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 812, Fri Apr 20 15:07:30 2001 UTC revision 1391, Wed Sep 24 21:31:51 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: 2003/09/24 16:31:00 CDT
18    Tag: blume-20030924-installer
19    Description:
20    
21    Another installer tweak:  All the ML code for the installer is now
22    compiled during CMB.make and put into a little library called
23    $smlnj/installer.cm.  The installation then simply invokes
24    
25       sml -m $smlnj/installer.cm
26    
27    and everything happens automagically.
28    
29    Win32:  ML code senses value of environment variable SMLNJ_HOME.
30    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
31            and BINDIR.
32    
33    The new scheme guarantees that the ML code responsible for the installation
34    is in sync with the APIs of the main system.  Also, the installer is
35    somewhat faster because the installer script is precompiled.
36    
37    ----------------------------------------------------------------------
38    Name: Matthias Blume (blume (at) tti - c (dot) org)
39    Date: 2003/09/24 15:35:00 CDT
40    Tag: blume-20030924-synsock
41    Description:
42    
43    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
44    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
45    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
46    implementations of non-blocking operations from CML's Socket
47    structure.
48    
49    ----------------------------------------------------------------------
50    Name: Matthias Blume (blume (at) tti - c (dot) org)
51    Date: 2003/09/24 15:10:05 CDT
52    Tag: blume-20030924-sockets
53    Description:
54    
55    1. Fixed SOCKET API and implementation to match Basis spec.
56       This required changing the internal representation of sockets to one
57       that remembers (for each socket file descriptor) whether it is currently
58       blocking or non-blocking.  This state is maintained lazily (i.e., a system
59       call is made only if the state actually needs to change).
60    
61    2. OS-specific details of sockets were moved into separate files, thus
62       making it possible to unify the bulk of the socket implementations
63       between Unix and Win32.
64    
65    3. CML's socket API changed accordingly.
66       (Note that we need to remove non-blocking functions from this API
67       since they are redundant in the case of CML!)
68    
69    4. CML's socket implementation now makes use of non-blocking functions
70       provided by Basis, thus removing all OS-dependent code from this part
71       of CML.
72    
73    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
74    
75    ----------------------------------------------------------------------
76    Name: Matthias Blume (blume (at) tti - c (dot) org)
77    Date: 2003/09/22 12:10:00 CDT
78    Tag: blume-20030922-110_43_2
79    Description:
80    
81    Made a new interim version and bootfiles for developer's bootstrapping
82    convenience.
83    
84       110.43.2 -- NEW BOOTFILES
85    
86    ----------------------------------------------------------------------
87    Name: Matthias Blume (blume (at) tti - c (dot) org)
88    Date: 2003/09/19 15:55:00 CDT
89    Tag: blume-20030919-cmdir
90    Description:
91    
92    1. new-install.sh -> install.sh
93    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
94    3. tweaked installer so that another name instead of .cm can be chosen
95       at install time (by setting the CM_DIR_ARC environment variable
96       during installation); once installation is complete, the name is
97       fixed
98    
99    ----------------------------------------------------------------------
100    Name: Matthias Blume (blume (at) tti - c (dot) org)
101    Date: 2003/09/18 16:00:00 CDT
102    Tag: blume-20030918-110_43_1
103    Description:
104    
105    Made a new interim version and bootfiles for developer's bootstrapping
106    convenience.
107    
108       110.43.1 -- NEW BOOTFILES
109    
110    ----------------------------------------------------------------------
111    Name: Matthias Blume (blume (at) tti - c (dot) org)
112    Date: 2003/09/18 15:20:00 CDT
113    Tag: blume-20030918-misc
114    Description:
115    
116    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
117       this seems to be controversial at the moment) and used those in
118       Posix.ProcEnv.times.
119    
120    2. Added Time.{from,to}Nanoseconds to Time.
121    
122    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
123       For example, fromLargeInt never needs to look at more than 3 "big
124       digits" to get its 53 bits of precision.
125    
126    ----------------------------------------------------------------------
127    Name: Matthias Blume (blume (at) tti - c (dot) org)
128    Date: 2003/09/17 16:30:00 CDT
129    Tag: blume-20030917-real32-slices
130    Description:
131    
132    Added an entry to the primitive environment
133    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
134    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
135    
136    Removed some of the "magic" constants in real64.sml and replaced them
137    with code that generates these values from their corresponding
138    integer counterparts.
139    
140    Made all(?) the slice-related changes to the Basis and made everything
141    compile again...
142    
143    ----------------------------------------------------------------------
144    Name: Matthias Blume (blume (at) tti - c (dot) org)
145    Date: 2003/09/15 17:45:00 CDT
146    Tag: blume-20030915-rbase
147    Description:
148    
149    Fixed bug in Real.fromLargeInt.
150    
151    ----------------------------------------------------------------------
152    Name: Matthias Blume (blume (at) tti - c (dot) org)
153    Date: 2003/09/13 18:11:00 CDT
154    Tag: blume-20030913-libinstall
155    Description:
156    
157    Minor bugfix in config/libinstall (set anchor with path to
158    standalone tool after installing it, otherwise libraries that
159    need ml-lex or ml-yacc won't compile the first time the installer
160    runs).
161    
162    ----------------------------------------------------------------------
163    Name: Matthias Blume (blume (at) tti - c (dot) org)
164    Date: 2003/09/12 11:45:00 CDT
165    Tag: blume-20030912-various
166    Description:
167    
168     - fixed bug in Real.toLargeInt
169     - fixed bug in Posix.ProcEnv.times
170     - changed inputLine functions to return an option
171     - minor installer improvements / bugfixes
172     - changed default @SMLalloc parameter for x86/celeron to 64k
173    
174    ----------------------------------------------------------------------
175    Name: Matthias Blume (blume (at) tti - c (dot) org)
176    Date: 2003/09/09 22:00:00 CDT
177    Tag: Release_110_43
178    Description:
179    
180        New working release 110.43.  New bootfiles.
181    
182    ----------------------------------------------------------------------
183    Name: Matthias Blume (blume (at) tti - c (dot) org)
184    Date: 2003/09/09 19:20:00 CDT
185    Tag: blume-20030909-installer
186    Description:
187    
188    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
189    Modified config/install.bat to take advantage of it.  Also modified
190    config/install.sh (and called it config/new-install.sh) to take advantage
191    of it on Unix systems.  (The SML code is (supposed to be) platform-
192    independent.)
193    
194    The installer can now install everything under Win32
195    as well as under *nix as long as it compiles.
196    
197    Other changes:
198    
199      - made CML compile again under Win32
200      - made eXene compile under Win32 (by providing a fake structure UnixSock
201        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
202      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
203        same as type int;  under Win32 it isn't
204      - fixed some slice-related problems in the win32-specific parts of CML
205      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
206        (under Win32, the volume name is case-insensitive, and the
207        OS.Path code compares volume names for equality)
208    
209    ----------------------------------------------------------------------
210    Name: Matthias Blume (blume (at) tti - c (dot) org)
211    Date: 2003/09/08 11:55:00 CDT
212    Tag: blume-20030908-fullpath
213    Description:
214    
215    Made Win32 version of OS.FileSys.fullPath return current directory
216    when given an empty string.  This is what the spec says, and incidentally,
217    CM depends on it.  (CM otherwise goes into an infinite loop in certain
218    cases when presented with the name of a non-existing .cm file.)
219    
220    ----------------------------------------------------------------------
221    Name: Matthias Blume (blume (at) tti - c (dot) org)
222    Date: 2003/09/04 16:30:00 CDT
223    Tag: blume-20030905-slices-etc
224    Description:
225    
226      1. Changed interface to vectors and arrays in Basis to match
227         (draft) Basis spec.
228      2. Added signatures and implementations of slices according to
229         Basis spec.
230      3. Edited source code throughout the system to make it compile again
231         under 1. and 2.  (In some cases code had to be added to have it
232         match the new signatures.)
233      4. MLRISC should be backward-compatible: the copies of the originals
234         of files that needed to change under 3. were retained, the .cm files
235         check the compiler version number and use old versions when
236         appropriate.
237      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
238         dirstream -> string option (in accordance with Basis spec).
239      6. When generating code that counts lines, ml-lex used function
240         CharVector.foldli, taking advantage of its old interface.
241         This has been replaced with the corresponding code from
242         CharVectorSlice.  (html-lex must be re-lexed!)
243      7. BitArray in smlnj-lib/Util has been extended/modified to match the
244         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
245      8. Removed temporary additions (fromInternal, toInternal) from the
246         (now obsolete) IntInf in smlnj-lib/Util.
247      9. Cleaned up structure Byte.
248     10. Added localOffset, scan, and fromString to Date (according to spec).
249         Cleaned/corrected implementation of Date.
250         (Still need to check for correctness; implement better canonicalizeDate.)
251     11. Added "scan" to signature IEEE_REAL.
252     12. Some improvements to IntInf [in particular: efficiency-hack for
253         mod and rem when second operand is 2 (for parity checks).]
254     13. Changed representation of type Time.time, using a single IntInf.int
255         value counting microseconds.  This considerably simplified the
256         implementation of structure Time.  We now support negative time
257         values; scan and fromString handle signs.
258     14. Functor PrimIO now takes two additional arguments (VectorSlice and
259         ArraySlice).
260    
261    ----------------------------------------------------------------------
262    Name: Matthias Blume (blume (at) tti - c (dot) org)
263    Date: 2003/08/28 17:00:00 CDT
264    Tag: blume-20030828-intinf
265    Description:
266    
267    This is a major update which comes with a version number bump
268    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
269    and an implementation of IntInf in the Basis.
270    
271    There are a fairly large number of related changes and updates throughout
272    the system:
273    
274    Basis:
275            - Implemented IntInf.
276            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
277            - Added some missing Real64 operations, most notably Real.toLargeInt.
278            - Added FixedInt as a synonym for Int32.
279    
280    compiler:
281            * Added support for a built-in intinf type.
282              - literals
283              - pattern matching
284              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
285              - overloading on literals and operations
286    
287            This required adding a primitive type intinf, some additional
288            primops, and implementations for several non-trivial intinf
289            operations in Core.  (The intinf type is completely abstract
290            to the compiler; all operations get delegated back to the Core.)
291    
292            * Intinf equality is handled by polyequal.  However, the compiler
293            does not print its usual warning in this case (since polyequal
294            is the right thing to do there).
295    
296            * Improved the organization of structure InlineT.
297    
298            * A word about conversion primops:
299            If conversions involving intinf do not cancel out during
300            CPS contract, then the compiler must insert calls to Core functions.
301            Since all core access must be resolved already during the FLINT
302            translate phase, it would be too late a the time of CPS contract
303            to add new Core calls.  For this reason, conversion primops
304            for intinf carry two arguments:  1. the numeric argument that
305            they are supposed to convert, and 2. the Core function that
306            can help with this conversion if necessary.  If CPS contract
307            eliminates a primop, then the associated Core function becomes
308            dead and goes away.  Intinf conversion primops that do not get
309            eliminated by CPS contract get rewritten into calls of their
310            core functions by a separate, new phase.
311    
312    interactive system:
313            - Control.Print.intinfDepth controls max length of intinf constants
314            being printed. (Analogous to Control.Print.stringDepth.)
315            - Cleanup in printutil and pputil: got rid of unused stuff and
316            duplicates; replaced some of the code with code that makes better
317            use of library functionality.
318    
319    CM:
320            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
321            are no longer silent.
322    
323    CKIT:
324            Fixed mismatched uses of Int32 and LargeInt.  I always decided
325            in favor of LargeInt -- which is now the same as IntInf.
