldli, taking advantage of its old interface.
492         This has been replaced with the corresponding code from
493         CharVectorSlice.  (html-lex must be re-lexed!)
494      7. BitArray in smlnj-lib/Util has been extended/modified to match the
495         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
496      8. Removed temporary additions (fromInternal, toInternal) from the
497         (now obsolete) IntInf in smlnj-lib/Util.
498      9. Cleaned up structure Byte.
499     10. Added localOffset, scan, and fromString to Date (according to spec).
500         Cleaned/corrected implementation of Date.
501         (Still need to check for correctness; implement better canonicalizeDate.)
502     11. Added "scan" to signature IEEE_REAL.
503     12. Some improvements to IntInf [in particular: efficiency-hack for
504         mod and rem when second operand is 2 (for parity checks).]
505     13. Changed representation of type Time.time, using a single IntInf.int
506         value counting microseconds.  This considerably simplified the
507         implementation of structure Time.  We now support negative time
508         values; scan and fromString handle signs.
509     14. Functor PrimIO now takes two additional arguments (VectorSlice and
510         ArraySlice).
511    
512    ----------------------------------------------------------------------
513    Name: Matthias Blume (blume (at) tti - c (dot) org)
514    Date: 2003/08/28 17:00:00 CDT
515    Tag: blume-20030828-intinf
516    Description:
517    
518    This is a major update which comes with a version number bump
519    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
520    and an implementation of IntInf in the Basis.
521    
522    There are a fairly large number of related changes and updates throughout
523    the system:
524    
525    Basis:
526            - Implemented IntInf.
527            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
528            - Added some missing Real64 operations, most notably Real.toLargeInt.
529            - Added FixedInt as a synonym for Int32.
530    
531    compiler:
532            * Added support for a built-in intinf type.
533              - literals
534              - pattern matching
535              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
536              - overloading on literals and operations
537    
538            This required adding a primitive type intinf, some additional
539            primops, and implementations for several non-trivial intinf
540            operations in Core.  (The intinf type is completely abstract
541            to the compiler; all operations get delegated back to the Core.)
542    
543            * Intinf equality is handled by polyequal.  However, the compiler
544            does not print its usual warning in this case (since polyequal
545            is the right thing to do there).
546    
547            * Improved the organization of structure InlineT.
548    
549            * A word about conversion primops:
550            If conversions involving intinf do not cancel out during
551            CPS contract, then the compiler must insert calls to Core functions.
552            Since all core access must be resolved already during the FLINT
553            translate phase, it would be too late a the time of CPS contract
554            to add new Core calls.  For this reason, conversion primops
555            for intinf carry two arguments:  1. the numeric argument that
556            they are supposed to convert, and 2. the Core function that
557            can help with this conversion if necessary.  If CPS contract
558            eliminates a primop, then the associated Core function becomes
559            dead and goes away.  Intinf conversion primops that do not get
560            eliminated by CPS contract get rewritten into calls of their
561            core functions by a separate, new phase.
562    
563    interactive system:
564            - Control.Print.intinfDepth controls max length of intinf constants
565            being printed. (Analogous to Control.Print.stringDepth.)
566            - Cleanup in printutil and pputil: got rid of unused stuff and
567            duplicates; replaced some of the code with code that makes better
568            use of library functionality.
569    
570    CM:
571            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
572            are no longer silent.
573    
574    CKIT:
575            Fixed mismatched uses of Int32 and LargeInt.  I always decided
576            in favor of LargeInt -- which is now the same as IntInf.
577            CKIT-knowledgable people should check whether this is what's
578            intended and otherwise change things back to using Int32 or
579            FixedInt.
580    
581    Throughout the code:
582            Started using IntInf.int literals and built-in operations
583            (e.g., comparison with 0) where this seems appropriate.
584    
585    
586    ----------------------------------------------------------------------
587    Name: Dave MacQueen (dbm@cs.uchicago.edu)
588    Date: 2003/08/13 11:36:00 CDT
589    Tag: dbm-20030813-mcz-merge1
590    Description:
591    
592    Merging changes from the mcz-branch development branch into trunk.
593    These changes involve replacement of the emulated old prettyprinter
594    interface with direct use of the SML/NJ Lib PP library, and fixing
595    of a couple of bugs (895, 1186) relating to error messages.  A new
596    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
597    has been added.
598    
599    ----------------------------------------------------------------------
600    Name: Matthias Blume (blume (at) tti - c (dot) org)
601    Date: 2003/08/11 15:45:00 CDT
602    Tag: blume-20030811-windows
603    Description:
604    
605    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
606    
607        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
608    
609    This patch restores SML/NJ's ability to run under win32.  There are a
610    number of changes, including fixes for several bugs that had gone
611    unnoticed until now:
612    
613      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
614        trying to run the win32 version from within cygwin.)
615    
616      - There are a number of simple .bat scripts that substitute for their
617        corresponding Unix shell-scripts.  (See below.)
618    
619      - The internals of ml-build have been modified slightly.  The main
620        difference is that instead of calling ".link-sml" (or link-sml.bat)
621        using OS.Process.system, the ML process delegates this task back
622        to the script.  Otherwise problems arise in mixed environments such
623        as Cygwin where scripts look and work like Unix scripts, but
624        where OS.Process.system cannot run them.
625    
626      - In CM, the srcpath pickler used native pathname syntax -- which
627        is incorrect in the case of cross-compilation.  The new pickle format
628        is independent of platform-specific naming conventions.
629    
630      - Path configuration files (such as lib/pathconfig) can now choose
631        between native and standard syntax.  Placing a line of the form
632    
633            standard!
634    
635        into the file causes all subsequent paths to be interpreted using
636        CM standard pathname syntax (= Unix conventions); a line
637    
638            native!
639    
640        switches back to native style.  This was needed so that
641        path config files can be written portably, see src/system/pathconfig.
642    
643      - Runtime system:
644    
645          - win32-filesys.c:  get_file_time and set_file_time now
646            access modification time, not creation time.
647    
648          - I/O code made aware of new array representation.
649    
650          - Bug fixes in X86.prim.masm.
651    
652       - src/system/makeml made aware of win32. (For use under cygwin
653            and other Unix-environments for windows.)
654    
655       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
656         which caused BinIO.inputAll to fail consistently.
657    
658    .bat scripts:
659    
660       Windows .bat scripts assume that SMLNJ_HOME is defined.
661    
662       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
663         applications (sml, ml-yacc, ml-lex).
664       - ml-build.bat: analogous to ml-build.
665       - config\install.bat:  Analogous to config/install.sh.  This requires
666         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
667         (nmake etc. must be on the path, and vcvars32 must have been run.)
668         Moreover, sources for ml-lex and ml-yacc need to exist under src,
669         and the bootfile hierarchy must have been unpacked under
670         sml.boot.x86-win32.
671         The script is very primitive and does a poor job at error checking.
672         It only installs the base system, ml-lex, and ml-yacc.  No other
673         libraries are being installed (i.e., you get only those that
674         are part of the compiler.)
675       - link-sml.bat:  analogous to .link-sml, but not currently used
676    
677    Unrelated bug fixes:
678    
679      - ml-nlffigen now exports structures ST_* corresponding to incomplete
680        types.
681      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
682        otherwise.)
683    
684    ----------------------------------------------------------------------
685    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
686    Date: 2003/06/17
687    Tag: macqueen-20030617-bug895
688    Description:
689    
690    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
691    Tag will be used for new development branch (mcz-branch) for
692    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
693    
694    ----------------------------------------------------------------------
695    Name: Matthias Blume (blume (at) tti - c (dot) org)
696    Date: 2003/05/27 16:55:00 CDT
697    Tag: blume-20030527-polyeq
698    Description:
699    
700    Tried to eliminated most cases of polymorphic equality.
701    
702    ----------------------------------------------------------------------
703    Name: Matthias Blume (blume (at) tti - c (dot) org)
704    Date: 2003/05/21 17:45:00 CDT
705    Tag: blume-20030517-complete
706    Description:
707    
708    Two changes:
709    
710      1. Added a flag for controlling whether non-exhaustive bindings will
711         be treated as errors (default is false).
712      2. Cleaned up the *entire* source tree so that CMB.make goes through
713         without a single non-exhaustive match- or bind warning.
714    
715    ----------------------------------------------------------------------
716    Name: Matthias Blume (blume (at) tti - c (dot) org)
717    Date: 2003/05/17 10:20:00 CDT
718    Tag: blume-20030517-absyn
719    Description:
720    
721    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
722    
723       This mainly affects the quality of error messages.  However, some
724       of the code is now more straightforward than before. (Treatment of
725       the above four constructs in translate.sml is much simpler than
726       the "macro-expansion" that was going on before.  Plus, the mach-
727       compiler no longer gets invoked just to be able to compile an
728       if-expression.)
729    
730    2. The ErrorMsg.Error exception is now caught and absorbed by the
731       interactive loop.
732    
733    ----------------------------------------------------------------------
734    Name: Allen Leung
735    Date: 2003/05/16 13:05:00 CDT
736    Tag: leunga-20030516-cygwin-runtime
737    Description:
738    
739        Ported the runtime system to cygwin, which uses the unix
740    x86-unix bin files.  Missing/buggy features:
741    
742       o getnetbyname, getnetbyaddr: these functions seem to be missing in
743         the Cygwin library.
744       o Ctrl-C handling may be flaky.
745       o Windows system calls and Windows I/O are not supported.
746    
747       A new set of binfiles is located at:
748    
749         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
750    
751       This is only needed for bootstrapping the cygwin version of smlnj.
752       Other x86 versions can use the existing binfiles.
753    
754    ----------------------------------------------------------------------
755    Name: Matthias Blume
756    Date: 2003/04/08 15:42:00 CDT
757    Tag: blume-20030408-listpair
758    Description:
759    
760    1. Added a target 'mlrisc' to installer.
761    
762    2. Added missing elements to structure ListPair.
763    
764    ----------------------------------------------------------------------
765    Name: Allen Leung
766    Date: 2003/01/07 10:40:00 EST
767    Tag: leunga-20030107-int-rem
768    Description:
769    
770    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
771    The arguments to the SUBL instruction were swapped.
772    
773    ----------------------------------------------------------------------
774    Name: Matthias Blume
775    Date: 2002/12/12 16:25:00 EST
776    Tag: blume-20021212-risc-ra
777    Description:
778    
779    Fixed a serious bug in the rewrite code for FP spilling/reloading that
780    sent the RA into an infinite loop when floating point registers get
781    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
782    110.37 and 110.38.)
783    There was another set of potential problems related to the handling of
784    MLRISC annotations (but those did not yet cause real problems, apparently).
785    
786    ----------------------------------------------------------------------
787    Name: Matthias Blume
788    Date: 2002/12/06 22:40:00 EST
789    Tag: blume-20021206-cm-fileid
790    Description:
791    
792    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
793    This fixes the problem of CM getting confused by files that suddenly
794    change their identity (e.g., by getting unlinked and recreated by some
795    text editor such as vi).  There might be a better/cheaper/cleaner way
796    of doing this, but for now this will have to do.
797    
798    ----------------------------------------------------------------------
799    Name: Matthias Blume
800    Date: 2002/10/28 09:50:00 EST
801    Tag: blume-20021028-typecheck
802    Description:
803    
804    Exported structure Typecheck from $smlnj/viscomp/core.cm.
805    
806    ----------------------------------------------------------------------
807    Name: Matthias Blume
808    Date: 2002/10/17 09:10:00 EDT
809    Tag: Release_110_42
810    Description:
811    
812    In good old tradition, there has been a slight hiccup so that we have
813    to patch 110.42 after the fact.   The old release tag has been replaced
814    (see below).
815    
816    The change solves a problem with two competing approaches the
817    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
818    each other's way.
819    
820    This change only affects the runtime system code and the installer script.
821    (No new bootfiles.)
822    
823    ----------------------------------------------------------------------
824    Name: Matthias Blume
825    Date: 2002/10/16 12:00:00 EDT
826    Tag: Release_110_42_removed
827    Description:
828    
829    New working release.  New bootfiles.
830    
831    ----------------------------------------------------------------------
832    Name: Matthias Blume
833    Date: 2002/10/10 13:10:00 EDT
834    Tag: blume-20021010-ppc-divs
835    Description:
836    
837    The mltree operator DIVS must be implemented with an overflow check on
838    the PPC because the hardware indicates divide-by-zero using "overflow" as
839    well.
840    
841    ----------------------------------------------------------------------
842    Name: Matthias Blume
843    Date: 2002/07/23 11:45:00
844    Tag: blume-20020723-smlnj-home
845    Description:
846    
847    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
848    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
849    CM_PATHCONFIG is also set), the path configuration file is assumed
850    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
851    the entire tree to some other place and everything will "just work".
852    
853    (Companion commands such as ml-build and ml-makedepend also sense this
854    variable.)
855    
856    ----------------------------------------------------------------------
857    Name: Matthias Blume
858    Date: 2002/07/12 21:19:00 EDT
859    Tag: blume-20020712-liveness
860    Description:
861    
862    Exported two useful "step" functions from liveness module (MLRISC).
863    
864    ----------------------------------------------------------------------
865    Name: Matthias Blume
866    Date: 2002/07/05 16:00 EDT
867    Tag: Release_110_41
868    Description:
869    
870    New working release.  New bootfiles.
871    
872    ----------------------------------------------------------------------
873    Name: Matthias Blume
874    Date: 2002/07/05 10:25:00 EDT
875    Tag: blume-20020705-btimp
876    Description:
877    
878    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
879    other clients can set up backtracing support.
880    
881    ----------------------------------------------------------------------
882    Name: Matthias Blume
883    Date: 2002/06/25 17:23:00 EDT
884    Tag: blume-20020625-fpmax
885    Description:
886    
887    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
888    case.  (The sense of the isNaN test was reversed -- which made min and
889    max always return their first argument.)
890    
891    ----------------------------------------------------------------------
892    Name: Matthias Blume
893    Date: 2002/06/11
894    Tag: blume-20020611-unixpath
895    Description:
896    
897    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
898    
899    ----------------------------------------------------------------------
900    Name: Matthias Blume
901    Date: 2002/06/10 16:35:00 EDT
902    Tag: blume-20020610-ieeereal
903    Description:
904    
905    I back-ported my implementation of IEEEReal.fromString from the
906    idlbasis-devel branch so that we can test it.
907    
908    Another small change is that ppDec tries to give more information
909    than just "<sig>" in the case of functors.  However, this code is
910    broken in some mysterious way if the functor's body's signature
911    has not been declared by ascription but gets inferred from the
912    implementation.  This needs fixing...
913    
914    ----------------------------------------------------------------------
915    Name: Matthias Blume
916    Date: 2002/05/31
917    Tag: blume-20020531-btrace-mode
918    Description:
919    
920    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
921    the wayside when I switched over to using Controls everywhere.)
922    
923    ----------------------------------------------------------------------
924    Name: Lal George
925    Date: 2002/05/23 12:21:40 EDT
926    Tag: george-20020523-visual-labels
927    Description:
928    
929    Labels are now displayed in the graphical output to make
930    the fall-through and target blocks obvious.
