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