326            CKIT-knowledgable people should check whether this is what's
327            intended and otherwise change things back to using Int32 or
328            FixedInt.
329    
330    Throughout the code:
331            Started using IntInf.int literals and built-in operations
332            (e.g., comparison with 0) where this seems appropriate.
333    
334    
335    ----------------------------------------------------------------------
336    Name: Dave MacQueen (dbm@cs.uchicago.edu)
337    Date: 2003/08/13 11:36:00 CDT
338    Tag: dbm-20030813-mcz-merge1
339    Description:
340    
341    Merging changes from the mcz-branch development branch into trunk.
342    These changes involve replacement of the emulated old prettyprinter
343    interface with direct use of the SML/NJ Lib PP library, and fixing
344    of a couple of bugs (895, 1186) relating to error messages.  A new
345    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
346    has been added.
347    
348    ----------------------------------------------------------------------
349    Name: Matthias Blume (blume (at) tti - c (dot) org)
350    Date: 2003/08/11 15:45:00 CDT
351    Tag: blume-20030811-windows
352    Description:
353    
354    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
355    
356        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
357    
358    This patch restores SML/NJ's ability to run under win32.  There are a
359    number of changes, including fixes for several bugs that had gone
360    unnoticed until now:
361    
362      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
363        trying to run the win32 version from within cygwin.)
364    
365      - There are a number of simple .bat scripts that substitute for their
366        corresponding Unix shell-scripts.  (See below.)
367    
368      - The internals of ml-build have been modified slightly.  The main
369        difference is that instead of calling ".link-sml" (or link-sml.bat)
370        using OS.Process.system, the ML process delegates this task back
371        to the script.  Otherwise problems arise in mixed environments such
372        as Cygwin where scripts look and work like Unix scripts, but
373        where OS.Process.system cannot run them.
374    
375      - In CM, the srcpath pickler used native pathname syntax -- which
376        is incorrect in the case of cross-compilation.  The new pickle format
377        is independent of platform-specific naming conventions.
378    
379      - Path configuration files (such as lib/pathconfig) can now choose
380        between native and standard syntax.  Placing a line of the form
381    
382            standard!
383    
384        into the file causes all subsequent paths to be interpreted using
385        CM standard pathname syntax (= Unix conventions); a line
386    
387            native!
388    
389        switches back to native style.  This was needed so that
390        path config files can be written portably, see src/system/pathconfig.
391    
392      - Runtime system:
393    
394          - win32-filesys.c:  get_file_time and set_file_time now
395            access modification time, not creation time.
396    
397          - I/O code made aware of new array representation.
398    
399          - Bug fixes in X86.prim.masm.
400    
401       - src/system/makeml made aware of win32. (For use under cygwin
402            and other Unix-environments for windows.)
403    
404       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
405         which caused BinIO.inputAll to fail consistently.
406    
407    .bat scripts:
408    
409       Windows .bat scripts assume that SMLNJ_HOME is defined.
410    
411       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
412         applications (sml, ml-yacc, ml-lex).
413       - ml-build.bat: analogous to ml-build.
414       - config\install.bat:  Analogous to config/install.sh.  This requires
415         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
416         (nmake etc. must be on the path, and vcvars32 must have been run.)
417         Moreover, sources for ml-lex and ml-yacc need to exist under src,
418         and the bootfile hierarchy must have been unpacked under
419         sml.boot.x86-win32.
420         The script is very primitive and does a poor job at error checking.
421         It only installs the base system, ml-lex, and ml-yacc.  No other
422         libraries are being installed (i.e., you get only those that
423         are part of the compiler.)
424       - link-sml.bat:  analogous to .link-sml, but not currently used
425    
426    Unrelated bug fixes:
427    
428      - ml-nlffigen now exports structures ST_* corresponding to incomplete
429        types.
430      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
431        otherwise.)
432    
433    ----------------------------------------------------------------------
434    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
435    Date: 2003/06/17
436    Tag: macqueen-20030617-bug895
437    Description:
438    
439    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
440    Tag will be used for new development branch (mcz-branch) for
441    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
442    
443    ----------------------------------------------------------------------
444    Name: Matthias Blume (blume (at) tti - c (dot) org)
445    Date: 2003/05/27 16:55:00 CDT
446    Tag: blume-20030527-polyeq
447    Description:
448    
449    Tried to eliminated most cases of polymorphic equality.
450    
451    ----------------------------------------------------------------------
452    Name: Matthias Blume (blume (at) tti - c (dot) org)
453    Date: 2003/05/21 17:45:00 CDT
454    Tag: blume-20030517-complete
455    Description:
456    
457    Two changes:
458    
459      1. Added a flag for controlling whether non-exhaustive bindings will
460         be treated as errors (default is false).
461      2. Cleaned up the *entire* source tree so that CMB.make goes through
462         without a single non-exhaustive match- or bind warning.
463    
464    ----------------------------------------------------------------------
465    Name: Matthias Blume (blume (at) tti - c (dot) org)
466    Date: 2003/05/17 10:20:00 CDT
467    Tag: blume-20030517-absyn
468    Description:
469    
470    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
471    
472       This mainly affects the quality of error messages.  However, some
473       of the code is now more straightforward than before. (Treatment of
474       the above four constructs in translate.sml is much simpler than
475       the "macro-expansion" that was going on before.  Plus, the mach-
476       compiler no longer gets invoked just to be able to compile an
477       if-expression.)
478    
479    2. The ErrorMsg.Error exception is now caught and absorbed by the
480       interactive loop.
481    
482    ----------------------------------------------------------------------
483    Name: Allen Leung
484    Date: 2003/05/16 13:05:00 CDT
485    Tag: leunga-20030516-cygwin-runtime
486    Description:
487    
488        Ported the runtime system to cygwin, which uses the unix
489    x86-unix bin files.  Missing/buggy features:
490    
491       o getnetbyname, getnetbyaddr: these functions seem to be missing in
492         the Cygwin library.
493       o Ctrl-C handling may be flaky.
494       o Windows system calls and Windows I/O are not supported.
495    
496       A new set of binfiles is located at:
497    
498         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
499    
500       This is only needed for bootstrapping the cygwin version of smlnj.
501       Other x86 versions can use the existing binfiles.
502    
503    ----------------------------------------------------------------------
504    Name: Matthias Blume
505    Date: 2003/04/08 15:42:00 CDT
506    Tag: blume-20030408-listpair
507    Description:
508    
509    1. Added a target 'mlrisc' to installer.
510    
511    2. Added missing elements to structure ListPair.
512    
513    ----------------------------------------------------------------------
514    Name: Allen Leung
515    Date: 2003/01/07 10:40:00 EST
516    Tag: leunga-20030107-int-rem
517    Description:
518    
519    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
520    The arguments to the SUBL instruction were swapped.
521    
522    ----------------------------------------------------------------------
523    Name: Matthias Blume
524    Date: 2002/12/12 16:25:00 EST
525    Tag: blume-20021212-risc-ra
526    Description:
527    
528    Fixed a serious bug in the rewrite code for FP spilling/reloading that
529    sent the RA into an infinite loop when floating point registers get
530    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
531    110.37 and 110.38.)
532    There was another set of potential problems related to the handling of
533    MLRISC annotations (but those did not yet cause real problems, apparently).
534    
535    ----------------------------------------------------------------------
536    Name: Matthias Blume
537    Date: 2002/12/06 22:40:00 EST
538    Tag: blume-20021206-cm-fileid
539    Description:
540    
541    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
542    This fixes the problem of CM getting confused by files that suddenly
543    change their identity (e.g., by getting unlinked and recreated by some
544    text editor such as vi).  There might be a better/cheaper/cleaner way
545    of doing this, but for now this will have to do.
546    
547    ----------------------------------------------------------------------
548    Name: Matthias Blume
549    Date: 2002/10/28 09:50:00 EST
550    Tag: blume-20021028-typecheck
551    Description:
552    
553    Exported structure Typecheck from $smlnj/viscomp/core.cm.
554    
555    ----------------------------------------------------------------------
556    Name: Matthias Blume
557    Date: 2002/10/17 09:10:00 EDT
558    Tag: Release_110_42
559    Description:
560    
561    In good old tradition, there has been a slight hiccup so that we have
562    to patch 110.42 after the fact.   The old release tag has been replaced
563    (see below).
564    
565    The change solves a problem with two competing approaches the
566    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
567    each other's way.
568    
569    This change only affects the runtime system code and the installer script.
570    (No new bootfiles.)
571    
572    ----------------------------------------------------------------------
573    Name: Matthias Blume
574    Date: 2002/10/16 12:00:00 EDT
575    Tag: Release_110_42_removed
576    Description:
577    
578    New working release.  New bootfiles.
579    
580    ----------------------------------------------------------------------
581    Name: Matthias Blume
582    Date: 2002/10/10 13:10:00 EDT
583    Tag: blume-20021010-ppc-divs
584    Description:
585    
586    The mltree operator DIVS must be implemented with an overflow check on
587    the PPC because the hardware indicates divide-by-zero using "overflow" as
588    well.
589    
590    ----------------------------------------------------------------------
591    Name: Matthias Blume
592    Date: 2002/07/23 11:45:00
593    Tag: blume-20020723-smlnj-home
594    Description:
595    
596    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
597    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
598    CM_PATHCONFIG is also set), the path configuration file is assumed
599    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
600    the entire tree to some other place and everything will "just work".
601    
602    (Companion commands such as ml-build and ml-makedepend also sense this
603    variable.)
604    
605    ----------------------------------------------------------------------
606    Name: Matthias Blume
607    Date: 2002/07/12 21:19:00 EDT
608    Tag: blume-20020712-liveness
609    Description:
610    
611    Exported two useful "step" functions from liveness module (MLRISC).
612    
613    ----------------------------------------------------------------------
614    Name: Matthias Blume
615    Date: 2002/07/05 16:00 EDT
616    Tag: Release_110_41
617    Description:
618    
619    New working release.  New bootfiles.
620    
621    ----------------------------------------------------------------------
622    Name: Matthias Blume
623    Date: 2002/07/05 10:25:00 EDT
624    Tag: blume-20020705-btimp
625    Description:
626    
627    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
628    other clients can set up backtracing support.
629    
630    ----------------------------------------------------------------------
631    Name: Matthias Blume
632    Date: 2002/06/25 17:23:00 EDT
633    Tag: blume-20020625-fpmax
634    Description:
635    
636    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
637    case.  (The sense of the isNaN test was reversed -- which made min and
638    max always return their first argument.)
639    
640    ----------------------------------------------------------------------
641    Name: Matthias Blume
642    Date: 2002/06/11
643    Tag: blume-20020611-unixpath
644    Description:
645    
646    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
647    
648    ----------------------------------------------------------------------
649    Name: Matthias Blume
650    Date: 2002/06/10 16:35:00 EDT
651    Tag: blume-20020610-ieeereal
652    Description:
653    
654    I back-ported my implementation of IEEEReal.fromString from the
655    idlbasis-devel branch so that we can test it.
656    
657    Another small change is that ppDec tries to give more information
658    than just "<sig>" in the case of functors.  However, this code is
659    broken in some mysterious way if the functor's body's signature
660    has not been declared by ascription but gets inferred from the
661    implementation.  This needs fixing...
662    
663    ----------------------------------------------------------------------
664    Name: Matthias Blume
665    Date: 2002/05/31
666    Tag: blume-20020531-btrace-mode
667    Description:
668    
669    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
670    the wayside when I switched over to using Controls everywhere.)
671    
672    ----------------------------------------------------------------------
673    Name: Lal George
674    Date: 2002/05/23 12:21:40 EDT
675    Tag: george-20020523-visual-labels
676    Description:
677    
678    Labels are now displayed in the graphical output to make
679    the fall-through and target blocks obvious.