931    
932    ----------------------------------------------------------------------
933    Name: Matthias Blume
934    Date: 2002/05/22 11:03:00 EDT
935    Tag: blume-20020522-shrink
936    Description:
937    
938    John tweaked yesterday's fix for 1131 to handle an out-of-memory
939    situation that comes up when allocating huge arrays.
940    
941    ----------------------------------------------------------------------
942    Name: Matthias Blume
943    Date: 2002/05/21 16:00:00 EDT
944    Tag: Release_110_40
945    Description:
946    
947    New working release (110.40).  New bootfiles.
948    
949    [Also: John Reppy fixed GC bug 1131.]
950    
951    ----------------------------------------------------------------------
952    Name: Matthias Blume
953    Date: 2002/05/21 12:35:00 EDT
954    Tag: blume-20020521-cmdoc
955    Description:
956    
957    CM documentation update.
958    
959    ----------------------------------------------------------------------
960    Name: Matthias Blume
961    Date: 2002/05/21 10:55:00 EDT
962    Tag: blume-20020521-misc
963    Description:
964    
965    - John tweaked runtime to be silent on heap export (except when
966      GC messages are on).
967    - I added a few more things (cross-compiling versions of CMB) to
968      config/preloads (as suggestions).
969    
970    ----------------------------------------------------------------------
971    Name: Matthias Blume
972    Date: 2002/05/20 22:25:00 EDT
973    Tag: blume-20020520-controls
974    Description:
975    
976    - Added ControlUtil structure to control-lib.cm.
977    - Use it throughout.
978    - Used Controls facility to define MLRISC controls (as opposed to
979      registering MLRISC control ref cells with Controls after the fact)
980    - Fixed messed-up controls priorities.
981    
982    * Removed again all the stuff from config/preloads that one wouldn't
983      be able to preload at the time the initial heap image is built.
984      (Many libraries, e.g., CML, do not exist yet at this time.  The
985       only libraries that can be preloaded via config/preloads are those
986       that come bundled with the bootfiles.)
987    
988    ----------------------------------------------------------------------
989    Name: Matthias Blume
990    Date: 2002/05/20 10:59:00 EDT
991    Tag: blume-20020520-preloads
992    Description:
993    
994    Added a lot of commented-out suggestions for things to be included
995    in config/preloads.
996    
997    ----------------------------------------------------------------------
998    Name: Allen Leung
999    Date: 2002/05/18 14:20:00 EDT
1000    Tag: leunga-20020518-mdl
1001    Description:
1002    
1003    o Made the mdl tool stuff compile and run again.
1004    o I've disabled all the stuff that depends on RTL specifications; they
1005      are all badly broken anyway.
1006    
1007    ----------------------------------------------------------------------
1008    Name: Matthias Blume
1009    Date: 2002/05/17 16:49:00 EDT
1010    Tag: blume-20020517-controls
1011    Description:
1012    
1013    0. John Reppy made several modifications to the SML/NJ library.
1014       In particular, there is a shiny new controls-lib.cm.
1015    
1016    1. Pushed new controls interface through compiler so that everything
1017       compiles again.
1018    
1019    2. Added FormatComb and FORMAT_COMB to the CML version of the
1020       SML/NJ library (so that CML compiles again).
1021    
1022    3. Modified init scripts because XXX_DEFAULT environment variables
1023       are no longer with us.  (Boot-time initialization is now done
1024       using the same environment variables that are also used for
1025       startup-time initialization of controls.)
1026    
1027    ----------------------------------------------------------------------
1028    Name: Lal George
1029    Date: 2002/05/15 09:20:10 EDT
1030    Tag: george-20020515-pseudo-op-decls
1031    Description:
1032    
1033            All pseudo-ops emitted before the first segment declaration
1034    such as TEXT, DATA, and BSS directives are assumed to be global
1035    declarations and are emitted first in the assembly file. This is
1036    useful in a number of situations where one has pseudo-ops that are not
1037    specific to any segment, and also works around the constraint that one
1038    cannot have client pseudo-ops in the TEXT segment.
1039    
1040            Because no segment is associated with these declarations it is
1041    an error to allocate any space or objects before the first segment
1042    directive and an exception will be raised. However, we cannot make
1043    this check for client pseudo-ops.
1044    
1045            These top level declarations are a field in the CFG graph_info.
1046    In theory you can continue to add to this field after the CFG has been
1047    built -- provided you know what you are doing;-)
1048    
1049    ----------------------------------------------------------------------
1050    Name: Matthias Blume
1051    Date: 2002/05/13 16:40:00 EDT
1052    Tag: blume-20020513-pp-etc
1053    Description:
1054    
1055    A few minor bugfixes:
1056    
1057      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1058        (Bogus printouts for redefined bindings still occur.  Compiler
1059        bug should no longer occur now.  We need to redo the prettyprinter
1060        from scratch.)
1061    
1062      - CM pathname printer now also adds escape sequences for ( and )
1063    
1064      - commend and docu fixes for ml-nlffi
1065    
1066    ----------------------------------------------------------------------
1067    Name: Matthias Blume
1068    Date: 2002/05/10 16:40:00 EDT
1069    Tag: blume-20020510-erg-textio
1070    Description:
1071    
1072    Applied the following bugfix provided by Emden Gansner:
1073    
1074        Output is corrupted when outputSubstr is used rather than output.
1075    
1076        The problem occurs when a substring
1077    
1078            ss = (s, dataStart, dataLen)
1079    
1080        where dataStart > 0, fills a stream buffer with avail bytes left.
1081        avail bytes of s, starting at index dataStart, are copied into the
1082        buffer, the buffer is flushed, and then the remaining dataLen-avail
1083        bytes of ss are copied into the beginning of the buffer. Instead of
1084        starting this copy at index dataStart+avail in s, the current code
1085        starts the copy at index avail.
1086    
1087        Fix:
1088        In text-io-fn.sml, change line 695 from
1089             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1090        to
1091             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1092    
1093    ----------------------------------------------------------------------
1094    Name: Matthias Blume
1095    Date: 2002/04/12 13:55:00 EDT
1096    Tag: blume-20020412-assyntax
1097    Description:
1098    
1099    1. Grabbed newer assyntax.h from the XFree86 project.
1100    2. Fiddled with how to compile X86.prim.asm without warnings.
1101    3. (Very) Minor cleanup in CM.
1102    
1103    ----------------------------------------------------------------------
1104    Name: Matthias Blume
1105    Date: 2002/04/01 (no joke!) 17:07:00 EST
1106    Tag: blume-20020401-x86div
1107    Description:
1108    
1109    Added full support for div/mod/rem/quot on the x86, using the machine
1110    instruction's two results (without clumsily recomputing the remainder)
1111    directly where appropriate.
1112    
1113    Some more extensive power-of-two support was added to the x86 instruction
1114    selector (avoiding expensive divs, mods, and muls where they can be
1115    replaced with cheaper shifts and masks).  However, this sort of thing
1116    ought to be done earlier, e.g., within the CPS optimizer so that
1117    all architectures benefit from it.
1118    
1119    The compiler compiles to a fixed point, but changes might be somewhat
1120    fragile nevertheless.  Please, report any strange things that you might
1121    see wrt. div/mod/quot/rem...
1122    
1123    ----------------------------------------------------------------------
1124    Name: Matthias Blume
1125    Date: 2002/03/29 17:22:00
1126    Tag: blume-20020329-div
1127    Description:
1128    
1129    Fixed my broken div/mod logic.  Unfortunately, this means that the
1130    inline code for div/mod now has one more comparison than before.
1131    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1132    The problem was with quotient = 0, because that alone does not tell
1133    us which way the rounding went.  One then has to look at whether
1134    remainder and divisor have the same sign...  :(
1135    
1136    Anyway, I replaced the bootfiles with fresh ones...
1137    
1138    ----------------------------------------------------------------------
1139    Name: Matthias Blume
1140    Date: 2002/03/29 14:10:00 EST
1141    Tag: blume-20020329-inlprims
1142    Description:
1143    
1144    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1145    
1146    Primops have changed. This means that the bin/boot-file formats have
1147    changed as well.
1148    
1149    To make sure that there is no confusion, I made a new version.
1150    
1151    
1152    CHANGES:
1153    
1154    * removed REMT from mltree (remainder should never overflow).
1155    
1156    * added primops to deal with divisions of all flavors to the frontend
1157    
1158    * handled these primops all the way through so they map to their respective
1159      MLRISC support
1160    
1161    * used these primops in the implementation of Int, Int32, Word, Word32
1162    
1163    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1164    
1165    * parameterized INLMIN, INLMAX, and INLABS by a numkind
1166    
1167    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1168      floating point
1169    
1170    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1171      and Real (but Real.abs maps to a separate floating-point-only primop)
1172    
1173    
1174    TODO items:
1175    
1176    * Hacked Alpha32 instruction selection, disabling the selection of REMx
1177      instructions because the machine instruction encoder cannot handle
1178      them.  (Hppa, PPC, and Sparc instruction selection did not handle
1179      REM in the first place, and REM is supported by the x86 machine coder.)
1180    
1181    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1182      selection phase.  (The two can be streamlined because the hardware
1183      delivers both quotient and remainder at the same time anyway.)
1184    
1185    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1186      (Currently the behavior is inconsistent both across architectures and
1187      wrt. the draft Basis spec.)
1188    
1189    * Word8 should eventually be handled natively, too.
1190    
1191    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1192      as if there currently is no execution path that could trigger it in
1193      SML/NJ.  (The assumptions underlying functions arith and promotable do not
1194      hold for things like multiplication and division.)
1195    
1196    ----------------------------------------------------------------------
1197    Name: Matthias Blume
1198    Date: 2002/03/27 16:27:00 EST
1199    Tag: blume-20020327-mlrisc-divisions
1200    Description:
1201    
1202    Added support for all four division operations (ML's div, mod, quot,
1203    and rem) to MLRISC.  In the course of doing so, I also rationalized
1204    the naming (no more annoying switch-around of DIV and QUOT), by
1205    parameterizing the operation by div_rounding_mode (which can be either
1206    DIV_TO_ZERO or DIV_TO_NEGINF).
1207    
1208    The generic MLTreeGen functor takes care of compiling all four
1209    operations down to only round-to-zero div.
1210    
1211    Missing pieces:
1212    
1213      * Doing something smarter than relying on MLTreeGen on architectures
1214        like, e.g., the x86 where hardware division delivers both quotient and
1215        remainder at the same time.  With this, the implementation of the
1216        round-to-neginf operations could be further streamlined.
1217    
1218      * Remove inlining support for div/mod/rem from the frontend and replace it
1219        with primops that get carried through to the backend.  Do this for all
1220        int and word types.
1221    
1222    ----------------------------------------------------------------------
1223    Name: Matthias Blume
1224    Date: 2002/03/25 17:25:00 EST
1225    Tag: blume-20020325-divmod
1226    Description:
1227    
1228    I improved (hopefully without breaking them) the implementation of Int.div,
1229    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1230    advantage of the following observations:
1231    
1232      Let  q = x quot y      r = x rem y
1233           d = x div  y      m = x mod y
1234    
1235    where "quot" is the round-to-zero version of integer division that
1236    hardware usually provides.  Then we have:
1237    
1238         r = x - q * y        where neither the * nor the - will overflow
1239         d = if q >= 0 orelse x = q * y then q else q - 1
1240                              where neither the * nor the - will overflow
1241         m = if q >= 0 orelse r = 0 then r else r + y
1242                              where the + will not overflow
1243    
1244    This results in substantial simplification of the generated code.
1245    The following table shows the number of CFG nodes and edges generated
1246    for
1247            fun f (x, y) = x OPER y
1248            (* with OPER \in div, mod, quot, rem *)
1249    
1250    
1251        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1252        --------------------------------------------------------
1253         div |         24 |         39 |         12 |         16
1254         mod |         41 |         71 |         12 |         16
1255        quot |          8 |         10 |          8 |         10
1256         rem |         10 |         14 |          8 |         10
1257    
1258    
1259    ----------------------------------------------------------------------
1260    Name: Matthias Blume
1261    Date: 2002/03/25 22:06:00 EST
1262    Tag: blume-20020325-cprotobug
1263    Description:
1264    
1265    Fixed a bug in cproto (c prototype decoder).
1266    
1267    ----------------------------------------------------------------------
1268    Name: Matthias Blume
1269    Date: 2002/03/25 16:00:00 EST
1270    Tag: blume-20020325-raw-primops
1271    Description:
1272    
1273    I did some cleanup to Allen's new primop code and
1274    replaced yesterday's bootfiles with new ones.
1275    (But they are stored in the same place.)
1276    
1277    ----------------------------------------------------------------------
1278    Name: Matthias Blume
1279    Date: 2002/03/24 22:40:00 EST
1280    Tag: blume-20020324-bootfiles
1281    Description:
1282    
1283    Made the bootfiles that Allen asked for.
1284    
1285    ----------------------------------------------------------------------
1286    Name: Allen Leung
1287    Date: 2002/03/23 15:50:00 EST
1288    Tag: leunga-20020323-flint-cps-rcc-primops
1289    Description:
1290    
1291      1. Changes to FLINT primops:
1292    
1293        (* make a call to a C-function;
1294         * The primop carries C function prototype information and specifies
1295         * which of its (ML-) arguments are floating point. C prototype
1296         * information is for use by the backend, ML information is for
1297         * use by the CPS converter. *)
1298      | RAW_CCALL of { c_proto: CTypes.c_proto,
1299                       ml_args: ccall_type list,
1300                       ml_res_opt: ccall_type option,
1301                       reentrant : bool
1302                     } option
1303       (* Allocate uninitialized storage on the heap.
1304        * The record is meant to hold short-lived C objects, i.e., they
1305        * are not ML pointers.  With the tag, the representation is
1306        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1307        *)
1308      | RAW_RECORD of {tag:bool,sz:int}
1309      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1310    
1311      2.  These CPS primops are now overloaded:
1312    
1313           rawload of {kind:numkind}
1314           rawstore of {kind:numkind}
1315    
1316          The one argument form is:
1317    
1318             rawload {kind} address
1319    
1320          The two argument form is:
1321    
1322             rawload {kind} [ml object, byte-offset]
1323    
1324      3. RAW_CCALL/RCC now takes two extra arguments:
1325    
1326         a. The first is whether the C call is reentrant, i.e., whether
1327            ML state should be saved and restored.
1328         b. The second argument is a string argument specifying the name of
1329            library and the C function.
1330    
1331         These things are currently not handled in the code generator, yet.
1332    
1333      4. In CProto,
1334    
1335         An encoding type of "bool" means "ml object" and is mapped into
1336         C prototype of PTR.  Note that "bool" is different than "string",
1337         even though "string" is also mapped into PTR, because "bool"
1338         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1339    
1340      5. Pickler/unpicker
1341    
1342         Changed to handle RAW_RECORD and newest RAW_CCALL
1343    
1344      6. MLRiscGen,
1345    
1346         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1347         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1348            in the file CodeGen/cpscompile/cps-c-calls.sml
1349    
1350      7. Added the conditional move operator
1351    
1352             condmove of branch
1353    
1354         to cps.  Generation of this is still buggy so it is currently
1355         disabled.
