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