680    
681    ----------------------------------------------------------------------
682    Name: Matthias Blume
683    Date: 2002/05/22 11:03:00 EDT
684    Tag: blume-20020522-shrink
685    Description:
686    
687    John tweaked yesterday's fix for 1131 to handle an out-of-memory
688    situation that comes up when allocating huge arrays.
689    
690    ----------------------------------------------------------------------
691    Name: Matthias Blume
692    Date: 2002/05/21 16:00:00 EDT
693    Tag: Release_110_40
694    Description:
695    
696    New working release (110.40).  New bootfiles.
697    
698    [Also: John Reppy fixed GC bug 1131.]
699    
700    ----------------------------------------------------------------------
701    Name: Matthias Blume
702    Date: 2002/05/21 12:35:00 EDT
703    Tag: blume-20020521-cmdoc
704    Description:
705    
706    CM documentation update.
707    
708    ----------------------------------------------------------------------
709    Name: Matthias Blume
710    Date: 2002/05/21 10:55:00 EDT
711    Tag: blume-20020521-misc
712    Description:
713    
714    - John tweaked runtime to be silent on heap export (except when
715      GC messages are on).
716    - I added a few more things (cross-compiling versions of CMB) to
717      config/preloads (as suggestions).
718    
719    ----------------------------------------------------------------------
720    Name: Matthias Blume
721    Date: 2002/05/20 22:25:00 EDT
722    Tag: blume-20020520-controls
723    Description:
724    
725    - Added ControlUtil structure to control-lib.cm.
726    - Use it throughout.
727    - Used Controls facility to define MLRISC controls (as opposed to
728      registering MLRISC control ref cells with Controls after the fact)
729    - Fixed messed-up controls priorities.
730    
731    * Removed again all the stuff from config/preloads that one wouldn't
732      be able to preload at the time the initial heap image is built.
733      (Many libraries, e.g., CML, do not exist yet at this time.  The
734       only libraries that can be preloaded via config/preloads are those
735       that come bundled with the bootfiles.)
736    
737    ----------------------------------------------------------------------
738    Name: Matthias Blume
739    Date: 2002/05/20 10:59:00 EDT
740    Tag: blume-20020520-preloads
741    Description:
742    
743    Added a lot of commented-out suggestions for things to be included
744    in config/preloads.
745    
746    ----------------------------------------------------------------------
747    Name: Allen Leung
748    Date: 2002/05/18 14:20:00 EDT
749    Tag: leunga-20020518-mdl
750    Description:
751    
752    o Made the mdl tool stuff compile and run again.
753    o I've disabled all the stuff that depends on RTL specifications; they
754      are all badly broken anyway.
755    
756    ----------------------------------------------------------------------
757    Name: Matthias Blume
758    Date: 2002/05/17 16:49:00 EDT
759    Tag: blume-20020517-controls
760    Description:
761    
762    0. John Reppy made several modifications to the SML/NJ library.
763       In particular, there is a shiny new controls-lib.cm.
764    
765    1. Pushed new controls interface through compiler so that everything
766       compiles again.
767    
768    2. Added FormatComb and FORMAT_COMB to the CML version of the
769       SML/NJ library (so that CML compiles again).
770    
771    3. Modified init scripts because XXX_DEFAULT environment variables
772       are no longer with us.  (Boot-time initialization is now done
773       using the same environment variables that are also used for
774       startup-time initialization of controls.)
775    
776    ----------------------------------------------------------------------
777    Name: Lal George
778    Date: 2002/05/15 09:20:10 EDT
779    Tag: george-20020515-pseudo-op-decls
780    Description:
781    
782            All pseudo-ops emitted before the first segment declaration
783    such as TEXT, DATA, and BSS directives are assumed to be global
784    declarations and are emitted first in the assembly file. This is
785    useful in a number of situations where one has pseudo-ops that are not
786    specific to any segment, and also works around the constraint that one
787    cannot have client pseudo-ops in the TEXT segment.
788    
789            Because no segment is associated with these declarations it is
790    an error to allocate any space or objects before the first segment
791    directive and an exception will be raised. However, we cannot make
792    this check for client pseudo-ops.
793    
794            These top level declarations are a field in the CFG graph_info.
795    In theory you can continue to add to this field after the CFG has been
796    built -- provided you know what you are doing;-)
797    
798    ----------------------------------------------------------------------
799    Name: Matthias Blume
800    Date: 2002/05/13 16:40:00 EDT
801    Tag: blume-20020513-pp-etc
802    Description:
803    
804    A few minor bugfixes:
805    
806      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
807        (Bogus printouts for redefined bindings still occur.  Compiler
808        bug should no longer occur now.  We need to redo the prettyprinter
809        from scratch.)
810    
811      - CM pathname printer now also adds escape sequences for ( and )
812    
813      - commend and docu fixes for ml-nlffi
814    
815    ----------------------------------------------------------------------
816    Name: Matthias Blume
817    Date: 2002/05/10 16:40:00 EDT
818    Tag: blume-20020510-erg-textio
819    Description:
820    
821    Applied the following bugfix provided by Emden Gansner:
822    
823        Output is corrupted when outputSubstr is used rather than output.
824    
825        The problem occurs when a substring
826    
827            ss = (s, dataStart, dataLen)
828    
829        where dataStart > 0, fills a stream buffer with avail bytes left.
830        avail bytes of s, starting at index dataStart, are copied into the
831        buffer, the buffer is flushed, and then the remaining dataLen-avail
832        bytes of ss are copied into the beginning of the buffer. Instead of
833        starting this copy at index dataStart+avail in s, the current code
834        starts the copy at index avail.
835    
836        Fix:
837        In text-io-fn.sml, change line 695 from
838             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
839        to
840             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
841    
842    ----------------------------------------------------------------------
843    Name: Matthias Blume
844    Date: 2002/04/12 13:55:00 EDT
845    Tag: blume-20020412-assyntax
846    Description:
847    
848    1. Grabbed newer assyntax.h from the XFree86 project.
849    2. Fiddled with how to compile X86.prim.asm without warnings.
850    3. (Very) Minor cleanup in CM.
851    
852    ----------------------------------------------------------------------
853    Name: Matthias Blume
854    Date: 2002/04/01 (no joke!) 17:07:00 EST
855    Tag: blume-20020401-x86div
856    Description:
857    
858    Added full support for div/mod/rem/quot on the x86, using the machine
859    instruction's two results (without clumsily recomputing the remainder)
860    directly where appropriate.
861    
862    Some more extensive power-of-two support was added to the x86 instruction
863    selector (avoiding expensive divs, mods, and muls where they can be
864    replaced with cheaper shifts and masks).  However, this sort of thing
865    ought to be done earlier, e.g., within the CPS optimizer so that
866    all architectures benefit from it.
867    
868    The compiler compiles to a fixed point, but changes might be somewhat
869    fragile nevertheless.  Please, report any strange things that you might
870    see wrt. div/mod/quot/rem...
871    
872    ----------------------------------------------------------------------
873    Name: Matthias Blume
874    Date: 2002/03/29 17:22:00
875    Tag: blume-20020329-div
876    Description:
877    
878    Fixed my broken div/mod logic.  Unfortunately, this means that the
879    inline code for div/mod now has one more comparison than before.
880    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
881    The problem was with quotient = 0, because that alone does not tell
882    us which way the rounding went.  One then has to look at whether
883    remainder and divisor have the same sign...  :(
884    
885    Anyway, I replaced the bootfiles with fresh ones...
886    
887    ----------------------------------------------------------------------
888    Name: Matthias Blume
889    Date: 2002/03/29 14:10:00 EST
890    Tag: blume-20020329-inlprims
891    Description:
892    
893    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
894    
895    Primops have changed. This means that the bin/boot-file formats have
896    changed as well.
897    
898    To make sure that there is no confusion, I made a new version.
899    
900    
901    CHANGES:
902    
903    * removed REMT from mltree (remainder should never overflow).
904    
905    * added primops to deal with divisions of all flavors to the frontend
906    
907    * handled these primops all the way through so they map to their respective
908      MLRISC support
909    
910    * used these primops in the implementation of Int, Int32, Word, Word32
911    
912    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
913    
914    * parameterized INLMIN, INLMAX, and INLABS by a numkind
915    
916    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
917      floating point
918    
919    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
920      and Real (but Real.abs maps to a separate floating-point-only primop)
921    
922    
923    TODO items:
924    
925    * Hacked Alpha32 instruction selection, disabling the selection of REMx
926      instructions because the machine instruction encoder cannot handle
927      them.  (Hppa, PPC, and Sparc instruction selection did not handle
928      REM in the first place, and REM is supported by the x86 machine coder.)
929    
930    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
931      selection phase.  (The two can be streamlined because the hardware
932      delivers both quotient and remainder at the same time anyway.)
933    
934    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
935      (Currently the behavior is inconsistent both across architectures and
936      wrt. the draft Basis spec.)
937    
938    * Word8 should eventually be handled natively, too.
939    
940    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
941      as if there currently is no execution path that could trigger it in
942      SML/NJ.  (The assumptions underlying functions arith and promotable do not
943      hold for things like multiplication and division.)
944    
945    ----------------------------------------------------------------------
946    Name: Matthias Blume
947    Date: 2002/03/27 16:27:00 EST
948    Tag: blume-20020327-mlrisc-divisions
949    Description:
950    
951    Added support for all four division operations (ML's div, mod, quot,
952    and rem) to MLRISC.  In the course of doing so, I also rationalized
953    the naming (no more annoying switch-around of DIV and QUOT), by
954    parameterizing the operation by div_rounding_mode (which can be either
955    DIV_TO_ZERO or DIV_TO_NEGINF).
956    
957    The generic MLTreeGen functor takes care of compiling all four
958    operations down to only round-to-zero div.
959    
960    Missing pieces:
961    
962      * Doing something smarter than relying on MLTreeGen on architectures
963        like, e.g., the x86 where hardware division delivers both quotient and
964        remainder at the same time.  With this, the implementation of the
965        round-to-neginf operations could be further streamlined.
966    
967      * Remove inlining support for div/mod/rem from the frontend and replace it
968        with primops that get carried through to the backend.  Do this for all
969        int and word types.
970    
971    ----------------------------------------------------------------------
972    Name: Matthias Blume
973    Date: 2002/03/25 17:25:00 EST
974    Tag: blume-20020325-divmod
975    Description:
976    
977    I improved (hopefully without breaking them) the implementation of Int.div,
978    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
979    advantage of the following observations:
980    
981      Let  q = x quot y      r = x rem y
982           d = x div  y      m = x mod y
983    
984    where "quot" is the round-to-zero version of integer division that
985    hardware usually provides.  Then we have:
986    
987         r = x - q * y        where neither the * nor the - will overflow
988         d = if q >= 0 orelse x = q * y then q else q - 1
989                              where neither the * nor the - will overflow
990         m = if q >= 0 orelse r = 0 then r else r + y
991                              where the + will not overflow
992    
993    This results in substantial simplification of the generated code.
994    The following table shows the number of CFG nodes and edges generated
995    for
996            fun f (x, y) = x OPER y
997            (* with OPER \in div, mod, quot, rem *)
998    
999    
1000        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1001        --------------------------------------------------------
1002         div |         24 |         39 |         12 |         16
1003         mod |         41 |         71 |         12 |         16
1004        quot |          8 |         10 |          8 |         10
1005         rem |         10 |         14 |          8 |         10
1006    
1007    
1008    ----------------------------------------------------------------------
1009    Name: Matthias Blume
1010    Date: 2002/03/25 22:06:00 EST
1011    Tag: blume-20020325-cprotobug
1012    Description:
1013    
1014    Fixed a bug in cproto (c prototype decoder).
1015    
1016    ----------------------------------------------------------------------
1017    Name: Matthias Blume
1018    Date: 2002/03/25 16:00:00 EST
1019    Tag: blume-20020325-raw-primops
1020    Description:
1021    
1022    I did some cleanup to Allen's new primop code and
1023    replaced yesterday's bootfiles with new ones.