1356    
1357    ----------------------------------------------------------------------
1358    Name: Lal George
1359    Date: 2002/03/22 14:18:25 EST
1360    Tag: george-20020322-cps-branch-prob
1361    Description:
1362    
1363    Implemented the Ball-Larus branch prediction-heuristics, and
1364    incorporated graphical viewers for control flow graphs.
1365    
1366    Ball-Larus Heuristics:
1367    ---------------------
1368    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1369    
1370    By design it uses the Dempster-Shafer theory for combining
1371    probabilities.  For example, in the function:
1372    
1373        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1374    
1375    the ball-larus heuristics predicts that the n=0 is unlikely
1376    (OH-heuristic), and the 'then' branch is unlikely because of the
1377    RH-heuristic -- giving the 'then' branch an even lower combined
1378    probability using the Dempster-Shafer theory.
1379    
1380    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1381    probability of the 'then' branch because of the loop in the else
1382    branch.
1383    
1384    
1385    Graphical Viewing:
1386    ------------------
1387    I merely plugged in Allen's graphical viewers into the compiler. The
1388    additional code is not much. At the top level, saying:
1389    
1390            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1391    
1392    will display the graphical view of the control flow graph just before
1393    back-patching.  daVinci must be in your path for this to work. If
1394    daVinci is not available, then the default viewer can be changed
1395    using:
1396    
1397            Control.MLRISC.getString "viewer"
1398    
1399    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1400    course, these viewers must be in your path.
1401    
1402    The above will display the compilation unit at the level of clusters,
1403    many of which are small, boring, and un-interesting. Also setting:
1404    
1405            Control.MLRISC.getInt "cfg-graphical-view_size"
1406    
1407    will display clusters that are larger than the value set by the above.
1408    
1409    
1410    ----------------------------------------------------------------------
1411    Name: Matthias Blume
1412    Date: 2002/03/21 22:20:00 EST
1413    Tag: blume-20020321-kmp-bugfix
1414    Description:
1415    
1416    Changed the interface to the KMP routine in PreString and fixed
1417    a minor bug in one place where it was used.
1418    
1419    ----------------------------------------------------------------------
1420    Name: Allen Leung
1421    Date: 2002/03/21 20:30:00 EST
1422    Tag: leunga-20020321-cfg
1423    Description:
1424    
1425      Fixed a potential problem in cfg edge splitting.
1426    
1427    ----------------------------------------------------------------------
1428    Name: Allen Leung
1429    Date: 2002/03/21 17:15:00 EST
1430    Tag: leunga-20020321-x86-fp-cfg
1431    Description:
1432    
1433      1. Recoded the buggy parts of x86-fp.
1434    
1435         a. All the block reordering code has been removed.
1436            We now depend on the block placement phases to do this work.
1437    
1438         b. Critical edge splitting code has been simplified and moved into the
1439            CFG modules, as where they belong.
1440    
1441         Both of these were quite buggy and complex.  The code is now much, much
1442         simpler.
1443    
1444      2. X86 backend.
1445    
1446         a. Added instructions for 64-bit support.  Instruction selection for
1447            64-bit has not been committed, however, since that
1448            requires changes to MLTREE which haven't been approved by
1449            Lal and John.
1450    
1451         b. Added support for FUCOMI and FUCOMIP when generating code for
1452            PentiumPro and above.  We only generate these instructions in
1453            the fast-fp mode.
1454    
1455         c. Added cases for JP and JNP in X86FreqProps.
1456    
1457      3. CFG
1458    
1459         CFG now has a bunch of methods for edge splitting and merging.
1460    
1461      4. Machine description.
1462    
1463         John's simplification of MLTREE_BASIS.fcond broke a few machine
1464         description things:
1465    
1466         rtl-build.{sig,sml} and hppa.mdl fixed.
1467    
1468         NOTE: the machine description stuff in the repository is still broken.
1469               Again, I can't put my fixes in because that involves
1470               changes to MLTREE.
1471    
1472    ----------------------------------------------------------------------
1473    Name: Matthias Blume
1474    Date: 2002/03/20 15:55:00 EST
1475    Tag: blume-20020320-kmp
1476    Description:
1477    
1478    Implemented Knuth-Morris-Pratt string matching in PreString and used
1479    it for String.isSubstring, Substring.isSubstring, and
1480    Substring.position.
1481    
1482    (Might need some stress-testing.  Simple examples worked fine.)
1483    
1484    ----------------------------------------------------------------------
1485    Name: Matthias Blume
1486    Date: 2002/03/19 16:37:00 EST
1487    Tag: blume-20020319-witnesses
1488    Description:
1489    
1490    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1491    
1492    This implements a generic mechanism for changing constness qualifiers
1493    anywhere within big C types without resorting to outright "casts".
1494    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1495    modify the constness at the outermost level.)
1496    The implementation of "convert" is based on the idea of "witness"
1497    values -- values that are not used by the operation but whose types
1498    "testify" to their applicability.  On the implementation side, "convert"
1499    is simply a projection (returning its second curried argument).  With
1500    cross-module inlining, it should not result in any machine code being
1501    generated.
1502    
1503    ----------------------------------------------------------------------
1504    Name: Matthias Blume
1505    Date: 2002/03/15 16:40:00 EST
1506    Tag: blume-20020315-basis
1507    Description:
1508    
1509    Provided (preliminary?) implementations for
1510    
1511      {String,Substring}.{concatWith,isSuffix,isSubstring}
1512    
1513    and
1514    
1515      Substring.full
1516    
1517    Those are in the Basis spec but they were missing in SML/NJ.
1518    
1519    ----------------------------------------------------------------------
1520    Name: Matthias Blume
1521    Date: 2002/03/14 21:30:00 EST
1522    Tag: blume-20020314-controls
1523    Description:
1524    
1525    Controls:
1526    ---------
1527    
1528    1. Factored out the recently-added Controls : CONTROLS stuff and put
1529       it into its own library $/controls-lib.cm.  The source tree for
1530       this is under src/smlnj-lib/Controls.
1531    
1532    2. Changed the names of types and functions in this interface, so they
1533       make a bit more "sense":
1534    
1535          module -> registry
1536          'a registry -> 'a group
1537    
1538    3. The interface now deals in ref cells only.  The getter/setter interface
1539       is (mostly) gone.
1540    
1541    4. Added a function that lets one register an already-existing ref cell.
1542    
1543    5. Made the corresponding modifications to the rest of the code so that
1544       everything compiles again.
1545    
1546    6. Changed the implementation of Controls.MLRISC back to something closer
1547       to the original.  In particular, this module (and therefore MLRISC)
1548       does not depend on Controls.  There now is some link-time code in
1549       int-sys.sml that registers the MLRISC controls with the Controls
1550       module.
1551    
1552    CM:
1553    ---
1554    
1555      * One can now specify the lambda-split aggressiveness in init.cmi.
1556    
1557    ----------------------------------------------------------------------
1558    Name: Allen Leung
1559    Date: 2002/03/13 17:30:00 EST
1560    Tag: leunga-20020313-x86-fp-unary
1561    Description:
1562    
1563    Bug fix for:
1564    
1565    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1566    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1567    > - fun f(x,(y,z)) = Real.~ y;
1568    > [autoloading]
1569    > [autoloading done]
1570    >       fchsl   (%eax), 184(%esp)
1571    > Error: MLRisc bug: X86MCEmitter.emitInstr
1572    >
1573    > uncaught exception Error
1574    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1575    
1576    The problem was that the code generator did not generate any fp registers
1577    in this case, and the ra didn't know that it needed to run the X86FP phase to
1578    translate the pseudo fp instruction.   This only happened with unary fp
1579    operators in certain situations.
1580    
1581    ----------------------------------------------------------------------
1582    Name: Matthias Blume
1583    Date: 2002/03/13 14:00:00 EST
1584    Tag: blume-20020313-overload-etc
1585    Description:
1586    
1587    1. Added _overload as a synonym for overload for backward compatibility.
1588       (Control.overloadKW must be true for either version to be accepted.)
1589    
1590    2. Fixed bug in install script that caused more things to be installed
1591       than what was requested in config/targets.
1592    
1593    3. Made CM aware of the (_)overload construct so that autoloading
1594       works.
1595    
1596    ----------------------------------------------------------------------
1597    Name: Matthias Blume
1598    Date: 2002/03/12 22:03:00 EST
1599    Tag: blume-20020312-url
1600    Description:
1601    
1602    Forgot to update BOOT and srcarchiveurl.
1603    
1604    ----------------------------------------------------------------------
1605    Name: Matthias Blume
1606    Date: 2002/03/12 17:30:00 EST
1607    Tag: blume-20020312-version110392
1608    Description:
1609    
1610    Yet another version number bump (because of small changes to the
1611    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1612    
1613    Changes:
1614    
1615      The new pid generation scheme described a few weeks ago was overly
1616      complicated.  I implemented a new mechanism that is simpler and
1617      provides a bit more "stability":  Once CM has seen a compilation
1618      unit, it keeps its identity constant (as long as you do not delete
1619      those crucial CM/GUID/* files).  This means that when you change
1620      an interface, compile, then go back to the old interface, and
1621      compile again, you arrive at the original pid.
1622    
1623      There now also is a mechanism that instructs CM to use the plain
1624      environment hash as a module's pid (effectively making its GUID
1625      the empty string).  For this, "noguid" must be specified as an
1626      option to the .sml file in question within its .cm file.
1627      This is most useful for code that is being generated by tools such
1628      as ml-nlffigen (because during development programmers tend to
1629      erase the tool's entire output directory tree including CM's cached
1630      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1631      revert to the old, broken behavior of SML/NJ, but in specific cases
1632      where there is no danger of interface confusion, its use is ok
1633      (I think).
1634    
1635      ml-nlffigen by default generates "noguid" annotations.  They can be
1636      turned off by specifying -guid in its command line.
1637    
1638    ----------------------------------------------------------------------
1639    Name: Lal George
1640    Date: 2002/03/12 12 14:42:36 EST
1641    Tag: george-20020312-frequency-computation
1642    Description:
1643    
1644    Integrated jump chaining and static block frequency into the
1645    compiler. More details and numbers later.
1646    
1647    ----------------------------------------------------------------------
1648    Name: Lal George
1649    Date: 2002/03/11 11 22:38:53 EST
1650    Tag: george-20020311-jump-chain-elim
1651    Description:
1652    
1653    Tested the jump chain elimination on all architectures (except the
1654    hppa).  This is on by default right now and is profitable for the
1655    alpha and x86, however, it may not be profitable for the sparc and ppc
1656    when compiling the compiler.
1657    
1658    The gc test will typically jump to a label at the end of the cluster,
1659    where there is another jump to an external cluster containing the actual
1660    code to invoke gc. This is to allow factoring of common gc invocation
1661    sequences. That is to say, we generate:
1662    
1663            f:
1664               testgc
1665               ja   L1      % jump if above to L1
1666    
1667            L1:
1668               jmp L2
1669    
1670    
1671    After jump chain elimination the 'ja L1' instructions is converted to
1672    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1673    up being implemented in their long form (if L2 is far away) using:
1674    
1675            jbe     L3      % jump if below or equal to L3
1676            jmp     L2
1677         L3:
1678            ...
1679    
1680    
1681    For large compilation units L2  may be far away.
1682    
1683    
1684    ----------------------------------------------------------------------
1685    Name: Matthias Blume
1686    Date: 2002/03/11 13:30:00 EST
1687    Tag: blume-20020311-mltreeeval
1688    Description:
1689    
1690    A functor parameter was missing.
1691    
1692    ----------------------------------------------------------------------
1693    Name: Allen Leung
1694    Date: 2002/03/11 10:30:00 EST
1695    Tag: leunga-20020311-runtime-string0
1696    Description:
1697    
1698       The representation of the empty string now points to a
1699    legal null terminated C string instead of unit.  It is now possible
1700    to convert an ML string into C string with InlineT.CharVector.getData.
1701    This compiles into one single machine instruction.
1702    
1703    ----------------------------------------------------------------------
1704    Name: Allen Leung
1705    Date: 2002/03/10 23:55:00 EST
1706    Tag: leunga-20020310-x86-call
1707    Description:
1708    
1709       Added machine generation for CALL instruction (relative displacement mode)
1710    
1711    ----------------------------------------------------------------------
1712    Name: Matthias Blume
1713    Date: 2002/03/08 16:05:00
1714    Tag: blume-20020308-entrypoints
1715    Description:
1716    
1717    Version number bumped to 110.39.1.  NEW BOOTFILES!
1718    
1719    Entrypoints: non-zero offset into a code object where execution should begin.
1720    
1721    - Added the notion of an entrypoint to CodeObj.
1722    - Added reading/writing of entrypoint info to Binfile.
1723    - Made runtime system bootloader aware of entrypoints.
1724    - Use the address of the label of the first function given to mlriscGen
1725      as the entrypoint.  This address is currently always 0, but it will
1726      not be 0 once we turn on block placement.
1727    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
1728      with entry points) from mlriscGen.
1729    
1730    ----------------------------------------------------------------------
1731    Name: Allen Leung
1732    Date: 2002/03/07 20:45:00 EST
1733    Tag: leunga-20020307-x86-cmov
1734    Description:
1735    
1736       Bug fixes for CMOVcc on x86.
1737    
1738       1. Added machine code generation for CMOVcc
1739       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1740       3. CMOVcc cannot have an immediate operand as argument.
1741    
1742    ----------------------------------------------------------------------
1743    Name: Matthias Blume
1744    Date: 2002/03/07 16:15:00 EST
1745    Tag: blume-20020307-controls
1746    Description:
1747    
1748    This is a very large but mostly boring patch which makes (almost)
1749    every tuneable compiler knob (i.e., pretty much everything under
1750    Control.* plus a few other things) configurable via both the command
1751    line and environment variables in the style CM did its configuration
1752    until now.
1753    
1754    Try starting sml with '-h' (or, if you are brave, '-H')
1755    
1756    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1757    implements the underlying generic mechanism.
1758    
1759    The interface to some of the existing such facilities has changed somewhat.
1760    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1761    (The getFoo interface is still there for backward-compatibility, but its
1762    use is deprecated.)
1763    
1764    The ml-build script passes -Cxxx=yyy command-line arguments through so
1765    that one can now twiddle the compiler settings when using this "batch"
1766    compiler.
1767    
1768    TODO items:
1769    
1770    We should go through and throw out all controls that are no longer
1771    connected to anything.  Moreover, we should go through and provide
1772    meaningful (and correct!) documentation strings for those controls
1773    that still are connected.
1774    
1775    Currently, multiple calls to Controls.new are accepted (only the first
1776    has any effect).  Eventually we should make sure that every control
1777    is being made (via Controls.new) exactly once.  Future access can then
1778    be done using Controls.acc.
1779    
1780    Finally, it would probably be a good idea to use the getter-setter
1781    interface to controls rather than ref cells.  For the time being, both
1782    styles are provided by the Controls module, but getter-setter pairs are
1783    better if thread-safety is of any concern because they can be wrapped.
1784    
1785    *****************************************
1786    
1787    One bug fix: The function blockPlacement in three of the MLRISC
1788    backpatch files used to be hard-wired to one of two possibilities at
1789    link time (according to the value of the placementFlag).  But (I
1790    think) it should rather sense the flag every time.
