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