1024    (But they are stored in the same place.)
1025    
1026    ----------------------------------------------------------------------
1027    Name: Matthias Blume
1028    Date: 2002/03/24 22:40:00 EST
1029    Tag: blume-20020324-bootfiles
1030    Description:
1031    
1032    Made the bootfiles that Allen asked for.
1033    
1034    ----------------------------------------------------------------------
1035    Name: Allen Leung
1036    Date: 2002/03/23 15:50:00 EST
1037    Tag: leunga-20020323-flint-cps-rcc-primops
1038    Description:
1039    
1040      1. Changes to FLINT primops:
1041    
1042        (* make a call to a C-function;
1043         * The primop carries C function prototype information and specifies
1044         * which of its (ML-) arguments are floating point. C prototype
1045         * information is for use by the backend, ML information is for
1046         * use by the CPS converter. *)
1047      | RAW_CCALL of { c_proto: CTypes.c_proto,
1048                       ml_args: ccall_type list,
1049                       ml_res_opt: ccall_type option,
1050                       reentrant : bool
1051                     } option
1052       (* Allocate uninitialized storage on the heap.
1053        * The record is meant to hold short-lived C objects, i.e., they
1054        * are not ML pointers.  With the tag, the representation is
1055        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1056        *)
1057      | RAW_RECORD of {tag:bool,sz:int}
1058      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1059    
1060      2.  These CPS primops are now overloaded:
1061    
1062           rawload of {kind:numkind}
1063           rawstore of {kind:numkind}
1064    
1065          The one argument form is:
1066    
1067             rawload {kind} address
1068    
1069          The two argument form is:
1070    
1071             rawload {kind} [ml object, byte-offset]
1072    
1073      3. RAW_CCALL/RCC now takes two extra arguments:
1074    
1075         a. The first is whether the C call is reentrant, i.e., whether
1076            ML state should be saved and restored.
1077         b. The second argument is a string argument specifying the name of
1078            library and the C function.
1079    
1080         These things are currently not handled in the code generator, yet.
1081    
1082      4. In CProto,
1083    
1084         An encoding type of "bool" means "ml object" and is mapped into
1085         C prototype of PTR.  Note that "bool" is different than "string",
1086         even though "string" is also mapped into PTR, because "bool"
1087         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1088    
1089      5. Pickler/unpicker
1090    
1091         Changed to handle RAW_RECORD and newest RAW_CCALL
1092    
1093      6. MLRiscGen,
1094    
1095         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1096         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1097            in the file CodeGen/cpscompile/cps-c-calls.sml
1098    
1099      7. Added the conditional move operator
1100    
1101             condmove of branch
1102    
1103         to cps.  Generation of this is still buggy so it is currently
1104         disabled.
1105    
1106    ----------------------------------------------------------------------
1107    Name: Lal George
1108    Date: 2002/03/22 14:18:25 EST
1109    Tag: george-20020322-cps-branch-prob
1110    Description:
1111    
1112    Implemented the Ball-Larus branch prediction-heuristics, and
1113    incorporated graphical viewers for control flow graphs.
1114    
1115    Ball-Larus Heuristics:
1116    ---------------------
1117    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1118    
1119    By design it uses the Dempster-Shafer theory for combining
1120    probabilities.  For example, in the function:
1121    
1122        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1123    
1124    the ball-larus heuristics predicts that the n=0 is unlikely
1125    (OH-heuristic), and the 'then' branch is unlikely because of the
1126    RH-heuristic -- giving the 'then' branch an even lower combined
1127    probability using the Dempster-Shafer theory.
1128    
1129    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1130    probability of the 'then' branch because of the loop in the else
1131    branch.
1132    
1133    
1134    Graphical Viewing:
1135    ------------------
1136    I merely plugged in Allen's graphical viewers into the compiler. The
1137    additional code is not much. At the top level, saying:
1138    
1139            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1140    
1141    will display the graphical view of the control flow graph just before
1142    back-patching.  daVinci must be in your path for this to work. If
1143    daVinci is not available, then the default viewer can be changed
1144    using:
1145    
1146            Control.MLRISC.getString "viewer"
1147    
1148    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1149    course, these viewers must be in your path.
1150    
1151    The above will display the compilation unit at the level of clusters,
1152    many of which are small, boring, and un-interesting. Also setting:
1153    
1154            Control.MLRISC.getInt "cfg-graphical-view_size"
1155    
1156    will display clusters that are larger than the value set by the above.
1157    
1158    
1159    ----------------------------------------------------------------------
1160    Name: Matthias Blume
1161    Date: 2002/03/21 22:20:00 EST
1162    Tag: blume-20020321-kmp-bugfix
1163    Description:
1164    
1165    Changed the interface to the KMP routine in PreString and fixed
1166    a minor bug in one place where it was used.
1167    
1168    ----------------------------------------------------------------------
1169    Name: Allen Leung
1170    Date: 2002/03/21 20:30:00 EST
1171    Tag: leunga-20020321-cfg
1172    Description:
1173    
1174      Fixed a potential problem in cfg edge splitting.
1175    
1176    ----------------------------------------------------------------------
1177    Name: Allen Leung
1178    Date: 2002/03/21 17:15:00 EST
1179    Tag: leunga-20020321-x86-fp-cfg
1180    Description:
1181    
1182      1. Recoded the buggy parts of x86-fp.
1183    
1184         a. All the block reordering code has been removed.
1185            We now depend on the block placement phases to do this work.
1186    
1187         b. Critical edge splitting code has been simplified and moved into the
1188            CFG modules, as where they belong.
1189    
1190         Both of these were quite buggy and complex.  The code is now much, much
1191         simpler.
1192    
1193      2. X86 backend.
1194    
1195         a. Added instructions for 64-bit support.  Instruction selection for
1196            64-bit has not been committed, however, since that
1197            requires changes to MLTREE which haven't been approved by
1198            Lal and John.
1199    
1200         b. Added support for FUCOMI and FUCOMIP when generating code for
1201            PentiumPro and above.  We only generate these instructions in
1202            the fast-fp mode.
1203    
1204         c. Added cases for JP and JNP in X86FreqProps.
1205    
1206      3. CFG
1207    
1208         CFG now has a bunch of methods for edge splitting and merging.
1209    
1210      4. Machine description.
1211    
1212         John's simplification of MLTREE_BASIS.fcond broke a few machine
1213         description things:
1214    
1215         rtl-build.{sig,sml} and hppa.mdl fixed.
1216    
1217         NOTE: the machine description stuff in the repository is still broken.
1218               Again, I can't put my fixes in because that involves
1219               changes to MLTREE.
1220    
1221    ----------------------------------------------------------------------
1222    Name: Matthias Blume
1223    Date: 2002/03/20 15:55:00 EST
1224    Tag: blume-20020320-kmp
1225    Description:
1226    
1227    Implemented Knuth-Morris-Pratt string matching in PreString and used
1228    it for String.isSubstring, Substring.isSubstring, and
1229    Substring.position.
1230    
1231    (Might need some stress-testing.  Simple examples worked fine.)
1232    
1233    ----------------------------------------------------------------------
1234    Name: Matthias Blume
1235    Date: 2002/03/19 16:37:00 EST
1236    Tag: blume-20020319-witnesses
1237    Description:
1238    
1239    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1240    
1241    This implements a generic mechanism for changing constness qualifiers
1242    anywhere within big C types without resorting to outright "casts".
1243    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1244    modify the constness at the outermost level.)
1245    The implementation of "convert" is based on the idea of "witness"
1246    values -- values that are not used by the operation but whose types
1247    "testify" to their applicability.  On the implementation side, "convert"
1248    is simply a projection (returning its second curried argument).  With
1249    cross-module inlining, it should not result in any machine code being
1250    generated.
1251    
1252    ----------------------------------------------------------------------
1253    Name: Matthias Blume
1254    Date: 2002/03/15 16:40:00 EST
1255    Tag: blume-20020315-basis
1256    Description:
1257    
1258    Provided (preliminary?) implementations for
1259    
1260      {String,Substring}.{concatWith,isSuffix,isSubstring}
1261    
1262    and
1263    
1264      Substring.full
1265    
1266    Those are in the Basis spec but they were missing in SML/NJ.
1267    
1268    ----------------------------------------------------------------------
1269    Name: Matthias Blume
1270    Date: 2002/03/14 21:30:00 EST
1271    Tag: blume-20020314-controls
1272    Description:
1273    
1274    Controls:
1275    ---------
1276    
1277    1. Factored out the recently-added Controls : CONTROLS stuff and put
1278       it into its own library $/controls-lib.cm.  The source tree for
1279       this is under src/smlnj-lib/Controls.
1280    
1281    2. Changed the names of types and functions in this interface, so they
1282       make a bit more "sense":
1283    
1284          module -> registry
1285          'a registry -> 'a group
1286    
1287    3. The interface now deals in ref cells only.  The getter/setter interface
1288       is (mostly) gone.
1289    
1290    4. Added a function that lets one register an already-existing ref cell.
1291    
1292    5. Made the corresponding modifications to the rest of the code so that
1293       everything compiles again.
1294    
1295    6. Changed the implementation of Controls.MLRISC back to something closer
1296       to the original.  In particular, this module (and therefore MLRISC)
1297       does not depend on Controls.  There now is some link-time code in
1298       int-sys.sml that registers the MLRISC controls with the Controls
1299       module.
1300    
1301    CM:
1302    ---
1303    
1304      * One can now specify the lambda-split aggressiveness in init.cmi.
1305    
1306    ----------------------------------------------------------------------
1307    Name: Allen Leung
1308    Date: 2002/03/13 17:30:00 EST
1309    Tag: leunga-20020313-x86-fp-unary
1310    Description:
1311    
1312    Bug fix for:
1313    
1314    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1315    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1316    > - fun f(x,(y,z)) = Real.~ y;
1317    > [autoloading]
1318    > [autoloading done]
1319    >       fchsl   (%eax), 184(%esp)
1320    > Error: MLRisc bug: X86MCEmitter.emitInstr
1321    >
1322    > uncaught exception Error
1323    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1324    
1325    The problem was that the code generator did not generate any fp registers
1326    in this case, and the ra didn't know that it needed to run the X86FP phase to
1327    translate the pseudo fp instruction.   This only happened with unary fp
1328    operators in certain situations.
1329    
1330    ----------------------------------------------------------------------
1331    Name: Matthias Blume
1332    Date: 2002/03/13 14:00:00 EST
1333    Tag: blume-20020313-overload-etc
1334    Description:
1335    
1336    1. Added _overload as a synonym for overload for backward compatibility.
1337       (Control.overloadKW must be true for either version to be accepted.)
1338    
1339    2. Fixed bug in install script that caused more things to be installed
1340       than what was requested in config/targets.
1341    
1342    3. Made CM aware of the (_)overload construct so that autoloading
1343       works.
1344    
1345    ----------------------------------------------------------------------
1346    Name: Matthias Blume
1347    Date: 2002/03/12 22:03:00 EST
1348    Tag: blume-20020312-url
1349    Description:
1350    
1351    Forgot to update BOOT and srcarchiveurl.
1352    
1353    ----------------------------------------------------------------------
1354    Name: Matthias Blume
1355    Date: 2002/03/12 17:30:00 EST
1356    Tag: blume-20020312-version110392
1357    Description:
1358    
1359    Yet another version number bump (because of small changes to the
1360    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1361    
1362    Changes:
1363    
1364      The new pid generation scheme described a few weeks ago was overly
1365      complicated.  I implemented a new mechanism that is simpler and
1366      provides a bit more "stability":  Once CM has seen a compilation
1367      unit, it keeps its identity constant (as long as you do not delete
1368      those crucial CM/GUID/* files).  This means that when you change
1369      an interface, compile, then go back to the old interface, and
1370      compile again, you arrive at the original pid.