1791    
1792    *****************************************
1793    
1794    Other assorted changes (by other people who did not supply a HISTORY entry):
1795    
1796    1. the cross-module inliner now works much better (Monnier)
1797    2. representation of weights, frequencies, and probabilities in MLRISC
1798       changed in preparation of using those for weighted block placement
1799       (Reppy, George)
1800    
1801    ----------------------------------------------------------------------
1802    Name: Lal George
1803    Date: 2002/03/07 14:44:24 EST 2002
1804    Tag: george-20020307-weighted-block-placement
1805    
1806    Tested the weighted block placement optimization on all architectures
1807    (except the hppa) using AMPL to generate the block and edge frequencies.
1808    Changes were required in the machine properties to correctly
1809    categorize trap instructions. There is an MLRISC flag
1810    "weighted-block-placement" that can be used to enable weighted block
1811    placement, but this will be ineffective without block/edge
1812    frequencies (coming soon).
1813    
1814    
1815    ----------------------------------------------------------------------
1816    Name: Lal George
1817    Date: 2002/03/05 17:24:48 EST
1818    Tag: george-20020305-linkage-cluster
1819    
1820    In order to support the block placement optimization, a new cluster
1821    is generated as the very first cluster (called the linkage cluster).
1822    It contains a single jump to the 'real' entry point for the compilation
1823    unit. Block placement has no effect on the linkage cluster itself, but
1824    all the other clusters  have full freedom in the manner in which they
1825    reorder blocks or functions.
1826    
1827    On the x86 the typical linkage code that is generated is:
1828       ----------------------
1829            .align 2
1830       L0:
1831            addl    $L1-L0, 72(%esp)
1832            jmp     L1
1833    
1834    
1835            .align  2
1836       L1:
1837       ----------------------
1838    
1839    72(%esp) is the memory location for the stdlink register. This
1840    must contain the address of the CPS function being called. In the
1841    above example, it contains the address of  L0; before
1842    calling L1 (the real entry point for the compilation unit), it
1843    must contain the address for L1, and hence
1844    
1845            addl $L1-L0, 72(%esp)
1846    
1847    I have tested this on all architectures except the hppa.The increase
1848    in code size is of course negligible
1849    
1850    ----------------------------------------------------------------------
1851    Name: Allen Leung
1852    Date: 2002/03/03 13:20:00 EST
1853    Tag: leunga-20020303-mlrisc-tools
1854    
1855      Added #[ ... ] expressions to mlrisc tools
1856    
1857    ----------------------------------------------------------------------
1858    Name: Matthias Blume
1859    Date: 2002/02/27 12:29:00 EST
1860    Tag: blume-20020227-cdebug
1861    Description:
1862    
1863    - made types in structure C and C_Debug to be equal
1864    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1865    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1866    
1867    ----------------------------------------------------------------------
1868    Name: Matthias Blume
1869    Date: 2002/02/26 12:00:00 EST
1870    Tag: blume-20020226-ffi
1871    Description:
1872    
1873    1. Fixed a minor bug in CM's "noweb" tool:
1874       If numbering is turned off, then truly don't number (i.e., do not
1875       supply the -L option to noweb).  The previous behavior was to supply
1876       -L'' -- which caused noweb to use the "default" line numbering scheme.
1877       Thanks to Chris Richards for pointing this out (and supplying the fix).
1878    
1879    2. Once again, I reworked some aspects of the FFI:
1880    
1881       A. The incomplete/complete type business:
1882    
1883       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
1884         gone!
1885       - ML types representing an incomplete type are now *equal* to
1886         ML types representing their corresponding complete types (just like
1887         in C).  This is still safe because ml-nlffigen will not generate
1888         RTTI for incomplete types, nor will it generate functions that
1889         require access to such RTTI.   But when ML code generated from both
1890         incomplete and complete versions of the C type meet, the ML types
1891         are trivially interoperable.
1892    
1893         NOTE:  These changes restore the full generality of the translation
1894         (which was previously lost when I eliminated functorization)!
1895    
1896       B. Enum types:
1897    
1898       - Structure C now has a type constructor "enum" that is similar to
1899         how the "su" constructor works.  However, "enum" is not a phantom
1900         type because each "T enum" has values (and is isomorphic to
1901         MLRep.Signed.int).
1902       - There are generic access operations for enum objects (using
1903         MLRep.Signed.int).
1904       - ml-nlffigen will generate a structure E_foo for each "enum foo".
1905         * The structure contains the definition of type "mlrep" (the ML-side
1906         representation type of the enum).  Normally, mlrep is the same
1907         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
1908         then mlrep will be defined as a datatype -- thus facilitating
1909         pattern matching on mlrep values.
1910         ("-ec" will be suppressed if there are duplicate values in an
1911          enumeration.)
1912         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
1913         will be generated for each C enum constant xxx.
1914         * Conversion functions m2i and i2m convert between mlrep and
1915         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
1916         * Coversion functions c and ml convert between mlrep and "tag enum".
1917         * Access functions (get/set) fetch and store mlrep values.
1918       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
1919         enumerations are merged into one single enumeration represented by
1920         structure E_'.
1921    
1922    ----------------------------------------------------------------------
1923    Name: Allen Leung
1924    Date: 2002/02/25 04:45:00 EST
1925    Tag: leunga-20020225-cps-spill
1926    
1927    This is a new implementation of the CPS spill phase.
1928    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
1929    In case of problems, replace it with the old file spill.sml
1930    
1931    The current compiler runs into some serious performance problems when
1932    constructing a large record.  This can happen when we try to compile a
1933    structure with many items.  Even a very simple structure like the following
1934    makes the compiler slow down.
1935    
1936        structure Foo = struct
1937           val x_1 = 0w1 : Word32.int
1938           val x_2 = 0w2 : Word32.int
1939           val x_3 = 0w3 : Word32.int
1940           ...
1941           val x_N = 0wN : Word32.int
1942        end
1943    
1944    The following table shows the compile time, from N=1000 to N=4000,
1945    with the old compiler:
1946    
1947    N
1948    1000   CPS 100 spill                           0.04u  0.00s  0.00g
1949           MLRISC ra                               0.06u  0.00s  0.05g
1950              (spills = 0 reloads = 0)
1951           TOTAL                                   0.63u  0.07s  0.21g
1952    
1953    1100   CPS 100 spill                           8.25u  0.32s  0.64g
1954           MLRISC ra                               5.68u  0.59s  3.93g
1955              (spills = 0 reloads = 0)
1956           TOTAL                                   14.71u  0.99s  4.81g
1957    
1958    1500   CPS 100 spill                           58.55u  2.34s  1.74g
1959           MLRISC ra                               5.54u  0.65s  3.91g
1960              (spills = 543 reloads = 1082)
1961           TOTAL                                   65.40u  3.13s  6.00g
1962    
1963    2000   CPS 100 spill                           126.69u  4.84s  3.08g
1964           MLRISC ra                               0.80u  0.10s  0.55g
1965              (spills = 42 reloads = 84)
1966           TOTAL                                   129.42u  5.10s  4.13g
1967    
1968    3000   CPS 100 spill                           675.59u  19.03s  11.64g
1969           MLRISC ra                               2.69u  0.27s  1.38g
1970              (spills = 62 reloads = 124)
1971           TOTAL                                   682.48u  19.61s  13.99g
1972    
1973    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
1974           MLRISC ra                               4.96u  0.27s  2.72g
1975              (spills = 85 reloads = 170)
1976           TOTAL                                   2375.26u  57.21s  48.00g
1977    
1978    As you can see the old cps spill module suffers from some serious
1979    performance problem.  But since I cannot decipher the old code fully,
1980    instead of patching the problems up, I'm reimplementing it
1981    with a different algorithm.  The new code is more modular,
1982    smaller when compiled, and substantially faster
1983    (O(n log n) time and O(n) space).  Timing of the new spill module:
1984    
1985    4000  CPS 100 spill                           0.02u  0.00s  0.00g
1986          MLRISC ra                               0.25u  0.02s  0.15g
1987             (spills=1 reloads=3)
1988          TOTAL                                   7.74u  0.34s  1.62g
1989    
1990    Implementation details:
1991    
1992    As far as I can tell, the purpose of the CPS spill module is to make sure the
1993    number of live variables at any program point (the bandwidth)
1994    does not exceed a certain limit, which is determined by the
1995    size of the spill area.
1996    
1997    When the bandwidth is too large, we decrease the register pressure by
1998    packing live variables into spill records.  How we achieve this is
1999    completely different than what we did in the old code.
2000    
2001    First, there is something about the MLRiscGen code generator
2002    that we should be aware of:
2003    
2004    o MLRiscGen performs code motion!
2005    
2006       In particular, it will move floating point computations and
2007       address computations involving only the heap pointer to
2008       their use sites (if there is only a single use).
2009       What this means is that if we have a CPS record construction
2010       statement
2011    
2012           RECORD(k,vl,w,e)
2013    
2014       we should never count the new record address w as live if w
2015       has only one use (which is often the case).
2016    
2017       We should do something similar to floating point, but the transformation
2018       there is much more complex, so I won't deal with that.
2019    
2020    Secondly, there are now two new cps primops at our disposal:
2021    
2022     1. rawrecord of record_kind option
2023        This pure operator allocates some uninitialized storage from the heap.
2024        There are two forms:
2025    
2026         rawrecord NONE [INT n]  allocates a tagless record of length n
2027         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2028                                     and initializes the tag.
2029    
2030     2. rawupdate of cty
2031          rawupdate cty (v,i,x)
2032          Assigns to x to the ith component of record v.
2033          The storelist is not updated.
2034    
2035    We use these new primops for both spilling and increment record construction.
2036    
2037     1. Spilling.
2038    
2039        This is implemented with a linear scan algorithm (but generalized
2040        to trees).  The algorithm will create a single spill record at the
2041        beginning of the cps function and use rawupdate to spill to it,
2042        and SELECT or SELp to reload from it.  So both spills and reloads
2043        are fine-grain operations.  In contrast, in the old algorithm
2044        "spills" have to be bundled together in records.
2045    
2046        Ideally, we should sink the spill record construction to where
2047        it is needed.  We can even split the spill record into multiple ones
2048        at the places where they are needed.  But CPS is not a good
2049        representation for global code motion, so I'll keep it simple and
2050        am not attempting this.
2051    
2052     2. Incremental record construction (aka record splitting).
2053    
2054        Long records with many component values which are simulatenously live
2055        (recall that single use record addresses are not considered to
2056         be live) are constructed with rawrecord and rawupdate.
2057        We allocate space on the heap with rawrecord first, then gradually
2058        fill it in with rawupdate.  This is the technique suggested to me
2059        by Matthias.
2060    
2061        Some restrictions on when this is applicable:
2062        1. It is not a VECTOR record.  The code generator currently does not handle
2063           this case. VECTOR record uses double indirection like arrays.
2064        2. All the record component values are defined in the same "basic block"
2065           as the record constructor.  This is to prevent speculative
2066           record construction.
2067    
2068    ----------------------------------------------------------------------
2069    Name: Allen Leung
2070    Date: 2002/02/22 01:02:00 EST
2071    Tag: leunga-20020222-mlrisc-tools
2072    
2073    Minor bug fixes in the parser and rewriter
2074    
2075    ----------------------------------------------------------------------
2076    Name: Allen Leung
2077    Date: 2002/02/21 20:20:00 EST
2078    Tag: leunga-20020221-peephole
2079    
2080    Regenerated the peephole files.  Some contained typos in the specification
2081    and some didn't compile because of pretty printing bugs in the old version
2082    of 'nowhere'.
2083    
2084    ----------------------------------------------------------------------
2085    Name: Allen Leung
2086    Date: 2002/02/19 20:20:00 EST
2087    Tag: leunga-20020219-mlrisc-tools
2088    Description:
2089    
2090       Minor bug fixes to the mlrisc-tools library:
2091    
2092       1.  Fixed up parsing colon suffixed keywords
2093       2.  Added the ability to shut the error messages up
2094       3.  Reimplemented the pretty printer and fixed up/improved
2095           the pretty printing of handle and -> types.
2096       4.  Fixed up generation of literal symbols in the nowhere tool.
2097       5.  Added some SML keywords to to sml.sty
2098    
2099    ----------------------------------------------------------------------
2100    Name: Matthias Blume
2101    Date: 2002/02/19 16:20:00 EST
2102    Tag: blume-20020219-cmffi
2103    Description:
2104    
2105    A wild mix of changes, some minor, some major:
2106    
2107    * All C FFI-related libraries are now anchored under $c:
2108        $/c.cm      --> $c/c.cm
2109        $/c-int.cm  --> $c/internals/c-int.cm
2110        $/memory.cm --> $c/memory/memory.cm
2111    
2112    * "make" tool (in CM) now treats its argument pathname slightly
2113      differently:
2114        1. If the native expansion is an absolute name, then before invoking
2115           the "make" command on it, CM will apply OS.Path.mkRelative
2116           (with relativeTo = OS.FileSys.getDir()) to it.
2117        2. The argument will be passed through to subsequent phases of CM
2118           processing without "going native".  In particular, if the argument
2119           was an anchored path, then "make" will not lose track of that anchor.
2120    
2121    * Compiler backends now "know" their respective C calling conventions
2122      instead of having to be told about it by ml-nlffigen.  This relieves
2123      ml-nlffigen from one of its burdens.
2124    
2125    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
2126    
2127    * Export C_DEBUG and C_Debug from $c/c.cm.
2128    
2129    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
2130      subtyping relationship between incomplete pointers and their complete
2131      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
2132      with the convention of instantiating 'o with ('t, 'c) obj whenever
2133      the pointer target type is complete.  In the incomplete case, 'o
2134      will be instantiated with some "'c iobj" -- a type obtained by
2135      using one of the functors PointerToIncompleteType or PointerToCompleteType.
2136    
2137      Operations that work on both incomplete and complete pointer types are
2138      typed as taking an 'o ptr while operations that require the target to
2139      be known are typed as taking some ('t, 'c) obj ptr.
2140    
2141      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
2142      where void is an eqtype without any values.  This makes it possible
2143      to work on voidptr values using functions meant to operate on light
2144      incomplete pointers.
2145    
2146    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
2147      been vastly simplified.
2148    
2149    ----------------------------------------------------------------------
2150    Name: Matthias Blume
2151    Date: 2002/02/19 10:48:00 EST
2152    Tag: blume-20020219-pqfix
2153    Description:
2154    
2155    Applied Chris Okasaki's bug fix for priority queues.
2156    
2157    ----------------------------------------------------------------------
2158    Name: Matthias Blume
2159    Date: 2002/02/15 17:05:00
2160    Tag: Release_110_39
2161    Description:
2162    
2163    Last-minute retagging is becoming a tradition... :-(
2164    
2165    This is the working release 110.39.
2166    
2167    ----------------------------------------------------------------------
2168    Name: Matthias Blume
2169    Date: 2002/02/15 16:00:00 EST
2170    Tag: Release_110_39-orig
2171    Description:
2172    
2173    Working release 110.39.  New bootfiles.
2174    
2175    (Update: There was a small bug in the installer so it wouldn't work
2176    with all shells.  So I retagged. -Matthias)
2177    
2178    ----------------------------------------------------------------------
2179    Name: Matthias Blume
2180    Date: 2002/02/15 14:17:00 EST
2181    Tag: blume-20020215-showbindings
2182    Description:
2183    
2184    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
2185    the latter can be useful for exploring what bindings are available at
2186    the interactive prompt.  (The first function returns only the list
2187    of symbols that are really bound, the second prints those but also the
2188    ones that CM's autoloading mechanism knows about.)
