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