1371    
1372      There now also is a mechanism that instructs CM to use the plain
1373      environment hash as a module's pid (effectively making its GUID
1374      the empty string).  For this, "noguid" must be specified as an
1375      option to the .sml file in question within its .cm file.
1376      This is most useful for code that is being generated by tools such
1377      as ml-nlffigen (because during development programmers tend to
1378      erase the tool's entire output directory tree including CM's cached
1379      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1380      revert to the old, broken behavior of SML/NJ, but in specific cases
1381      where there is no danger of interface confusion, its use is ok
1382      (I think).
1383    
1384      ml-nlffigen by default generates "noguid" annotations.  They can be
1385      turned off by specifying -guid in its command line.
1386    
1387    ----------------------------------------------------------------------
1388    Name: Lal George
1389    Date: 2002/03/12 12 14:42:36 EST
1390    Tag: george-20020312-frequency-computation
1391    Description:
1392    
1393    Integrated jump chaining and static block frequency into the
1394    compiler. More details and numbers later.
1395    
1396    ----------------------------------------------------------------------
1397    Name: Lal George
1398    Date: 2002/03/11 11 22:38:53 EST
1399    Tag: george-20020311-jump-chain-elim
1400    Description:
1401    
1402    Tested the jump chain elimination on all architectures (except the
1403    hppa).  This is on by default right now and is profitable for the
1404    alpha and x86, however, it may not be profitable for the sparc and ppc
1405    when compiling the compiler.
1406    
1407    The gc test will typically jump to a label at the end of the cluster,
1408    where there is another jump to an external cluster containing the actual
1409    code to invoke gc. This is to allow factoring of common gc invocation
1410    sequences. That is to say, we generate:
1411    
1412            f:
1413               testgc
1414               ja   L1      % jump if above to L1
1415    
1416            L1:
1417               jmp L2
1418    
1419    
1420    After jump chain elimination the 'ja L1' instructions is converted to
1421    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1422    up being implemented in their long form (if L2 is far away) using:
1423    
1424            jbe     L3      % jump if below or equal to L3
1425            jmp     L2
1426         L3:
1427            ...
1428    
1429    
1430    For large compilation units L2  may be far away.
1431    
1432    
1433    ----------------------------------------------------------------------
1434    Name: Matthias Blume
1435    Date: 2002/03/11 13:30:00 EST
1436    Tag: blume-20020311-mltreeeval
1437    Description:
1438    
1439    A functor parameter was missing.
1440    
1441    ----------------------------------------------------------------------
1442    Name: Allen Leung
1443    Date: 2002/03/11 10:30:00 EST
1444    Tag: leunga-20020311-runtime-string0
1445    Description:
1446    
1447       The representation of the empty string now points to a
1448    legal null terminated C string instead of unit.  It is now possible
1449    to convert an ML string into C string with InlineT.CharVector.getData.
1450    This compiles into one single machine instruction.
1451    
1452    ----------------------------------------------------------------------
1453    Name: Allen Leung
1454    Date: 2002/03/10 23:55:00 EST
1455    Tag: leunga-20020310-x86-call
1456    Description:
1457    
1458       Added machine generation for CALL instruction (relative displacement mode)
1459    
1460    ----------------------------------------------------------------------
1461    Name: Matthias Blume
1462    Date: 2002/03/08 16:05:00
1463    Tag: blume-20020308-entrypoints
1464    Description:
1465    
1466    Version number bumped to 110.39.1.  NEW BOOTFILES!
1467    
1468    Entrypoints: non-zero offset into a code object where execution should begin.
1469    
1470    - Added the notion of an entrypoint to CodeObj.
1471    - Added reading/writing of entrypoint info to Binfile.
1472    - Made runtime system bootloader aware of entrypoints.
1473    - Use the address of the label of the first function given to mlriscGen
1474      as the entrypoint.  This address is currently always 0, but it will
1475      not be 0 once we turn on block placement.
1476    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
1477      with entry points) from mlriscGen.
1478    
1479    ----------------------------------------------------------------------
1480    Name: Allen Leung
1481    Date: 2002/03/07 20:45:00 EST
1482    Tag: leunga-20020307-x86-cmov
1483    Description:
1484    
1485       Bug fixes for CMOVcc on x86.
1486    
1487       1. Added machine code generation for CMOVcc
1488       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1489       3. CMOVcc cannot have an immediate operand as argument.
1490    
1491    ----------------------------------------------------------------------
1492    Name: Matthias Blume
1493    Date: 2002/03/07 16:15:00 EST
1494    Tag: blume-20020307-controls
1495    Description:
1496    
1497    This is a very large but mostly boring patch which makes (almost)
1498    every tuneable compiler knob (i.e., pretty much everything under
1499    Control.* plus a few other things) configurable via both the command
1500    line and environment variables in the style CM did its configuration
1501    until now.
1502    
1503    Try starting sml with '-h' (or, if you are brave, '-H')
1504    
1505    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1506    implements the underlying generic mechanism.
1507    
1508    The interface to some of the existing such facilities has changed somewhat.
1509    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1510    (The getFoo interface is still there for backward-compatibility, but its
1511    use is deprecated.)
1512    
1513    The ml-build script passes -Cxxx=yyy command-line arguments through so
1514    that one can now twiddle the compiler settings when using this "batch"
1515    compiler.
1516    
1517    TODO items:
1518    
1519    We should go through and throw out all controls that are no longer
1520    connected to anything.  Moreover, we should go through and provide
1521    meaningful (and correct!) documentation strings for those controls
1522    that still are connected.
1523    
1524    Currently, multiple calls to Controls.new are accepted (only the first
1525    has any effect).  Eventually we should make sure that every control
1526    is being made (via Controls.new) exactly once.  Future access can then
1527    be done using Controls.acc.
1528    
1529    Finally, it would probably be a good idea to use the getter-setter
1530    interface to controls rather than ref cells.  For the time being, both
1531    styles are provided by the Controls module, but getter-setter pairs are
1532    better if thread-safety is of any concern because they can be wrapped.
1533    
1534    *****************************************
1535    
1536    One bug fix: The function blockPlacement in three of the MLRISC
1537    backpatch files used to be hard-wired to one of two possibilities at
1538    link time (according to the value of the placementFlag).  But (I
1539    think) it should rather sense the flag every time.
1540    
1541    *****************************************
1542    
1543    Other assorted changes (by other people who did not supply a HISTORY entry):
1544    
1545    1. the cross-module inliner now works much better (Monnier)
1546    2. representation of weights, frequencies, and probabilities in MLRISC
1547       changed in preparation of using those for weighted block placement
1548       (Reppy, George)
1549    
1550    ----------------------------------------------------------------------
1551    Name: Lal George
1552    Date: 2002/03/07 14:44:24 EST 2002
1553    Tag: george-20020307-weighted-block-placement
1554    
1555    Tested the weighted block placement optimization on all architectures
1556    (except the hppa) using AMPL to generate the block and edge frequencies.
1557    Changes were required in the machine properties to correctly
1558    categorize trap instructions. There is an MLRISC flag
1559    "weighted-block-placement" that can be used to enable weighted block
1560    placement, but this will be ineffective without block/edge
1561    frequencies (coming soon).
1562    
1563    
1564    ----------------------------------------------------------------------
1565    Name: Lal George
1566    Date: 2002/03/05 17:24:48 EST
1567    Tag: george-20020305-linkage-cluster
1568    
1569    In order to support the block placement optimization, a new cluster
1570    is generated as the very first cluster (called the linkage cluster).
1571    It contains a single jump to the 'real' entry point for the compilation
1572    unit. Block placement has no effect on the linkage cluster itself, but
1573    all the other clusters  have full freedom in the manner in which they
1574    reorder blocks or functions.
1575    
1576    On the x86 the typical linkage code that is generated is:
1577       ----------------------
1578            .align 2
1579       L0:
1580            addl    $L1-L0, 72(%esp)
1581            jmp     L1
1582    
1583    
1584            .align  2
1585       L1:
1586       ----------------------
1587    
1588    72(%esp) is the memory location for the stdlink register. This
1589    must contain the address of the CPS function being called. In the
1590    above example, it contains the address of  L0; before
1591    calling L1 (the real entry point for the compilation unit), it
1592    must contain the address for L1, and hence
1593    
1594            addl $L1-L0, 72(%esp)
1595    
1596    I have tested this on all architectures except the hppa.The increase
1597    in code size is of course negligible
1598    
1599    ----------------------------------------------------------------------
1600    Name: Allen Leung
1601    Date: 2002/03/03 13:20:00 EST
1602    Tag: leunga-20020303-mlrisc-tools
1603    
1604      Added #[ ... ] expressions to mlrisc tools
1605    
1606    ----------------------------------------------------------------------
1607    Name: Matthias Blume
1608    Date: 2002/02/27 12:29:00 EST
1609    Tag: blume-20020227-cdebug
1610    Description:
1611    
1612    - made types in structure C and C_Debug to be equal
1613    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1614    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1615    
1616    ----------------------------------------------------------------------
1617    Name: Matthias Blume
1618    Date: 2002/02/26 12:00:00 EST
1619    Tag: blume-20020226-ffi
1620    Description:
1621    
1622    1. Fixed a minor bug in CM's "noweb" tool:
1623       If numbering is turned off, then truly don't number (i.e., do not
1624       supply the -L option to noweb).  The previous behavior was to supply
1625       -L'' -- which caused noweb to use the "default" line numbering scheme.
1626       Thanks to Chris Richards for pointing this out (and supplying the fix).
1627    
1628    2. Once again, I reworked some aspects of the FFI:
1629    
1630       A. The incomplete/complete type business:
1631    
1632       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
1633         gone!
1634       - ML types representing an incomplete type are now *equal* to
1635         ML types representing their corresponding complete types (just like
1636         in C).  This is still safe because ml-nlffigen will not generate
1637         RTTI for incomplete types, nor will it generate functions that
1638         require access to such RTTI.   But when ML code generated from both
1639         incomplete and complete versions of the C type meet, the ML types
1640         are trivially interoperable.
1641    
1642         NOTE:  These changes restore the full generality of the translation
1643         (which was previously lost when I eliminated functorization)!
1644    
1645       B. Enum types:
1646    
1647       - Structure C now has a type constructor "enum" that is similar to
1648         how the "su" constructor works.  However, "enum" is not a phantom
1649         type because each "T enum" has values (and is isomorphic to
1650         MLRep.Signed.int).
1651       - There are generic access operations for enum objects (using
1652         MLRep.Signed.int).
1653       - ml-nlffigen will generate a structure E_foo for each "enum foo".
1654         * The structure contains the definition of type "mlrep" (the ML-side
1655         representation type of the enum).  Normally, mlrep is the same
1656         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
1657         then mlrep will be defined as a datatype -- thus facilitating
1658         pattern matching on mlrep values.
1659         ("-ec" will be suppressed if there are duplicate values in an
1660          enumeration.)
1661         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
1662         will be generated for each C enum constant xxx.
1663         * Conversion functions m2i and i2m convert between mlrep and
1664         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
1665         * Coversion functions c and ml convert between mlrep and "tag enum".
1666         * Access functions (get/set) fetch and store mlrep values.
1667       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
1668         enumerations are merged into one single enumeration represented by
1669         structure E_'.