2189    
2190    ----------------------------------------------------------------------
2191    Name: Matthias Blume
2192    Date: 2002/02/15 12:08:00 EST
2193    Tag: blume-20020215-iptrs
2194    Description:
2195    
2196    Two improvements to ml-nlffigen:
2197    
2198      1. Write files only if they do not exist or if their current contents
2199         do not coincide with what's being written.  (That is, avoid messing
2200         with the time stamps unless absolutely necessary.)
2201    
2202      2. Implement a "repository" mechanism for generated files related
2203         to "incomplete pointer types".   See the README file for details.
2204    
2205    ----------------------------------------------------------------------
2206    Name: Matthias Blume
2207    Date: 2002/02/14 11:50:00 EST
2208    Tag: blume-20020214-quote
2209    Description:
2210    
2211    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
2212    because of the new and improved tag generation scheme.  (Thanks to Allen
2213    Leung for pointing it out.)
2214    
2215    ----------------------------------------------------------------------
2216    Name: Lal George
2217    Date: 2002/02/14 09:55:27 EST 2002
2218    Tag: george-20020214-isabelle-bug
2219    Description:
2220    
2221    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2222    of Isabelle on the x86.
2223    
2224    From Allen:
2225    -----------
2226     I've found the problem:
2227    
2228         in ra-core.sml, I use the counter "blocked" to keep track of the
2229         true number of elements in the freeze queue.  When the counter goes
2230         to zero, I skip examining the queue.  But I've messed up the
2231         bookkeeping in combine():
2232    
2233             else ();
2234             case !ucol of
2235               PSEUDO => (if !cntv > 0 then
2236                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2237                                        ^^^^^^^^^^^^^^^^^^^^^^^
2238                      moveu := mergeMoveList(!movev, !moveu)
2239                     )
2240                  else ();
2241    
2242         combine() is called to coalesce two nodes u and v.
2243         I think I was thinking that if the move counts of u and v are both
2244         greater than zero then after they are coalesced then one node is
2245         removed from the freeze queue.  Apparently I was thinking that
2246         both u and v are of low degree, but that's clearly not necessarily true.
2247    
2248    
2249    02/12/2002:
2250        Here's the patch.  HOL now compiles.
2251    
2252        I don't know how this impact on performance (compile
2253        time or runtime).  This bug caused the RA (especially on the x86)
2254        to go thru the potential spill phase when there are still nodes on the
2255        freeze queue.
2256    
2257    
2258    
2259    
2260    ----------------------------------------------------------------------
2261    Name: Matthias Blume
2262    Date: 2002/02/13 22:40:00 EST
2263    Tag: blume-20020213-fptr-rtti
2264    Description:
2265    
2266    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2267    updates.
2268    
2269    ----------------------------------------------------------------------
2270    Name: Matthias Blume
2271    Date: 2002/02/13 16:41:00 EST
2272    Tag: blume-20020213-cmlpq
2273    Description:
2274    
2275    Added new priority queue export symbols (which have just been added to
2276    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2277    not compile and the installer would choke.)
2278    
2279    ----------------------------------------------------------------------
2280    Name: Matthias Blume
2281    Date: 2002/02/13 16:15:00 EST
2282    Tag: blume-20020213-various
2283    Description:
2284    
2285    1. More tweaks to ml-nlffigen:
2286    
2287       - better internal datastructures (resulting in slight speedup)
2288       - "-match" option requires exact match
2289       - "localized" gensym counters (untagged structs/unions nested within
2290         other structs/unions or within typedefs get a fresh counter; their
2291         tag will be prefixed by a concatenation of their parents' tags)
2292       - bug fixes (related to calculation of transitive closure of types
2293         to be included in the output)
2294    
2295    2. Minor Basis updates:
2296    
2297       - added implementations for List.collate and Option.app
2298    
2299    ----------------------------------------------------------------------
2300    Name: Matthias Blume
2301    Date: 2002/02/11 15:55:00 EST
2302    Tag: blume-20020211-gensym
2303    Description:
2304    
2305    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2306    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2307    names (ML structure names that correspond to unnamed C structs, unions,
2308    and enums), so that separate runs of ml-nlffigen do not clash.
2309    
2310    ----------------------------------------------------------------------
2311    Name: Matthias Blume
2312    Date: 2002/02/11 12:05:00 EST
2313    Tag: blume-20020211-gensml
2314    Description:
2315    
2316    A quick fix for a problem with GenSML (in the pgraph-util library):
2317    Make generation of toplevel "local" optional.  (Strictly speaking,
2318    signature definitions within "local" are not legal SML.)
2319    
2320    Other than that: updates to INSTALL and cm/TODO.
2321    
2322    ----------------------------------------------------------------------
2323    Name: Matthias Blume
2324    Date: 2002/02/08 15:00:00 EST
2325    Tag: blume-20020208-uniquepid
2326    Description:
2327    
2328    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2329    
2330    1. The installer (config/install.sh) has gotten smarter:
2331    
2332         - Configuration options are a bit easier to specify now
2333           (in config/targets).
2334         - Bug in recognizing .tar.bz2 files fixed.
2335         - Installer automatically resolves dependencies between
2336           configuration options (e.g., if you ask for eXene, you will
2337           also get cml -- regardless whether you asked for it or not).
2338         - Installer can run in "quieter mode" by setting the environment
2339           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2340           "completely silent", though.
2341         - Build HashCons library as part of smlnj-lib.
2342    
2343    2. A new scheme for assigning persistent identifiers to compilation
2344       units (and, by extension, to types etc.) has been put into place.
2345       This fixes a long-standing bug where types and even dynamic values
2346       can get internally confused, thereby compromising type safety
2347       (abstraction) and dynamic correctness.  See
2348    
2349         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2350    
2351       for an example of how things could go wrong until now.
2352    
2353       The downside of the new scheme is that pids are not quite as
2354       persistent as they used to be: CM will generate a fresh pid
2355       for every compilation unit that it thinks it sees for the first
2356       time.  That means that if you compile starting from a clean, fresh
2357       source tree at two different times, you end up with different
2358       binaries.
2359    
2360       Cutoff recompilation, however, has not been compromised because
2361       CM keeps pid information in special caches between runs.
2362    
2363    ----------------------------------------------------------------------
2364    Name: Lal George
2365    Date: 2002/02/07 15:34:13 EST 2002
2366    Tag: <none>
2367    Description:
2368    
2369    Compilers that generate assembly code may produce  global labels
2370    whose value is resolved at link time. The various peephole optimization
2371    modules did not take this in account.
2372    
2373    TODO. The Labels.addrOf function should really return an option
2374    type so that clients are forced to deal with this issue, rather
2375    than an exception being raised.
2376    
2377    ----------------------------------------------------------------------
2378    Name: Lal George
2379    Date: 2002/02/06 13:55:02 EST
2380    Tag: george-20020206-ra-breakup
2381    Description:
2382    
2383    1. A bug fix from Allen.
2384    
2385        A typo causes extra fstp %st(0)'s to be generated at compensation
2386        edges, which might cause stack underflow traps at runtime.  This
2387        occurs in fft where there are extraneous fstps right before the 'into'
2388        trap instruction (in this case they are harmless since none of the
2389        integers overflow.)
2390    
2391    2. Pulled out various utility modules that were embedded in the modules
2392       of the register allocator. I need these modules for other purposes, but
2393       they are not complete enough to put into a library (just yet).
2394    ----------------------------------------------------------------------
2395    Name: Matthias Blume
2396    Date: 2002/01/31 16:05:00 EST
2397    Tag: blume-20020131-sparc-ccalls
2398    Description:
2399    
2400    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2401       of extra stack space by mistake.  Fixed.
2402    
2403    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2404    
2405    ----------------------------------------------------------------------
2406    Name: Allen Leung
2407    Date: 2002/01/30
2408    Tag: leunga-20020130-nowhere-bug-fix
2409    Description:
2410    
2411       MLRISC bug fixes:
2412       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2413       2. MachineInt.fromString was negating its value.
2414    
2415    ----------------------------------------------------------------------
2416    Name: Matthias Blume
2417    Date: 2002/01/29
2418    Tag: blume-20020129-INSTALL
2419    Description:
2420    
2421    - Added somewhat detailed installation instructions (file INSTALL).
2422    - Fixed curl-detection bug in config/install.sh.
2423    - It is now possible to select the URL getter using the URLGETTER
2424      environment variable:
2425    
2426          not set / "unknown"      --> automatic detection (script tries wget,
2427                                       curl, and lynx)
2428          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2429                                       how to properly invoke them)
2430          other                    --> use $URLGETTER directly, it must take
2431                                       precisely two command-line arguments
2432                                       (source URL and destination file name)
2433    
2434    ----------------------------------------------------------------------
2435    Name: Matthias Blume
2436    Date: 2002/01/28
2437    Tag: blume-20020128-sparc-ccalls
2438    Description:
2439    
2440    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2441    - Make use of the allocParam argument in sparc-c-calls.
2442    
2443    ----------------------------------------------------------------------
2444    Name: Matthias Blume
2445    Date: 2002/01/28
2446    Tag: blume-20020128-allocParam
2447    Description:
2448    
2449    John Reppy:  Changes c-calls API to accept client-callback for
2450    allocating extra stack space.
2451    me: Corresponding changes to mlriscGen (using a dummy argument that
2452        does not change the current behavior).
2453    
2454    ----------------------------------------------------------------------
2455    Name: Matthias Blume
2456    Date: 2002/01/28 12:00:00
2457    Tag: Release_110_38
2458    Description:
2459    
2460    This time for real!!!
2461    
2462    ----------------------------------------------------------------------
2463    Name: Matthias Blume
2464    Date: 2002/01/28 10:56:00 EST
2465    Tag: blume-20020128-retraction
2466    Description:
2467    
2468    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
2469       with blume-Release_110_38-retracted.)
2470    
2471    1. Fixed a problem with incorrect rounding modes in real64.sml.
2472       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2473    
2474    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2475       and enums fixed.  The naming of corresponding ML identifiers should
2476       now be consistent again.
2477    
2478    ----------------------------------------------------------------------
2479    Name: Allen Leung
2480    Date: 2002/01/27
2481    Tag: leunga-20020127-nowhere
2482    Description:
2483    
2484       Added a target called nowhere in the configuration scripts.
2485       Enabling this will build the MLRISC 'nowhere' tool (for translating
2486       programs with where-clauses into legal SML code) during installation.
2487    
2488    ----------------------------------------------------------------------
2489    Name: Matthias Blume
2490    Date: 2002/01/25 21:27:00 EST
2491    Tag: blume-Release_110_38-retracted
2492    Description:
2493    
2494    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
2495    
2496    README will be added later.
2497    
2498    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
2499    (Release_110_38) removed.  Reason: Last-minute bug fixes.
2500    
2501    ----------------------------------------------------------------------
2502    Name: Matthias Blume
2503    Date: 2002/01/25
2504    Tag: blume-20020125-ffi
2505    Description:
2506    
2507    A large number of tweaks and improvements to ml-nlffi-lib and
2508    ml-nlffigen:
2509    
2510       - ML represenation types have been streamlined
2511       - getter and setter functions work with concrete values, not abstract
2512         ones where possible
2513       - ml-nlffigen command line more flexible (see README file there)
2514       - some bugs have been fixed (hopefully)
2515    
2516    ----------------------------------------------------------------------
2517    Name: Lal George
2518    Date: 2002/01/24
2519    Tag: george-20020124-risc-ra-interface
2520    Description:
2521    
2522       There is a dramatic simplification in the interface to the
2523       register allocator for RISC architectures as a result of making
2524       parallel copy instructions explicit.
2525    
2526    ----------------------------------------------------------------------
2527    Name: Matthias Blume
2528    Date: 2002/01/22
2529    Tag: blume-20020122-x86-ccalls
2530    Description:
2531    
2532    Bug fix for c-calls on x86 (having to do with how char- and
2533    short-arguments are being handled).
2534    
2535    ----------------------------------------------------------------------
2536    Name: Matthias Blume
2537    Date: 2002/01/21
2538    Tag: blume-20020121-ff
2539    Description:
2540    
2541    Another day of fiddling with the FFI...
2542    
2543    1. Bug fix/workaround:  CKIT does not complain about negative array
2544       dimensions, so ml-nlffigen has to guard itself against this possibility.
2545       (Otherwise a negative dimension would send it into an infinite loop.)
2546    
2547    2. Some of the abstract types (light objects, light pointers, most "base"
2548       types) in structure C are now eqtypes.
2549    
2550    3. Added constructors and test functions for NULL function pointers.
2551    
2552    ----------------------------------------------------------------------
2553    Name: Matthias Blume
2554    Date: 2002/01/18
2555    Tag: blume-20020118-ready-for-new-release
2556    Description:
2557    
2558    Made config/srcarchiveurl point to a new place.  (Will provide boot
2559    files shortly.)
2560    
2561    Maybe we christen this to be 110.38?
2562    
2563    ----------------------------------------------------------------------
2564    Name: Matthias Blume
2565    Date: 2002/01/18
2566    Tag: blume-20020118-more-ffifiddle
2567    Description:
2568    
2569    Today's FFI fiddling:
2570    
2571      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
2572        C.Set.* that use concrete (MLRep.*) arguments and results instead
2573        of abstract ones.
2574    
2575      - Provided word-style bit operations etc. for "int" representation
2576        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
2577        and Long.
2578    
2579    ----------------------------------------------------------------------
2580    Name: Matthias Blume
2581    Date: 2002/01/18
2582    Tag: blume-20020118-use-x86-fp
2583    Description:
2584    
2585    Now that x86-fast-fp seems to be working, I turned it back on again
2586    by default.  (Seems to work fine now, even with the FFI.)
2587    
2588    Other than that, I added some documentation about the FFI to
2589    src/ml-nlffigen/README and updated the FFI test examples in
2590    src/ml-nlffi-lib/Tests/*.
2591    
2592    ----------------------------------------------------------------------
2593    Name: Allen Leung
2594    Date: 2002/01/17
2595    Tag: leunga-20020117-x86-fast-fp-call
2596    Description:
2597    
2598       1. Fixed a problem with handling return fp values when x86's fast fp
2599          mode is turned on.
2600    
2601       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
2602          of %f32.
2603    
2604       3. Added a constructor INT32lit to the ast of MLRISC tools.
2605    
2606    ----------------------------------------------------------------------
2607    Name: Matthias Blume
2608    Date: 2002/01/16
2609    Tag: blume-20020116-ffifiddle
2610    Description:
2611    
2612    More fiddling with the FFI interface:
2613    
2614     - Make constness 'c instead of rw wherever possible.  This eliminates
2615       the need for certain explicit coercions.  (However, due to ML's
2616       value polymorphism, there will still be many cases where explicit
2617       coercions are necessary.  Phantom types are not the whole answer
2618       to modeling a subtyping relationship in ML.)
2619    
2620     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
2621    
2622     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
2623       clumsy workaround for varargs problem.)