1670    
1671    ----------------------------------------------------------------------
1672    Name: Allen Leung
1673    Date: 2002/02/25 04:45:00 EST
1674    Tag: leunga-20020225-cps-spill
1675    
1676    This is a new implementation of the CPS spill phase.
1677    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
1678    In case of problems, replace it with the old file spill.sml
1679    
1680    The current compiler runs into some serious performance problems when
1681    constructing a large record.  This can happen when we try to compile a
1682    structure with many items.  Even a very simple structure like the following
1683    makes the compiler slow down.
1684    
1685        structure Foo = struct
1686           val x_1 = 0w1 : Word32.int
1687           val x_2 = 0w2 : Word32.int
1688           val x_3 = 0w3 : Word32.int
1689           ...
1690           val x_N = 0wN : Word32.int
1691        end
1692    
1693    The following table shows the compile time, from N=1000 to N=4000,
1694    with the old compiler:
1695    
1696    N
1697    1000   CPS 100 spill                           0.04u  0.00s  0.00g
1698           MLRISC ra                               0.06u  0.00s  0.05g
1699              (spills = 0 reloads = 0)
1700           TOTAL                                   0.63u  0.07s  0.21g
1701    
1702    1100   CPS 100 spill                           8.25u  0.32s  0.64g
1703           MLRISC ra                               5.68u  0.59s  3.93g
1704              (spills = 0 reloads = 0)
1705           TOTAL                                   14.71u  0.99s  4.81g
1706    
1707    1500   CPS 100 spill                           58.55u  2.34s  1.74g
1708           MLRISC ra                               5.54u  0.65s  3.91g
1709              (spills = 543 reloads = 1082)
1710           TOTAL                                   65.40u  3.13s  6.00g
1711    
1712    2000   CPS 100 spill                           126.69u  4.84s  3.08g
1713           MLRISC ra                               0.80u  0.10s  0.55g
1714              (spills = 42 reloads = 84)
1715           TOTAL                                   129.42u  5.10s  4.13g
1716    
1717    3000   CPS 100 spill                           675.59u  19.03s  11.64g
1718           MLRISC ra                               2.69u  0.27s  1.38g
1719              (spills = 62 reloads = 124)
1720           TOTAL                                   682.48u  19.61s  13.99g
1721    
1722    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
1723           MLRISC ra                               4.96u  0.27s  2.72g
1724              (spills = 85 reloads = 170)
1725           TOTAL                                   2375.26u  57.21s  48.00g
1726    
1727    As you can see the old cps spill module suffers from some serious
1728    performance problem.  But since I cannot decipher the old code fully,
1729    instead of patching the problems up, I'm reimplementing it
1730    with a different algorithm.  The new code is more modular,
1731    smaller when compiled, and substantially faster
1732    (O(n log n) time and O(n) space).  Timing of the new spill module:
1733    
1734    4000  CPS 100 spill                           0.02u  0.00s  0.00g
1735          MLRISC ra                               0.25u  0.02s  0.15g
1736             (spills=1 reloads=3)
1737          TOTAL                                   7.74u  0.34s  1.62g
1738    
1739    Implementation details:
1740    
1741    As far as I can tell, the purpose of the CPS spill module is to make sure the
1742    number of live variables at any program point (the bandwidth)
1743    does not exceed a certain limit, which is determined by the
1744    size of the spill area.
1745    
1746    When the bandwidth is too large, we decrease the register pressure by
1747    packing live variables into spill records.  How we achieve this is
1748    completely different than what we did in the old code.
1749    
1750    First, there is something about the MLRiscGen code generator
1751    that we should be aware of:
1752    
1753    o MLRiscGen performs code motion!
1754    
1755       In particular, it will move floating point computations and
1756       address computations involving only the heap pointer to
1757       their use sites (if there is only a single use).
1758       What this means is that if we have a CPS record construction
1759       statement
1760    
1761           RECORD(k,vl,w,e)
1762    
1763       we should never count the new record address w as live if w
1764       has only one use (which is often the case).
1765    
1766       We should do something similar to floating point, but the transformation
1767       there is much more complex, so I won't deal with that.
1768    
1769    Secondly, there are now two new cps primops at our disposal:
1770    
1771     1. rawrecord of record_kind option
1772        This pure operator allocates some uninitialized storage from the heap.
1773        There are two forms:
1774    
1775         rawrecord NONE [INT n]  allocates a tagless record of length n
1776         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
1777                                     and initializes the tag.
1778    
1779     2. rawupdate of cty
1780          rawupdate cty (v,i,x)
1781          Assigns to x to the ith component of record v.
1782          The storelist is not updated.
1783    
1784    We use these new primops for both spilling and increment record construction.
1785    
1786     1. Spilling.
1787    
1788        This is implemented with a linear scan algorithm (but generalized
1789        to trees).  The algorithm will create a single spill record at the
1790        beginning of the cps function and use rawupdate to spill to it,
1791        and SELECT or SELp to reload from it.  So both spills and reloads
1792        are fine-grain operations.  In contrast, in the old algorithm
1793        "spills" have to be bundled together in records.
1794    
1795        Ideally, we should sink the spill record construction to where
1796        it is needed.  We can even split the spill record into multiple ones
1797        at the places where they are needed.  But CPS is not a good
1798        representation for global code motion, so I'll keep it simple and
1799        am not attempting this.
1800    
1801     2. Incremental record construction (aka record splitting).
1802    
1803        Long records with many component values which are simulatenously live
1804        (recall that single use record addresses are not considered to
1805         be live) are constructed with rawrecord and rawupdate.
1806        We allocate space on the heap with rawrecord first, then gradually
1807        fill it in with rawupdate.  This is the technique suggested to me
1808        by Matthias.
1809    
1810        Some restrictions on when this is applicable:
1811        1. It is not a VECTOR record.  The code generator currently does not handle
1812           this case. VECTOR record uses double indirection like arrays.
1813        2. All the record component values are defined in the same "basic block"
1814           as the record constructor.  This is to prevent speculative
1815           record construction.
1816    
1817    ----------------------------------------------------------------------
1818    Name: Allen Leung
1819    Date: 2002/02/22 01:02:00 EST
1820    Tag: leunga-20020222-mlrisc-tools
1821    
1822    Minor bug fixes in the parser and rewriter
1823    
1824    ----------------------------------------------------------------------
1825    Name: Allen Leung
1826    Date: 2002/02/21 20:20:00 EST
1827    Tag: leunga-20020221-peephole
1828    
1829    Regenerated the peephole files.  Some contained typos in the specification
1830    and some didn't compile because of pretty printing bugs in the old version
1831    of 'nowhere'.
1832    
1833    ----------------------------------------------------------------------
1834    Name: Allen Leung
1835    Date: 2002/02/19 20:20:00 EST
1836    Tag: leunga-20020219-mlrisc-tools
1837    Description:
1838    
1839       Minor bug fixes to the mlrisc-tools library:
1840    
1841       1.  Fixed up parsing colon suffixed keywords
1842       2.  Added the ability to shut the error messages up
1843       3.  Reimplemented the pretty printer and fixed up/improved
1844           the pretty printing of handle and -> types.
1845       4.  Fixed up generation of literal symbols in the nowhere tool.
1846       5.  Added some SML keywords to to sml.sty
1847    
1848    ----------------------------------------------------------------------
1849    Name: Matthias Blume
1850    Date: 2002/02/19 16:20:00 EST
1851    Tag: blume-20020219-cmffi
1852    Description:
1853    
1854    A wild mix of changes, some minor, some major:
1855    
1856    * All C FFI-related libraries are now anchored under $c:
1857        $/c.cm      --> $c/c.cm
1858        $/c-int.cm  --> $c/internals/c-int.cm
1859        $/memory.cm --> $c/memory/memory.cm
1860    
1861    * "make" tool (in CM) now treats its argument pathname slightly
1862      differently:
1863        1. If the native expansion is an absolute name, then before invoking
1864           the "make" command on it, CM will apply OS.Path.mkRelative
1865           (with relativeTo = OS.FileSys.getDir()) to it.
1866        2. The argument will be passed through to subsequent phases of CM
1867           processing without "going native".  In particular, if the argument
1868           was an anchored path, then "make" will not lose track of that anchor.
1869    
1870    * Compiler backends now "know" their respective C calling conventions
1871      instead of having to be told about it by ml-nlffigen.  This relieves
1872      ml-nlffigen from one of its burdens.
1873    
1874    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
1875    
1876    * Export C_DEBUG and C_Debug from $c/c.cm.
1877    
1878    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
1879      subtyping relationship between incomplete pointers and their complete
1880      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
1881      with the convention of instantiating 'o with ('t, 'c) obj whenever
1882      the pointer target type is complete.  In the incomplete case, 'o
1883      will be instantiated with some "'c iobj" -- a type obtained by
1884      using one of the functors PointerToIncompleteType or PointerToCompleteType.
1885    
1886      Operations that work on both incomplete and complete pointer types are
1887      typed as taking an 'o ptr while operations that require the target to
1888      be known are typed as taking some ('t, 'c) obj ptr.
1889    
1890      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
1891      where void is an eqtype without any values.  This makes it possible
1892      to work on voidptr values using functions meant to operate on light
1893      incomplete pointers.
1894    
1895    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
1896      been vastly simplified.
1897    
1898    ----------------------------------------------------------------------
1899    Name: Matthias Blume
1900    Date: 2002/02/19 10:48:00 EST
1901    Tag: blume-20020219-pqfix
1902    Description:
1903    
1904    Applied Chris Okasaki's bug fix for priority queues.
1905    
1906    ----------------------------------------------------------------------
1907    Name: Matthias Blume
1908    Date: 2002/02/15 17:05:00
1909    Tag: Release_110_39
1910    Description:
1911    
1912    Last-minute retagging is becoming a tradition... :-(
1913    
1914    This is the working release 110.39.
1915    
1916    ----------------------------------------------------------------------
1917    Name: Matthias Blume
1918    Date: 2002/02/15 16:00:00 EST
1919    Tag: Release_110_39-orig
1920    Description:
1921    
1922    Working release 110.39.  New bootfiles.
1923    
1924    (Update: There was a small bug in the installer so it wouldn't work
1925    with all shells.  So I retagged. -Matthias)
1926    
1927    ----------------------------------------------------------------------
1928    Name: Matthias Blume
1929    Date: 2002/02/15 14:17:00 EST
1930    Tag: blume-20020215-showbindings
1931    Description:
1932    
1933    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
1934    the latter can be useful for exploring what bindings are available at
1935    the interactive prompt.  (The first function returns only the list
1936    of symbols that are really bound, the second prints those but also the
1937    ones that CM's autoloading mechanism knows about.)
1938    
1939    ----------------------------------------------------------------------
1940    Name: Matthias Blume
1941    Date: 2002/02/15 12:08:00 EST
1942    Tag: blume-20020215-iptrs
1943    Description:
1944    
1945    Two improvements to ml-nlffigen:
1946    
1947      1. Write files only if they do not exist or if their current contents
1948         do not coincide with what's being written.  (That is, avoid messing
1949         with the time stamps unless absolutely necessary.)
1950    
1951      2. Implement a "repository" mechanism for generated files related
1952         to "incomplete pointer types".   See the README file for details.
1953    
1954    ----------------------------------------------------------------------
1955    Name: Matthias Blume
1956    Date: 2002/02/14 11:50:00 EST
1957    Tag: blume-20020214-quote
1958    Description:
1959    
1960    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
1961    because of the new and improved tag generation scheme.  (Thanks to Allen
1962    Leung for pointing it out.)
1963    
1964    ----------------------------------------------------------------------
1965    Name: Lal George
1966    Date: 2002/02/14 09:55:27 EST 2002
1967    Tag: george-20020214-isabelle-bug
1968    Description:
1969    
1970    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
1971    of Isabelle on the x86.
1972    
1973    From Allen:
1974    -----------
1975     I've found the problem:
1976    
1977         in ra-core.sml, I use the counter "blocked" to keep track of the
1978         true number of elements in the freeze queue.  When the counter goes
1979         to zero, I skip examining the queue.  But I've messed up the
1980         bookkeeping in combine():
1981    
1982             else ();
1983             case !ucol of
1984               PSEUDO => (if !cntv > 0 then
1985                     (if !cntu > 0 then blocked := !blocked - 1 else ();
1986                                        ^^^^^^^^^^^^^^^^^^^^^^^
1987                      moveu := mergeMoveList(!movev, !moveu)
1988                     )
1989                  else ();
1990    
1991         combine() is called to coalesce two nodes u and v.
1992         I think I was thinking that if the move counts of u and v are both
1993         greater than zero then after they are coalesced then one node is
1994         removed from the freeze queue.  Apparently I was thinking that
1995         both u and v are of low degree, but that's clearly not necessarily true.
1996    
1997    
1998    02/12/2002:
1999        Here's the patch.  HOL now compiles.
2000    
2001        I don't know how this impact on performance (compile
2002        time or runtime).  This bug caused the RA (especially on the x86)
2003        to go thru the potential spill phase when there are still nodes on the
2004        freeze queue.
2005    
2006    
2007    
2008    
2009    ----------------------------------------------------------------------
2010    Name: Matthias Blume
2011    Date: 2002/02/13 22:40:00 EST
2012    Tag: blume-20020213-fptr-rtti
2013    Description:
2014    
2015    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2016    updates.
2017    
2018    ----------------------------------------------------------------------
2019    Name: Matthias Blume
2020    Date: 2002/02/13 16:41:00 EST
2021    Tag: blume-20020213-cmlpq
2022    Description:
2023    
2024    Added new priority queue export symbols (which have just been added to
2025    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2026    not compile and the installer would choke.)
2027    
2028    ----------------------------------------------------------------------
2029    Name: Matthias Blume
2030    Date: 2002/02/13 16:15:00 EST
2031    Tag: blume-20020213-various
2032    Description:
2033    
2034    1. More tweaks to ml-nlffigen:
2035    
2036       - better internal datastructures (resulting in slight speedup)
2037       - "-match" option requires exact match
2038       - "localized" gensym counters (untagged structs/unions nested within
2039         other structs/unions or within typedefs get a fresh counter; their
2040         tag will be prefixed by a concatenation of their parents' tags)
2041       - bug fixes (related to calculation of transitive closure of types
2042         to be included in the output)
2043    
2044    2. Minor Basis updates:
2045    
2046       - added implementations for List.collate and Option.app
2047    
2048    ----------------------------------------------------------------------
2049    Name: Matthias Blume
2050    Date: 2002/02/11 15:55:00 EST
2051    Tag: blume-20020211-gensym
2052    Description:
2053    
2054    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2055    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2056    names (ML structure names that correspond to unnamed C structs, unions,
2057    and enums), so that separate runs of ml-nlffigen do not clash.
2058    
2059    ----------------------------------------------------------------------
2060    Name: Matthias Blume
2061    Date: 2002/02/11 12:05:00 EST
2062    Tag: blume-20020211-gensml
2063    Description:
2064    
2065    A quick fix for a problem with GenSML (in the pgraph-util library):
2066    Make generation of toplevel "local" optional.  (Strictly speaking,
2067    signature definitions within "local" are not legal SML.)
2068    
2069    Other than that: updates to INSTALL and cm/TODO.
2070    
2071    ----------------------------------------------------------------------
2072    Name: Matthias Blume
2073    Date: 2002/02/08 15:00:00 EST
2074    Tag: blume-20020208-uniquepid
2075    Description:
2076    
2077    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2078    
2079    1. The installer (config/install.sh) has gotten smarter:
2080    
2081         - Configuration options are a bit easier to specify now
2082           (in config/targets).
2083         - Bug in recognizing .tar.bz2 files fixed.
2084         - Installer automatically resolves dependencies between
2085           configuration options (e.g., if you ask for eXene, you will
2086           also get cml -- regardless whether you asked for it or not).
2087         - Installer can run in "quieter mode" by setting the environment
2088           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2089           "completely silent", though.
2090         - Build HashCons library as part of smlnj-lib.
2091    
2092    2. A new scheme for assigning persistent identifiers to compilation
2093       units (and, by extension, to types etc.) has been put into place.
2094       This fixes a long-standing bug where types and even dynamic values
2095       can get internally confused, thereby compromising type safety
2096       (abstraction) and dynamic correctness.  See
2097    
2098         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2099    
2100       for an example of how things could go wrong until now.
2101    
2102       The downside of the new scheme is that pids are not quite as
2103       persistent as they used to be: CM will generate a fresh pid
2104       for every compilation unit that it thinks it sees for the first
2105       time.  That means that if you compile starting from a clean, fresh
2106       source tree at two different times, you end up with different
2107       binaries.
2108    
2109       Cutoff recompilation, however, has not been compromised because
2110       CM keeps pid information in special caches between runs.
2111    
2112    ----------------------------------------------------------------------
2113    Name: Lal George
2114    Date: 2002/02/07 15:34:13 EST 2002
2115    Tag: <none>
2116    Description:
2117    
2118    Compilers that generate assembly code may produce  global labels
2119    whose value is resolved at link time. The various peephole optimization
2120    modules did not take this in account.
2121    
2122    TODO. The Labels.addrOf function should really return an option
2123    type so that clients are forced to deal with this issue, rather
2124    than an exception being raised.
2125    
2126    ----------------------------------------------------------------------
2127    Name: Lal George
2128    Date: 2002/02/06 13:55:02 EST
2129    Tag: george-20020206-ra-breakup
2130    Description:
2131    
2132    1. A bug fix from Allen.
2133    
2134        A typo causes extra fstp %st(0)'s to be generated at compensation
2135        edges, which might cause stack underflow traps at runtime.  This
2136        occurs in fft where there are extraneous fstps right before the 'into'
2137        trap instruction (in this case they are harmless since none of the
2138        integers overflow.)
2139    
2140    2. Pulled out various utility modules that were embedded in the modules
2141       of the register allocator. I need these modules for other purposes, but
2142       they are not complete enough to put into a library (just yet).
2143    ----------------------------------------------------------------------
2144    Name: Matthias Blume
2145    Date: 2002/01/31 16:05:00 EST
2146    Tag: blume-20020131-sparc-ccalls
2147    Description:
2148    
2149    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2150       of extra stack space by mistake.  Fixed.
2151    
2152    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2153    
2154    ----------------------------------------------------------------------
2155    Name: Allen Leung
2156    Date: 2002/01/30
2157    Tag: leunga-20020130-nowhere-bug-fix
2158    Description:
2159    
2160       MLRISC bug fixes:
2161       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2162       2. MachineInt.fromString was negating its value.
2163    
2164    ----------------------------------------------------------------------
2165    Name: Matthias Blume
2166    Date: 2002/01/29
2167    Tag: blume-20020129-INSTALL
2168    Description:
2169    
2170    - Added somewhat detailed installation instructions (file INSTALL).
2171    - Fixed curl-detection bug in config/install.sh.
2172    - It is now possible to select the URL getter using the URLGETTER
2173      environment variable:
2174    
2175          not set / "unknown"      --> automatic detection (script tries wget,
2176                                       curl, and lynx)
2177          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2178                                       how to properly invoke them)
2179          other                    --> use $URLGETTER directly, it must take
2180                                       precisely two command-line arguments
2181                                       (source URL and destination file name)
2182    
2183    ----------------------------------------------------------------------
2184    Name: Matthias Blume
2185    Date: 2002/01/28
2186    Tag: blume-20020128-sparc-ccalls
2187    Description:
2188    
2189    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2190    - Make use of the allocParam argument in sparc-c-calls.
2191    
2192    ----------------------------------------------------------------------
2193    Name: Matthias Blume
2194    Date: 2002/01/28
2195    Tag: blume-20020128-allocParam
2196    Description:
2197    
2198    John Reppy:  Changes c-calls API to accept client-callback for
2199    allocating extra stack space.
2200    me: Corresponding changes to mlriscGen (using a dummy argument that
2201        does not change the current behavior).
2202    
2203    ----------------------------------------------------------------------
2204    Name: Matthias Blume
2205    Date: 2002/01/28 12:00:00
2206    Tag: Release_110_38
2207    Description:
2208    
2209    This time for real!!!
2210    
2211    ----------------------------------------------------------------------
2212    Name: Matthias Blume
2213    Date: 2002/01/28 10:56:00 EST
2214    Tag: blume-20020128-retraction
2215    Description:
2216    
2217    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
2218       with blume-Release_110_38-retracted.)
2219    
2220    1. Fixed a problem with incorrect rounding modes in real64.sml.
2221       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2222    
2223    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2224       and enums fixed.  The naming of corresponding ML identifiers should
2225       now be consistent again.
2226    
2227    ----------------------------------------------------------------------
2228    Name: Allen Leung
2229    Date: 2002/01/27
2230    Tag: leunga-20020127-nowhere
2231    Description:
2232    
2233       Added a target called nowhere in the configuration scripts.
2234       Enabling this will build the MLRISC 'nowhere' tool (for translating
2235       programs with where-clauses into legal SML code) during installation.
2236    
2237    ----------------------------------------------------------------------
2238    Name: Matthias Blume
2239    Date: 2002/01/25 21:27:00 EST
2240    Tag: blume-Release_110_38-retracted
2241    Description:
2242    
2243    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
2244    
2245    README will be added later.
2246    
2247    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
2248    (Release_110_38) removed.  Reason: Last-minute bug fixes.
2249    
2250    ----------------------------------------------------------------------
2251    Name: Matthias Blume
2252    Date: 2002/01/25
2253    Tag: blume-20020125-ffi
2254    Description:
2255    
2256    A large number of tweaks and improvements to ml-nlffi-lib and
2257    ml-nlffigen:
2258    
2259       - ML represenation types have been streamlined
2260       - getter and setter functions work with concrete values, not abstract
2261         ones where possible
2262       - ml-nlffigen command line more flexible (see README file there)
2263       - some bugs have been fixed (hopefully)
2264    
2265    ----------------------------------------------------------------------
2266    Name: Lal George
2267    Date: 2002/01/24
2268    Tag: george-20020124-risc-ra-interface
2269    Description:
2270    
2271       There is a dramatic simplification in the interface to the
2272       register allocator for RISC architectures as a result of making
2273       parallel copy instructions explicit.
2274    
2275    ----------------------------------------------------------------------
2276    Name: Matthias Blume
2277    Date: 2002/01/22
2278    Tag: blume-20020122-x86-ccalls
2279    Description:
2280    
2281    Bug fix for c-calls on x86 (having to do with how char- and
2282    short-arguments are being handled).
2283    
2284    ----------------------------------------------------------------------
2285    Name: Matthias Blume
2286    Date: 2002/01/21
2287    Tag: blume-20020121-ff
2288    Description:
2289    
2290    Another day of fiddling with the FFI...
2291    
2292    1. Bug fix/workaround:  CKIT does not complain about negative array
2293       dimensions, so ml-nlffigen has to guard itself against this possibility.
2294       (Otherwise a negative dimension would send it into an infinite loop.)
2295    
2296    2. Some of the abstract types (light objects, light pointers, most "base"
2297       types) in structure C are now eqtypes.
2298    
2299    3. Added constructors and test functions for NULL function pointers.
2300    
2301    ----------------------------------------------------------------------
2302    Name: Matthias Blume
2303    Date: 2002/01/18
2304    Tag: blume-20020118-ready-for-new-release
2305    Description:
2306    
2307    Made config/srcarchiveurl point to a new place.  (Will provide boot
2308    files shortly.)