2624    
2625    ----------------------------------------------------------------------
2626    Name: Lal George
2627    Date: 2002/01/15
2628    Tag: <none>
2629    Description:
2630    
2631    1. Since COPY instructions are no longer native to the architecture,
2632       a generic functor can be used to implement the expandCopies function.
2633    
2634    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
2635       TEXT segment.
2636    
2637    ----------------------------------------------------------------------
2638    Name: Matthias Blume
2639    Date: 2002/01/15
2640    Tag: blume-20020115-ffiupdates
2641    Description:
2642    
2643    1. Fix for bug resulting in single-precision float values being returned
2644       incorrectly from FFI calls.
2645    
2646    2. Small modifications to C FFI API:
2647    
2648        - memory-allocation routines return straight objects (no options)
2649          and raise an exception in out-of-memory situations
2650        - unsafe extensions to cast between function pointers and pointers
2651          from/to ints
2652        - added structure C_Debug as an alternative to structure C where
2653          pointer-dereferencing (|*| and |*!) always check for null-pointers
2654        - added open_lib' to DynLinkage;  open_lib' works like open_lib
2655          but also takes a (possibly empty) list of existing library handles
2656          that the current library depends on
2657    
2658    ----------------------------------------------------------------------
2659    Name: Matthias Blume
2660    Date: 2002/01/10
2661    Tag: blume-20020110-newffigen
2662    Description:
2663    
2664    1. Updates to portable graph code.
2665    
2666    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
2667       more scalable now so that even huge interfaces such as the one
2668       for GTK compile in finite time and space. :-)
2669       See src/ml-nlffigen/README for details on what's new.
2670    
2671    ----------------------------------------------------------------------
2672    Name: Lal George
2673    Date: 2001/01/09 14:31:35 EST 2002
2674    Tag: george-20011206-rm-native-copy
2675    Description:
2676    
2677            Removed the native COPY and FCOPY instructions
2678            from all the architectures and replaced it with the
2679            explicit COPY instruction from the previous commit.
2680    
2681            It is now possible to simplify many of the optimizations
2682            modules that manipulate copies. This has not been
2683            done in this change.
2684    
2685    ----------------------------------------------------------------------
2686    Name: Lal George
2687    Date: 2001/12/06 16:50:13 EST 2001
2688    Tag: george-20011206-mlrisc-instruction
2689    Description:
2690    
2691    Changed the representation of instructions from being fully abstract
2692    to being partially concrete. That is to say:
2693    
2694      from
2695            type instruction
2696    
2697      to
2698            type instr                              (* machine instruction *)
2699    
2700            datatype instruction =
2701                LIVE of {regs: C.cellset, spilled: C.cellset}
2702              | KILL of {regs: C.cellset, spilled: C.cellset}
2703              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
2704              | ANNOTATION of {i: instruction, a: Annotations.annotation}
2705              | INSTR of instr
2706    
2707    This makes the handling of certain special instructions that appear on
2708    all architectures easier and uniform.
2709    
2710    LIVE and KILL say that a list of registers are live or killed at the
2711    program point where they appear. No spill code is generated when an
2712    element of the 'regs' field is spilled, but the register is moved to
2713    the 'spilled' (which is present, more for debugging than anything else).
2714    
2715    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
2716    We used to generate:
2717    
2718            DEFFREG f1
2719            f1 := f2 + f3
2720            trapb
2721    
2722    but now generate:
2723    
2724            f1 := f2 + f3
2725            trapb
2726            LIVE {regs=[f1,f2,f3], spilled=[]}
2727    
2728    Furthermore, the DEFFREG (hack) required that all floating point instruction
2729    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
2730    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
2731    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
2732    this architecture).
2733    
2734    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
2735    all the architectures. This will result in further simplification of the
2736    register allocator that must be aware of them for coalescing purposes, and
2737    will also simplify certain aspects of the machine description that provides
2738    callbacks related to parallel copies.
2739    
2740    ANNOTATION should be obvious, and now INSTR represents the honest to God
2741    machine instruction set!
2742    
2743    The <arch>/instructions/<arch>Instr.sml files define certain utility
2744    functions for making porting easier -- essentially converting upper case
2745    to lower case. All machine instructions (of type instr) are in upper case,
2746    and the lower case form generates an MLRISC instruction. For example on
2747    the alpha we have:
2748    
2749      datatype instr =
2750         LDA of {r:cell, b:cell, d:operand}
2751       | ...
2752    
2753      val lda : {r:cell, b:cell, d:operand} -> instruction
2754        ...
2755    
2756    where lda is just (INSTR o LDA), etc.
2757    
2758    ----------------------------------------------------------------------
2759    Name: Matthias Blume
2760    Date: 2001/11/22 21:40:00 EST
2761    Tag: Release_110_37
2762    Description:
2763    
2764    Release 110.37.  This time for real.
2765    
2766    ----------------------------------------------------------------------
2767    Name: Matthias Blume
2768    Date: 2001/11/21 16:35:00 EST
2769    Tag: blume-20011121-foot-in-mouth
2770    Description:
2771    
2772    Removed the "Release_110_37" tag because of a serious bug.
2773    This will be re-tagged once the bug is fixed.
2774    
2775    ----------------------------------------------------------------------
2776    Name: Matthias Blume
2777    Date: 2001/11/21 16:14:00 EST
2778    Tag: blume-20011121-forgottenfile
2779    Description:
2780    
2781    Forgot to add a file.  (Just a .tex-file -- part of
2782    the CM manual source.)
2783    
2784    ----------------------------------------------------------------------
2785    Name: Matthias Blume
2786    Date: 2001/11/21 16:10:00 EST
2787    Tag: blume-20011121-invalid_110_37
2788    Description:
2789    
2790    Note: I removed the original tag "Release_110_37" from this commit
2791          because we found a serious bug in all non-x86 backends.
2792          - Matthias
2793    
2794    1. Modifications to the SML/NJ code generator and to the runtime system
2795       so that code object name strings are directly inserted into code
2796       objects at code generation time.  The only business the runtime system
2797       has with this is now to read the name strings on occasions.
2798       (The encoding of the name string has also changed somewhat.)
2799    
2800    2. CM now implements a simple "set calculus" for specifying export lists.
2801       In particular, it is now possible to refer to the export lists of
2802       other libraries/groups/sources and form unions as well as differences.
2803       See the latest CM manual for details.
2804    
2805    3. An separate notion of "proxy" libraries has again be eliminated from
2806       CM's model.  (Proxy libraries are now simply a special case of using
2807       the export list calculus.)
2808    
2809    4. Some of the existing libraries now take advantage of the new set
2810       calculus.
2811       (Notice that not all libraries have been converted because some
2812       of the existing .cm-files are supposed to be backward compatible
2813       with 110.0.x.)
2814    
2815    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
2816       "General.exnMessage"!  The former relies on a certain hook to be
2817       initialized, and that often does not happen in the stand-alone case.)
2818    
2819    ----------------------------------------------------------------------
2820    Name: Lal George
2821    Date: 2001/11/21  13:56:18 EST
2822    Tag: george-2001121-pseudo-ops
2823    Description:
2824    
2825      Implemented a complete redesign of MLRISC pseudo-ops. Now there
2826      ought to never be any question of incompatabilities with
2827      pseudo-op syntax expected by host assemblers.
2828    
2829      For now, only modules supporting GAS syntax are implemented
2830      but more should follow, such as MASM, and vendor assembler
2831      syntax, e.g. IBM as, Sun as, etc.
2832    
2833    ----------------------------------------------------------------------
2834    Name: Matthias Blume
2835    Date: 2001/11/14 11:52:00 EST
2836    Tag: blume-20011114-srcname
2837    Description:
2838    
2839    1. Routed the name of the current source file to mlriscgen where it
2840       should be directly emitted into the code object.  (This last part
2841       is yet to be done.)
2842    
2843    2. Some cleanup of the pgraph code to make it match the proposal that
2844       I put out the other day.  (The proposal notwithstanding, things are
2845       still in flux here.)
2846    
2847    ----------------------------------------------------------------------
2848    Name: Lal George
2849    Date: 2001/11/14 09:44:04 EST
2850    Tag:
2851    Description:
2852    
2853      Fix for a backpatching bug reported by Allen.
2854    
2855      Because the boundary between short and long span-dependent
2856      instructions is +/- 128, there are an astounding number of
2857      span-dependent instructions whose size is over estimated.
2858    
2859      Allen came up with the idea of letting the size of span
2860      dependent instructions be non-monotonic, for a maxIter
2861      number of times, after which the size must be monotonically
2862      increasing.
2863    
2864      This table shows the number of span-dependent instructions
2865      whose size was over-estimated as a function of maxIter, for the
2866      file Parse/parse/ml.grm.sml:
2867    
2868         maxIter            # of instructions:
2869            10                      687
2870            20                      438
2871            30                      198
2872            40                        0
2873    
2874      In compiling the compiler, there is no significant difference in
2875      compilation speed between maxIter=10 and maxIter=40. Actually,
2876      my measurements showed that maxIter=40 was a tad faster than
2877      maxIter=10! Also 96% of the  files in the compiler reach a fix
2878      point within 13 iterations, so fixing maxIter at 40, while high,
2879      is okay.
2880    
2881    ----------------------------------------------------------------------
2882    Name: Matthias Blume
2883    Date: 2001/10/31 15:25:00 EST
2884    Tag: blume-20011031-pgraph
2885    Description:
2886    
2887    CKIT:
2888    * Changed the "Function" constructor of type Ast.ctype to carry optional
2889      argument identifiers.
2890    * Changed the return type of TypeUtil.getFunction accordingly.
2891    * Type equality ignores the argument names.
2892    * TypeUtil.composite tries to preserve argument names but gives up quickly
2893      if there is a mismatch.
2894    
2895    installation script:
2896    * attempts to use "curl" if available (unless "wget" is available as well)
2897    
2898    CM:
2899    * has an experimental implementation of "portable graphs" which I will
2900      soon propose as an implementation-independent library format
2901    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
2902    
2903    NLFFI-LIB:
2904    * some cleanup (all cosmetic)
2905    
2906    NLFFIGEN:
2907    * temporarily disabled the mechanism that suppresses ML output for
2908      C definitions whose identifiers start with an underscore character
2909    * generate val bindings for enum constants
2910    * user can request that only one style (light or heavy) is being used;
2911      default is to use both (command-line arguments: -heavy and -light)
2912    * fixed bug in handling of function types involving incomplete pointers
2913    * generate ML entry points that take record arguments (i.e., using
2914      named arguments) for C functions that have a prototype with named
2915      arguments
2916      (see changes to CKIT)
2917    
2918    ----------------------------------------------------------------------
2919    Name: Allen Leung
2920    Date: 2001/10/27 20:34:00 EDT
2921    Tag: leunga-20011027-x86-fast-fp-call
2922    Description:
2923    
2924       Fixed the bug described in blume-20010920-slowfp.
2925    
2926       The fix involves
2927          1. generating FCOPYs in FSTP in ia32-svid
2928          2. marking a CALL with the appropriate annotation
2929    
2930    ----------------------------------------------------------------------
2931    Name: Matthias Blume
2932    Date: 2001/10/16 11:32:00 EDT
2933    Tag: blume-20011016-netbsd
2934    Description:
2935    
2936    Underscore patch from Chris Richards (fixing problem with compiling
2937    runtime system under recent NetBSD).
2938    
2939    ----------------------------------------------------------------------
2940    Name: Allen Leung
2941    Date: 2001/10/12 17:18:32 EDT 2001
2942    Tag: leung-20011012-x86-printflowgraph
2943    Description:
2944    
2945    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
2946    
2947    ----------------------------------------------------------------------
2948    Name: Lal George
2949    Date: 2001/10/11 23:51:34 EDT
2950    Tag: george-20011011-too-many-instrs
2951    Description:
2952    
2953    The representation of a program point never expected to see more
2954    than 65536 instructions in a basic block!
2955    
2956    ----------------------------------------------------------------------
2957    Name: Lal George
2958    Date: 2001/10/09 09:41:37 EDT
2959    Tag: george-20011008-mlrisc-labels
2960    Description:
2961    
2962    Changed the machine description files to support printing of
2963    local and global labels in assembly code, based on host assembler
2964    conventions.
2965    
2966    ----------------------------------------------------------------------
2967    Name: Matthias Blume
2968    Date: 2001/09/25 15:25:00 EDT
2969    Tag: blume-20010925-exninfo
2970    Description:
2971    
2972    I provided a non-hook implementation of exnName (at the toplevel) and
2973    made the "dummy" implementation of exnMessage (at the toplevel) more
2974    useful: if nothing gets "hooked in", then at least you are going to
2975    see the exception name and a message indicating why you don't see more.
2976    
2977    [For the time being, programs that need exnMessage and want to use
2978    ml-build should either use General.exnMessage (strongly recommended) or
2979    refer to structure General at some other point so that CM sees a
2980    static dependency.]
2981    
2982    [Similar remarks go for "print" and "use":  If you want to use their
2983    functionality in stand-alone programs generated by ml-build, then use
2984    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
2985    
2986    ----------------------------------------------------------------------
2987    Name: Matthias Blume
2988    Date: 2001/09/20 17:28:00 EDT
2989    Tag: blume-20010920-slowfp
2990    Description:
2991    
2992    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
2993    
2994    ----------------------------------------------------------------------
2995    Name: Matthias Blume
2996    Date: 2001/09/20 17:20:00 EDT
2997    Tag: blume-20010920-canonicalpaths
2998    Description:
2999    
3000    0. Updated the BOOT file (something that I forgot to do earlier).
3001    
3002    1. Small internal change to CM so that it avoids "/../" in filenames
3003       as much as possible (but only where it is safe).
3004    
3005    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
3006       that arguments that contain delimiters are passed through correctly.
3007       This change also means that all "special" arguments of the form
3008       @SMLxxx... must come first.
3009    
3010    3. Changed install script to put relative anchor names for tool commands
3011       into pathconfig.
3012    
3013    ----------------------------------------------------------------------
3014    Name: Matthias Blume
3015    Date: 2001/09/18 15:35:00 EDT
3016    Tag: blume-20010918-readme11036
3017    Description:
3018    
3019    Added README files.
3020    
3021    ----------------------------------------------------------------------
3022    Name: Matthias Blume
3023    Date: 2001/09/18 11:45:00 EDT
3024    Tag: Release_110_36 (retag)
3025    Description:
3026    
3027    Fixed mistake in config/preloads. Retagged as 110.36.
3028    
3029    ----------------------------------------------------------------------
3030    Name: Matthias Blume
3031    Date: 2001/09/18 09:40:00 EDT
3032    Tag: Release_110_36_orig (tag changed)
3033    Description:
3034    
3035    New version (110.36).  New bootfiles.
3036    
3037    ----------------------------------------------------------------------
3038    Name: Matthias Blume
3039    Date: 2001/09/14 16:15:00 EDT
3040    Tag: blume-20010914-x86fastfp
3041    Description:
3042    
3043    John committed some changes that Allen made, in particular a (hopefully)
3044    correctly working version of the x86-fp module.
3045    
3046    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
3047    flag to "true".  Everything seems to compile to a fixpoint ok, and
3048    "mandelbrot" speeds up by about 15%.
3049    
3050    ----------------------------------------------------------------------
3051    Name: Matthias Blume
3052    Date: 2001/09/13 11:20:00 EDT
3053    Tag: blume-20010913-minimal
3054    Description:
3055    
3056    1. Stefan Monnier's patch to fix a miscompilation problem that
3057       was brought to light by John Reppy's work on Moby.
3058    
3059    2. Implemented a minimal "structure Compiler" that contains just
3060       "version" and "architecture".  The minimal version will be
3061       available when the full version is not.  This is for backward-
3062       compatibility with code that wants to test Compiler.version.
3063    
3064    ----------------------------------------------------------------------
3065    Name: Matthias Blume
3066    Date: 2001/08/28 14:03:00 EDT
3067    Tag: blume-20010828-ml-lex
3068    Description:
3069    
3070    Fix for bug 1581, received from Neophytos Michael.
3071    
3072    ----------------------------------------------------------------------
3073    Name: Matthias Blume
3074    Date: 2001/08/27 11:20:00 EDT
3075    Tag: blume-20010827-readme11035
3076    Description:
3077    
3078    Fleshed out the README file for 110.35.
3079    
3080    ----------------------------------------------------------------------
3081    Name: Matthias Blume
3082    Date: 2001/08/24 17:10:00 EDT
3083    Tag: Release_110_35
3084    Description:
3085    
3086    New version number (110.35).  New bootfiles.
3087    
3088    ----------------------------------------------------------------------
3089    Name: Lal George
3090    Date: 2001/08/24 13:47:18 EDT 2001
3091    Tag: george-20010824-MLRISC-graphs
3092    Description:
3093    
3094     removed clusters from MLRISC completely and replaced with graphs.
3095    
3096    ----------------------------------------------------------------------
3097    Name: Matthias Blume
3098    Date: 2001/08/23 17:50:00 EDT
3099    Tag: blume-20010823-toplevel
3100    Description:
3101    
3102    - some reorganization of the code that implements various kinds of
3103      environments in the compiler (static, dynamic, symbolic, combined)
3104    - re-implemented the EnvRef module so that evalStream works properly
3105      (if the stream contains references to "use", "CM.make", etc.)
3106    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
3107    
3108    ----------------------------------------------------------------------
3109    Name: Matthias Blume
3110    Date: 2001/08/20 15:50 EDT
3111    Tag: blume20010820-slipup
3112    Description:
3113    
3114    I forgot to commit a few files.  Here they are...
3115    
3116    ----------------------------------------------------------------------
3117    Name: Matthias Blume
3118    Date: 2001/08/20 15:35:00 EDT
3119    Tag: blume-20010820-debugprof
3120    Description:
3121    
3122    !!!! NEW BOOTFILES !!!!
3123    
3124    This is another round of reorganizing the compiler sources.  This
3125    time the main goal was to factor out all the "instrumentation"
3126    passes (for profiling and backtracing) into their own library.
3127    The difficulty was to do it in such a way that it does not depend
3128    on elaborate.cm but only on elabdata.cm.
3129    
3130    Therefore there have been further changes to both elaborate.cm and
3131    elabdata.cm -- more "generic" things have been moved from the former
3132    to the latter.  As a result, I was forced to split the assignment
3133    of numbers indicating "primtyc"s into two portions: SML-generic and
3134    SML/NJ-specific.  Since it would have been awkward to maintain,
3135    I bit the bullet and actually _changed_ the mapping between these
3136    numbers and primtycs.  The bottom line of this is that you need
3137    a new set of bin- and bootfiles.
3138    
3139    I have built new bootfiles for all architectures, so doing a fresh
3140    checkout and config/install.sh should be all you need.
3141    
3142    The newly created library's name is
3143    
3144        $smlnj/viscomp/debugprof.cm
3145    
3146    and its sources live under
3147    
3148        src/compiler/DebugProf
3149    
3150    ----------------------------------------------------------------------
3151    Name: Matthias Blume
3152    Date: 2001/08/15 17:15:00 EDT
3153    Tag: blume-20010815-compreorg
3154    Description:
3155    
3156    This is a first cut at reorganizing the CM libraries that make up the
3157    core of the compiler.  The idea is to separate out pieces that could
3158    be used independently by tools, e.g., the parser, the typechecker, etc.
3159    
3160    The current status is a step in this direction, but it is not quite
3161    satisfactory yet.  Expect more changes in the future.
3162    
3163    Here is the current (new) organization...
3164    
3165        What used to be $smlnj/viscomp/core.cm is now divided into
3166        six CM libraries:
3167    
3168             $smlnj/viscomp/basics.cm
3169                           /parser.cm
3170                           /elabdata.cm
3171                           /elaborate.cm
3172                           /execute.cm
3173                           /core.cm
3174    
3175        The CM files for these libraries live under src/system/smlnj/viscomp.
3176        All these libraries are proxy libraries that contain precisely
3177        one CM library component.  Here are the locations of the components
3178        (all within the src/compiler tree):
3179    
3180             Basics/basics.cm
3181             Parse/parser.cm
3182             ElabData/elabdata.cm
3183             Elaborator/elaborate.cm
3184             Execution/execute.cm
3185             core.cm
3186    
3187         [This organization is the same that has been used already
3188         for a while for the architecture-specific parts of the visible
3189         compiler and for the old version of core.cm.]
3190    
3191         As you will notice, many source files have been moved from their
3192         respective original locations to a new home in one of the above
3193         subtrees.
3194    
3195         The division of labor between the new libraries is the following:
3196    
3197             basics.cm:
3198                - Simple, basic definitions that pertain to many (or all) of
3199                  the other libraries.
3200             parser.cm:
3201                - The SML parser, producing output of type Ast.dec.
3202                - The type family for Ast is also defined and exported here.
3203             elabdata.cm:
3204                - The datatypes that describe input and output of the elaborator.
3205                  This includes types, absyn, and static environments.
3206             elaborator.cm:
3207                - The SML/NJ type checker and elaborator.
3208                  This maps an Ast.dec (with a given static environment) to
3209                  an Absyn.dec (with a new static environment).
3210                - This libraries implements certain modules that used to be
3211                  structures as functors (to remove dependencies on FLINT).
3212             execute.cm:
3213                - Everything having to do with executing binary code objects.
3214                - Dynamic environments.
3215             core.cm:
3216                - SML/NJ-specific instantiations of the elaborator and MLRISC.
3217                - Top-level modules.
3218                - FLINT (this should eventually become its own library)
3219    
3220    Notes:
3221    
3222    I am not 100% happy with the way I separated the elaborator (and its
3223    data structures) from FLINT.  Two instances of the same problem:
3224    
3225        1. Data structures contain certain fields that carry FLINT-specific
3226           information.  I hacked around this using exn and the property list
3227           module from smlnj-lib.  But the fact that there are middle-end
3228           specific fields around at all is a bit annoying.
3229    
3230        2. The elaborator calculates certain FLINT-related information.  I tried
3231           to make this as abstract as I could using functorization, but, again,
3232           the fact that the elaborator has to perform calculations on behalf
3233           of the middle-end at all is not nice.
3234    
3235        3. Having to used exn and property lists is unfortunate because it
3236           weakens type checking.  The other alternative (parameterizing
3237           nearly *everything*) is not appealing, though.
3238    
3239    I removed the "rebinding =" warning hack because due to the new organization
3240    it was awkward to maintain it.  As a result, the compiler now issues some of
3241    these warnings when compiling init.cmi during bootstrap compilation. On
3242    the plus side, you also get a warning when you do, for example:
3243       val op = = Int32.+
3244    which was not the case up to now.
3245    
3246    I placed "assign" and "deref" into the _Core structure so that the
3247    code that deals with the "lazy" keyword can find them there.  This
3248    removes the need for having access to the primitive environment
3249    during elaboration.
3250    
3251    ----------------------------------------------------------------------
3252    Name: Matthias Blume
3253    Date: 2001/08/13
3254    Tag: blume-20010813-closures
3255    Description:
3256    
3257    This fix was sent to us by Zhong Shao.  It is supposed to improve the
3258    performance of certain loops by avoiding needless closure allocation.
3259    
3260    ----------------------------------------------------------------------
3261    Name: Lal George
3262    Date: 2001/07/31 10:03:23 EDT 2001
3263    Tag: george-20010731-x86-fmalloc
3264    Description: Fixed bug in x86 calls
3265    
3266        There was a bug where call instructions would mysteriously
3267        vanish. The call instruction had to be one that returned
3268        a floating point value.
3269    
3270    ----------------------------------------------------------------------
3271    Name: Lal George
3272    Date: 2001/07/19 16:36:29 EDT 2001
3273    Tag: george-20010719-simple-cells
3274    Description:
3275    
3276    I have dramatically simplified the interface for CELLS in MLRISC.
3277    
3278    In summary, the cells interface is broken up into three parts:
3279    
3280      1. CellsBasis : CELLS_BASIS
3281    
3282            CellsBasis is a top level structure and common for all
3283            architectures.  it contains the definitions of basic datatypes
3284            and utility  functions over these types.
3285    
3286      2. functor Cells() : CELLS
3287    
3288            Cells generates an interface for CELLS that incorporates the
3289            specific resources on the target architecture, such as the
3290            presence of special register classes, their number and size,
3291            and various useful substructures.
3292    
3293      3. <ARCH>CELLS
3294    
3295            e.g. SparcCells: SPARCCELLS
3296    
3297            <ARCH>CELLS usually contains additional bindings for special
3298            registers  on the architecture, such as:
3299    
3300                    val r0 : cell           (* register zero *)
3301                    val y : cell            (* Y register *)
3302                    val psr : cell          (* processor status register *)
3303                    ...
3304    
3305            The structure returned by applying the Cells functor is opened
3306            in this interface.
3307    
3308    The main implication of all this is that the datatypes for cells is
3309    split between CellsBasis and CELLS -- a fairly simple change for user
3310    code.
3311    
3312    In the old scheme the CELLS interface had a definitional binding of
3313    the form:
3314    
3315            signature CELLS = sig
3316    
3317               structure CellsBasis = CellsBasis
3318    
3319               ...
3320    
3321            end
3322    
3323    With all the sharing constraints that goes on in MLRISC, this old
3324    design  quickly leads to errors such as:
3325    
3326            "structure definition spec inside of sharing ... "
3327    
3328    
3329    and appears to require an unacceptable amount of sharing and where
3330    constraint hackery.
3331    
3332    I think this error message (the interaction of definitional specs and
3333    sharing) requires more explanation on our web page.
3334    
3335    ----------------------------------------------------------------------
3336    Name: Matthias Blume
3337    Date: 2001/07/19 15:00:00 EDT
3338    Tag: blume-20010719-libreorg
3339    Description:
3340    
3341    This update puts together a fairly extensive but straightforward change
3342    to the way the libraries that implement the interactive system are
3343    organized:
3344    
3345       The biggest change is the elimination of structure Compiler.  As a
3346       replacement for this structure, there is now a CM library
3347       (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
3348       that exports all the substructures of the original structure Compiler
3349       directly.  So instead of saying Compiler.Foo.bar one now simply
3350       says Foo.bar.  (The CM libraries actually export a collection of
3351       structures that is richer than the collection of substructures of
3352       structure Compiler.)
3353    
3354       To make the transition smooth, there is a separate library called
3355       $smlnj/compiler/compiler.cm which puts together and exports the
3356       original structure Compiler (or at least something very close to it).
3357    
3358       There are five members of the original structure Compiler
3359       that are not exported directly but which instead became members
3360       of a new structure Backend (described by signature BACKEND).  These are:
3361       structure Profile (: PROFILE), structure Compile (: COMPILE), structure
3362       Interact (: INTERACT), structure Machine (: MACHINE), and val
3363       architecture (: string).
3364    
3365       Structure Compiler.Version has become structure CompilerVersion.
3366    
3367       Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
3368       by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
3369       or x86, respectively.
3370       Each of these exports the same frontend structures that
3371       $smlnj/compiler.cm exports.  But they do not have a structure Backend
3372       and instead export some structure <Arch>Backend where <Arch> is Alpha32,
3373       Hppa, PPC, Sparc, or X86, respectively.
3374    
3375       Library $smlnj/compiler/all.cm exports the union of the exports of
3376       $smlnj/compiler/<arch>.cm
3377    
3378       There are no structures <Arch>Compiler anymore, use
3379       $smlnj/compiler/<arch>.cm instead.
3380    
3381       Library host-compiler-0.cm is gone.  Instead, the internal library
3382       that instantiates CM is now called cm0.cm.  Selection of the host
3383       compiler (backend) is no longer done here but. (Responsibility for it
3384       now lies with $smlnj/compiler/current.cm.  This seems to be more
3385       logical.)
3386    
3387       Many individual files have been moved or renamed.  Some files have
3388       been split into multiple files, and some "dead" files have been deleted.
3389    
3390    Aside from these changes to library organization, there are also changes
3391    to the way the code itself is organized:
3392    
3393       Structure Binfile has been re-implemented in such a way that it no
3394       longer needs any knowledge of the compiler.  It exclusively deals
3395       with the details of binfile layout.  It no longer invokes the
3396       compiler (for the purpose of creating new prospective binfile
3397       content), and it no longer has any knowledge of how to interpret
3398       pickles.
3399    
3400       Structure Compile (: COMPILE) has been stripped down to the bare
3401       essentials of compilation.  It no longer deals with linking/execution.
3402       The interface has been cleaned up considerably.
3403    
3404       Utility routines for dealing with linking and execution have been
3405       moved into their own substructures.
3406    
3407       (The ultimate goal of these changes is to provide a light-weight
3408       binfile loader/linker (at least for, e.g., stable libraries) that
3409       does not require CM or the compiler to be present.)
3410    
3411    CM documentation has been updated to reflect the changes to library
3412    organization.
3413    
3414    ----------------------------------------------------------------------
3415    Name: Matthias Blume
3416    Date: 2001/07/10 17:30:00 EDT
3417    Tag: Release_110_34
3418    Description:
3419    
3420    Minor tweak to 110.34 (re-tagged):
3421    
3422      - README.html file added to CVS repository
3423      - runtime compiles properly under FreeBSD 3.X and 4.X
3424    
3425    ----------------------------------------------------------------------
3426    Name: Matthias Blume
3427    Date: 2001/07/10 17:30:00 EDT
3428    Tag: Release_110_34
3429    Description:
3430    
3431    New version number (110.34). New bootfiles.
3432    
3433    ----------------------------------------------------------------------
3434    Name: Matthias Blume
3435    Date: 2001/07/09 16:00:00 EDT
3436    Tag: blume-20010709-more-varargs
3437    Description:
3438    
3439    I changed the handling of varargs in ml-nlffigen again:
3440    The ellipsis ... will now simply be ignored (with an accompanying warning).
3441    
3442    The immediate effect is that you can actually call a varargs function
3443    from ML -- but you can't actually supply any arguments beyond the ones
3444    specified explicitly.  (For example, you can call printf with its format
3445    string, but you cannot pass additional arguments.)
3446    
3447    This behavior is only marginally more useful than the one before, but
3448    it has the advantage that a function or, more importantly, a function
3449    type never gets dropped on the floor, thus avoiding follow-up problems with
3450    other types that refer to the offending one.
3451    
3452    ----------------------------------------------------------------------
3453    Name: Matthias Blume
3454    Date: 2001/07/09 11:25:00 EDT
3455    Tag: blume-20010709-varargs
3456    Description:
3457    
3458    1. ckit-lib.cm now exports structure Error
3459    2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
3460       with a warning accompanied by a source location.  Moreover, it
3461       merely skips the offending function or type and proceeds with the
3462       rest of its work.u  As a result, one can safely feed C code containing
3463       "..." to ml-nlffigen.
3464    3. There are some internal improvements to CM, providing slightly
3465       more general string substitutions in the tools subsystem.
3466    
3467    ----------------------------------------------------------------------
3468    Name: Matthias Blume
3469    Date: 2001/06/27 15:10:00 EDT
3470    Tag: blume-20010627-concur
3471    Description:
3472    
3473    Fixed a small bug in CM's handling of parallel compilation.
3474    (You could observe the bug by Control-C-interrupting an ordinary
3475    CMB.make or CM.stabilize and then attaching some compile servers.
3476    The result was that all of a sudden the previously interrupted
3477    compilation would continue on its own.  This was because of
3478    an over-optimization: CM did not bother to clean out certain queues
3479    when no servers were attached "anyway", resulting in the contents
3480    of these queues to grab control when new servers did get attached.)
3481    
3482    There is also another minor update to the CM manual.
3483    
3484    ----------------------------------------------------------------------
3485    Name: Matthias Blume
3486    Date: 2001/06/26 16:15:00 EDT
3487    Tag: blume-20010626-cmdoc
3488    Description:
3489    
3490    Minor typo fixed in CM manual (syntax diagram for libraries).
3491    
3492    ----------------------------------------------------------------------
3493    Name: Matthias Blume
3494    Date: 2001/06/25 22:55:00 EDT
3495    Tag: blume-20010625-x86pc
3496    Description:
3497    
3498    Fixed a nasty bug in the X86 assembly code that caused signal
3499    handlers to fail (crash) randomly.
3500    
3501    ----------------------------------------------------------------------
3502    Name: Matthias Blume
3503    Date: 2001/06/25 12:05:00 EDT
3504    Tag: blume-20010625-nlffigen
3505    Description:
3506    
3507    This update fixes a number of minor bugs in ml-nlffigen as reported by
3508    Nick Carter <nbc@andrew.cmu.edu>.
3509    
3510      1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
3511      2. Default names for generated files are now derived from the name of
3512         the C file *without its directory*.  In particular, this causes generated
3513         files to be placed locally even if the C file is in some system directory.
3514      3. Default names for generated signatures and structures are also derived
3515         from the C file name without its directory.  This avoids silly things
3516         like "structure GL/GL".
3517         (Other silly names are still possible because ml-nlffigen does not do
3518          a thorough check of whether generated names are legal ML identifiers.
3519          When in doubt, use command line arguments to force particular names.)
3520    
3521    ----------------------------------------------------------------------
3522    Name: Matthias Blume
3523    Date: 2001/06/21 12:25:00 EDT
3524    Tag: blume-20010621-eXene
3525    Description:
3526    
3527    eXene now compiles and (sort of) works again.
3528    
3529    The library name (for version > 110.33) is $/eXene.cm.
3530    
3531    I also added an new example in src/eXene/examples/nbody.  See the
3532    README file there for details.
3533    
3534    ----------------------------------------------------------------------
3535    Name: Matthias Blume
3536    Date: 2001/06/20 16:40:00 EDT
3537    Tag: blume-20010620-cml
3538    Description:
3539    
3540    CML now compiles and works again.
3541    
3542    Libraries (for version > 110.33):
3543    
3544      $cml/cml.cm            Main CML library.
3545      $cml/basis.cm          CML's version of $/basis.cm.
3546      $cml/cml-internal.cm   Internal helper library.
3547      $cml/core-cml.cm       Internal helper library.
3548      $cml-lib/trace-cml.cm  Tracing facility.
3549      $cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
3550    
3551    The installer (config/install.sh) has been taught how to properly
3552    install this stuff.
3553    
3554    ----------------------------------------------------------------------
3555    Name: Matthias Blume
3556    Date: 2001/06/19 17:55:00 EDT
3557    Tag: blume-20010619-instantiate
3558    Description:
3559    
3560    This un-breaks the fix for bug 1432.
3561    (The bug was originally fixed in 110.9 but I broke it again some
3562    time after that.)
3563    
3564    ----------------------------------------------------------------------
3565    Name: Matthias Blume
3566    Date: 2001/06/19 17:25:00 EDT
3567    Tag: blume-20010619-signals
3568    Description:
3569    
3570    This should (hopefully) fix the long-standing signal handling bug.
3571    (The runtime system was constructing a continuation record with an
3572    incorrect descriptor which would cause the GC to drop data on the floor...)
3573    
3574    ----------------------------------------------------------------------
3575    Name: Matthias Blume
3576    Date: 2001/06/15 15:05:00 EDT
3577    Tag: blume-20010615-moresparc
3578    Description:
3579    
3580    Here is a short late-hour update related to Sparc c-calls:
3581    
3582     -- made handling of double-word arguments a bit smarter
3583    
3584     -- instruction selection phase tries to collapse certain clumsily
3585        constructed ML-Trees; typical example:
3586    
3587            ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
3588    
3589        This currently has no further impact on SML/NJ since mlriscGen does
3590        not seem to generate such patterns in the first place, and c-calls
3591        (which did generate them in the beginning) has meanwhile been fixed
3592        so as to avoid them as well.
3593    
3594    ----------------------------------------------------------------------
3595    Name: Matthias Blume
3596    Date: 2001/06/15 15:05:00 EDT
3597    Tag: blume-20010615-sparc
3598    Description:
3599    
3600    The purpose of this update is to provide an implementation of NLFFI
3601    on Sparc machines.
3602    
3603    Here are the changes in detail:
3604    
3605       * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
3606       the Sparc implementation of the c-calls API.
3607       * The Sparc backend of SML/NJ has been modified to uniformely use %fp
3608       for accessing the ML frame.  Thus, we have a real frame pointer and
3609       can freely modify %sp without need for an omit-frame-ptr phase.
3610       The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
3611       this case.
3612       * ml-nlffigen has been taught to produce code for different architectures
3613       and calling conventions.
3614       * In a way similar to what was done in the x86 case, the Sparc
3615       backend uses its own specific extension to mltree.  (For example,
3616       it needs to be able to generate UNIMP instructions which are part
3617       of the calling convention.)
3618       * ml-nlffi-lib was reorganized to make it more modular (in particular,
3619       to make it easier to plug in new machine- and os-dependent parts).
3620    
3621    There are some other fairly unrelated bug fixes and cleanups as well:
3622    
3623       * I further hacked the .cm files for MLRISC tools (like MDLGen) so
3624       that they properly share their libraries with existing SML/NJ libraries.
3625       * I fixed a minor cosmetic bug in CM, supressing certain spurious
3626       follow-up error messages.
3627       * Updates to CM/CMB documentation.
3628    
3629    TODO items:
3630    
3631       * MLRISC should use a different register as its asmTemp on the Sparc.
3632         (The current %o2 is a really bad choice because it is part of the
3633         calling conventions, so things might interfere in unexpected ways.)
3634    
3635    ----------------------------------------------------------------------
3636    Name: Matthias Blume
3637    Date: 2001/06/07
3638    Tag: blume-20010607-calls
3639    Description:
3640    
3641    A number of internal changes related to C calls and calling conventions:
3642    
3643    1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
3644       number of bytes popped implicitly (by the callee).  In most cases
3645       this field is 0 but on x86/win32 it is some non-zero value.  This
3646       is information provided for the benefit of the "omit-frameptr" pass.
3647    2. The CALL instruction on the x86 carries a similar "pops" field.
3648       The instruction selection phase copies its value from the ML-Tree
3649       CALL statement.
3650    3. On all other architectures, the instruction selection phase checks
3651       whether "pops=0" and complains if not.
3652    4. The c-calls implementation for x86 now accepts two calling conventions:
3653       "ccall" and "stdcall".  When "ccall" is selected, the caller cleans
3654       up after the call and pops is set to 0.  For "stdcall", the caller
3655       does nothing, leaving the cleanup to the callee; pops is set to
3656       the number of bytes that were pushed onto the stack.
3657    5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
3658       distinguish between "ccall" and "stdcall".
3659    6. The UNIMP instruction has been added to the supported Sparc instruction
3660       set. (This is needed for implementing the official C calling convention
3661       on this architecture.)
3662    7. I fixed some of the .cm files under src/MLRISC/Tools to make them
3663       work with the latest CM.
3664    
3665    ----------------------------------------------------------------------
3666    Name: Matthias Blume
3667    Date: 2001/06/05 15:10:00 EDT
3668    Tag: blume-20010605-cm-index
3669    Description:
3670    
3671    0. The "lambdasplit" parameter for class "sml" in CM has been documented.
3672    
3673    1. CM can now generate "index files".  These are human-readable files
3674       that list on a per-.cm-file basis each toplevel symbol defined or
3675       imported.  The location of the index file for
3676       <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
3677       To enable index-file generation, set CM.Control.generate_index to true
3678       or export an environment-symbol: export CM_GENERATE_INDEX=true.
3679    
3680       The CM manual has been updated accordingly.
3681    
3682    2. I made some slight modifications to the c-calls API in MLRISC.
3683    
3684         a) There is now a callback to support saving/restoring of
3685            dedicated but caller-save registers around the actual call
3686            instruction.
3687         b) One can optionally specify a comment-annotation for the
3688            call instruction.
3689    
3690    3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
3691       (For example, the comment annotation shows the C prototype of
3692        the function being called.)
3693    
3694    ----------------------------------------------------------------------
3695    Name: Matthias Blume
3696    Date: 2001/06/01 13:30:00 EDT
3697    Tag: blume-20010601-nlffi-cleanup
3698    Description:
3699    
3700    This is mostly a cleanup of MLFFI stuff:
3701    
3702       - some signature files have been put into a more exposed place
3703       - the ugly 'f type parameter is gone (simplifies types tremendously!)
3704       - ml-nlffigen changed accordingly
3705       - tutorial updated
3706    
3707    Other changes:
3708    
3709       - author's affiliation in CM manual(s) updated
3710       - some more recognized keywords added to Allen's sml.sty
3711    
3712    ----------------------------------------------------------------------
3713    Name: Matthias Blume
3714    Date: 2001/05/25 15:30:00 EDT
3715    Tag: blume-20010525-iptr
3716    Description:
3717    
3718      - put the official 110.33-README (as it appears on the ftp server) under
3719        CVS
3720      - fixed a small bug related to incomplete pointer types in
3721        ml-nlffigen
3722      - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
3723        (it does not need the 'f type parameter)
3724    
3725    ----------------------------------------------------------------------
3726    Name: Matthias Blume
3727    Date: 2001/05/23 14:30:00 EDT
3728    Tag: Release_110_33
3729    Description:
3730    
3731    New version number (110.33).  New bootfiles.
3732    
3733    ----------------------------------------------------------------------
3734    Name: Matthias Blume
3735    Date: 2001/05/22 18:06:00 EDT
3736    Tag: blume-20010522-targets
3737    Description:
3738    
3739    Made install.sh use file config/targets.customized if it exists, falling
3740    back to config/targets if it doesn't.  This way one can have a customized
3741    version of the targets file without touching the "real thing", thus
3742    eliminating the constant fear of accidentally checking something bogus
3743    back into the CVS repository...  (File config/targets.customized must
3744    not be added to the repository!)
3745    
3746    ----------------------------------------------------------------------
3747    Name: Matthias Blume
3748    Date: 2001/05/22 16:30:00 EDT
3749    Tag: blume-20010522-minitut
3750    Description:
3751    
3752    1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
3753       struct returns.
3754    2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
3755       incomplete, preliminary documentation for NLFFI.
3756    
3757    ----------------------------------------------------------------------
3758    Name: Matthias Blume
3759    Date: 2001/05/14 11:30:00 EDT
3760    Tag: blume-20010514-script
3761    Description:
3762    
3763    Some bugs in install script fixed.
3764    
3765    In addition to that I also made a slight change to the NLFFI API:
3766    Functors generated by ml-nlffigen now take the dynamic library as a
3767    straight functor argument, not as a suspended one.  (The original
3768    functor code used to force the suspension right away anyway, so there
3769    was nothing gained by this complication of the interface.)
3770    
3771    ----------------------------------------------------------------------
3772    Name: Matthias Blume
3773    Date: 2001/05/11 14:35:00 EDT
3774    Tag: blume-20010511-ml-nlffi
3775    Description:
3776    
3777    I finally took the plunge and added my new FFI code to the main
3778    repository.  For x86-linux it is now ready for prime-time.
3779    
3780    There are two new subdirectories of "src":
3781    
3782      - ml-nlffi-lib:
3783           The utility library for programs using the FFI interface.
3784           Here is the implementation of $/c.cm and its associated low-level
3785           partners $/c-int.cm and $/memory.cm.
3786      - ml-nlffigen:
3787           A stand-alone program for generating ML glue code from C source
3788           code.
3789    
3790    Building ml-nlffigen requires $/ckit-lib.cm.
3791    
3792    The config/install.sh script has been updates to do the Right Thing
3793    (hopefully).
3794    
3795    Notice that the source tree for the C-Kit will not be put under "src"
3796    but directly under the installation root directory.  (This is the
3797    structure that currently exists on the CVS server when you check out
3798    module "sml".)  Fortunately, config/install.sh knows about this oddity.
3799    
3800    Bugs: No documentation yet.
3801    
3802    ----------------------------------------------------------------------
3803    Name: Matthias Blume
3804    Date: 2001/05/09 16:35:00 EDT
3805    Tag: blume-20010509-cpscontract
3806    Description:
3807    
3808    Fixed a bug in the accounting code in cpsopt/contract.sml.  (The
3809    wrapper/unwrapper elimination did not decrement usage counts and some
3810    dead variables got overlooked by the dead-up logic.)
3811    
3812    ----------------------------------------------------------------------
3813    Name: Lal George
3814    Date: 2001/05/08  17:26:09 EDT
3815    Tag: george-20010508-omit-frameptr
3816    Description:
3817    
3818    Changes to implement the omit-frame-pointer optimization to support
3819    raw C calls. For now, there is only support on the Intel x86, but
3820    other architectures will follow as more experience is gained with this.
3821    
3822    
3823    ----------------------------------------------------------------------
3824    Name: Matthias Blume
3825    Date: 2001/05/07 14:40:00 EDT
3826    Tag: blume-20010507-proxies
3827    Description:
3828    
3829    I made into "proxy libraries" all libraries that qualify for such a
3830    change.  (A qualifying library is a library that has another library or
3831    groups as its sole member and repeats that member's export list
3832    verbatim.  A proxy library avoids this repetition by omitting its export
3833    list, effectively inheriting the list that its (only) member exports.
3834    See the CM manual for more explanation.)
3835    The main effect is that explicit export lists for these libraries
3836    do not have to be kepts in sync, making maintenance a bit easier.
3837    
3838    I also added copyright notices to many .cm-files.
3839    
3840</