2309    
2310    Maybe we christen this to be 110.38?
2311    
2312    ----------------------------------------------------------------------
2313    Name: Matthias Blume
2314    Date: 2002/01/18
2315    Tag: blume-20020118-more-ffifiddle
2316    Description:
2317    
2318    Today's FFI fiddling:
2319    
2320      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
2321        C.Set.* that use concrete (MLRep.*) arguments and results instead
2322        of abstract ones.
2323    
2324      - Provided word-style bit operations etc. for "int" representation
2325        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
2326        and Long.
2327    
2328    ----------------------------------------------------------------------
2329    Name: Matthias Blume
2330    Date: 2002/01/18
2331    Tag: blume-20020118-use-x86-fp
2332    Description:
2333    
2334    Now that x86-fast-fp seems to be working, I turned it back on again
2335    by default.  (Seems to work fine now, even with the FFI.)
2336    
2337    Other than that, I added some documentation about the FFI to
2338    src/ml-nlffigen/README and updated the FFI test examples in
2339    src/ml-nlffi-lib/Tests/*.
2340    
2341    ----------------------------------------------------------------------
2342    Name: Allen Leung
2343    Date: 2002/01/17
2344    Tag: leunga-20020117-x86-fast-fp-call
2345    Description:
2346    
2347       1. Fixed a problem with handling return fp values when x86's fast fp
2348          mode is turned on.
2349    
2350       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
2351          of %f32.
2352    
2353       3. Added a constructor INT32lit to the ast of MLRISC tools.
2354    
2355    ----------------------------------------------------------------------
2356    Name: Matthias Blume
2357    Date: 2002/01/16
2358    Tag: blume-20020116-ffifiddle
2359    Description:
2360    
2361    More fiddling with the FFI interface:
2362    
2363     - Make constness 'c instead of rw wherever possible.  This eliminates
2364       the need for certain explicit coercions.  (However, due to ML's
2365       value polymorphism, there will still be many cases where explicit
2366       coercions are necessary.  Phantom types are not the whole answer
2367       to modeling a subtyping relationship in ML.)
2368    
2369     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
2370    
2371     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
2372       clumsy workaround for varargs problem.)
2373    
2374    ----------------------------------------------------------------------
2375    Name: Lal George
2376    Date: 2002/01/15
2377    Tag: <none>
2378    Description:
2379    
2380    1. Since COPY instructions are no longer native to the architecture,
2381       a generic functor can be used to implement the expandCopies function.
2382    
2383    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
2384       TEXT segment.
2385    
2386    ----------------------------------------------------------------------
2387    Name: Matthias Blume
2388    Date: 2002/01/15
2389    Tag: blume-20020115-ffiupdates
2390    Description:
2391    
2392    1. Fix for bug resulting in single-precision float values being returned
2393       incorrectly from FFI calls.
2394    
2395    2. Small modifications to C FFI API:
2396    
2397        - memory-allocation routines return straight objects (no options)
2398          and raise an exception in out-of-memory situations
2399        - unsafe extensions to cast between function pointers and pointers
2400          from/to ints
2401        - added structure C_Debug as an alternative to structure C where
2402          pointer-dereferencing (|*| and |*!) always check for null-pointers
2403        - added open_lib' to DynLinkage;  open_lib' works like open_lib
2404          but also takes a (possibly empty) list of existing library handles
2405          that the current library depends on
2406    
2407    ----------------------------------------------------------------------
2408    Name: Matthias Blume
2409    Date: 2002/01/10
2410    Tag: blume-20020110-newffigen
2411    Description:
2412    
2413    1. Updates to portable graph code.
2414    
2415    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
2416       more scalable now so that even huge interfaces such as the one
2417       for GTK compile in finite time and space. :-)
2418       See src/ml-nlffigen/README for details on what's new.
2419    
2420    ----------------------------------------------------------------------
2421    Name: Lal George
2422    Date: 2001/01/09 14:31:35 EST 2002
2423    Tag: george-20011206-rm-native-copy
2424    Description:
2425    
2426            Removed the native COPY and FCOPY instructions
2427            from all the architectures and replaced it with the
2428            explicit COPY instruction from the previous commit.
2429    
2430            It is now possible to simplify many of the optimizations
2431            modules that manipulate copies. This has not been
2432            done in this change.
2433    
2434    ----------------------------------------------------------------------
2435    Name: Lal George
2436    Date: 2001/12/06 16:50:13 EST 2001
2437    Tag: george-20011206-mlrisc-instruction
2438    Description:
2439    
2440    Changed the representation of instructions from being fully abstract
2441    to being partially concrete. That is to say:
2442    
2443      from
2444            type instruction
2445    
2446      to
2447            type instr                              (* machine instruction *)
2448    
2449            datatype instruction =
2450                LIVE of {regs: C.cellset, spilled: C.cellset}
2451              | KILL of {regs: C.cellset, spilled: C.cellset}
2452              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
2453              | ANNOTATION of {i: instruction, a: Annotations.annotation}
2454              | INSTR of instr
2455    
2456    This makes the handling of certain special instructions that appear on
2457    all architectures easier and uniform.
2458    
2459    LIVE and KILL say that a list of registers are live or killed at the
2460    program point where they appear. No spill code is generated when an
2461    element of the 'regs' field is spilled, but the register is moved to
2462    the 'spilled' (which is present, more for debugging than anything else).
2463    
2464    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
2465    We used to generate:
2466    
2467            DEFFREG f1
2468            f1 := f2 + f3
2469            trapb
2470    
2471    but now generate:
2472    
2473            f1 := f2 + f3
2474            trapb
2475            LIVE {regs=[f1,f2,f3], spilled=[]}
2476    
2477    Furthermore, the DEFFREG (hack) required that all floating point instruction
2478    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
2479    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
2480    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
2481    this architecture).
2482    
2483    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
2484    all the architectures. This will result in further simplification of the
2485    register allocator that must be aware of them for coalescing purposes, and
2486    will also simplify certain aspects of the machine description that provides
2487    callbacks related to parallel copies.
2488    
2489    ANNOTATION should be obvious, and now INSTR represents the honest to God
2490    machine instruction set!
2491    
2492    The <arch>/instructions/<arch>Instr.sml files define certain utility
2493    functions for making porting easier -- essentially converting upper case
2494    to lower case. All machine instructions (of type instr) are in upper case,
2495    and the lower case form generates an MLRISC instruction. For example on
2496    the alpha we have:
2497    
2498      datatype instr =
2499         LDA of {r:cell, b:cell, d:operand}
2500       | ...
2501    
2502      val lda : {r:cell, b:cell, d:operand} -> instruction
2503        ...
2504    
2505    where lda is just (INSTR o LDA), etc.
2506    
2507    ----------------------------------------------------------------------
2508    Name: Matthias Blume
2509    Date: 2001/11/22 21:40:00 EST
2510    Tag: Release_110_37
2511    Description:
2512    
2513    Release 110.37.  This time for real.
2514    
2515    ----------------------------------------------------------------------
2516    Name: Matthias Blume
2517    Date: 2001/11/21 16:35:00 EST
2518    Tag: blume-20011121-foot-in-mouth
2519    Description:
2520    
2521    Removed the "Release_110_37" tag because of a serious bug.
2522    This will be re-tagged once the bug is fixed.
2523    
2524    ----------------------------------------------------------------------
2525    Name: Matthias Blume
2526    Date: 2001/11/21 16:14:00 EST
2527    Tag: blume-20011121-forgottenfile
2528    Description:
2529    
2530    Forgot to add a file.  (Just a .tex-file -- part of
2531    the CM manual source.)
2532    
2533    ----------------------------------------------------------------------
2534    Name: Matthias Blume
2535    Date: 2001/11/21 16:10:00 EST
2536    Tag: blume-20011121-invalid_110_37
2537    Description:
2538    
2539    Note: I removed the original tag "Release_110_37" from this commit
2540          because we found a serious bug in all non-x86 backends.
2541          - Matthias
2542    
2543    1. Modifications to the SML/NJ code generator and to the runtime system
2544       so that code object name strings are directly inserted into code
2545       objects at code generation time.  The only business the runtime system
2546       has with this is now to read the name strings on occasions.
2547       (The encoding of the name string has also changed somewhat.)
2548    
2549    2. CM now implements a simple "set calculus" for specifying export lists.
2550       In particular, it is now possible to refer to the export lists of
2551       other libraries/groups/sources and form unions as well as differences.
2552       See the latest CM manual for details.
2553    
2554    3. An separate notion of "proxy" libraries has again be eliminated from
2555       CM's model.  (Proxy libraries are now simply a special case of using
2556       the export list calculus.)
2557    
2558    4. Some of the existing libraries now take advantage of the new set
2559       calculus.
2560       (Notice that not all libraries have been converted because some
2561       of the existing .cm-files are supposed to be backward compatible
2562       with 110.0.x.)
2563    
2564    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
2565       "General.exnMessage"!  The former relies on a certain hook to be
2566       initialized, and that often does not happen in the stand-alone case.)
2567    
2568    ----------------------------------------------------------------------
2569    Name: Lal George
2570    Date: 2001/11/21  13:56:18 EST
2571    Tag: george-2001121-pseudo-ops
2572    Description:
2573    
2574      Implemented a complete redesign of MLRISC pseudo-ops. Now there
2575      ought to never be any question of incompatabilities with
2576      pseudo-op syntax expected by host assemblers.
2577    
2578      For now, only modules supporting GAS syntax are implemented
2579      but more should follow, such as MASM, and vendor assembler
2580      syntax, e.g. IBM as, Sun as, etc.
2581    
2582    ----------------------------------------------------------------------
2583    Name: Matthias Blume
2584    Date: 2001/11/14 11:52:00 EST
2585    Tag: blume-20011114-srcname
2586    Description:
2587    
2588    1. Routed the name of the current source file to mlriscgen where it
2589       should be directly emitted into the code object.  (This last part
2590       is yet to be done.)
2591    
2592    2. Some cleanup of the pgraph code to make it match the proposal that
2593       I put out the other day.  (The proposal notwithstanding, things are
2594       still in flux here.)
2595    
2596    ----------------------------------------------------------------------
2597    Name: Lal George
2598    Date: 2001/11/14 09:44:04 EST
2599    Tag:
2600    Description:
2601    
2602      Fix for a backpatching bug reported by Allen.
2603    
2604      Because the boundary between short and long span-dependent
2605      instructions is +/- 128, there are an astounding number of
2606      span-dependent instructions whose size is over estimated.
2607    
2608      Allen came up with the idea of letting the size of span
2609      dependent instructions be non-monotonic, for a maxIter
2610      number of times, after which the size must be monotonically
2611      increasing.
2612    
2613      This table shows the number of span-dependent instructions
2614      whose size was over-estimated as a function of maxIter, for the
2615      file Parse/parse/ml.grm.sml:
2616    
2617         maxIter            # of instructions:
2618            10                      687
2619            20                      438
2620            30                      198
2621            40                        0
2622    
2623      In compiling the compiler, there is no significant difference in
2624      compilation speed between maxIter=10 and maxIter=40. Actually,
2625      my measurements showed that maxIter=40 was a tad faster than
2626      maxIter=10! Also 96% of the  files in the compiler reach a fix
2627      point within 13 iterations, so fixing maxIter at 40, while high,
2628      is okay.
2629    
2630    ----------------------------------------------------------------------
2631    Name: Matthias Blume
2632    Date: 2001/10/31 15:25:00 EST
2633    Tag: blume-20011031-pgraph
2634    Description:
2635    
2636    CKIT:
2637    * Changed the "Function" constructor of type Ast.ctype to carry optional
2638      argument identifiers.
2639    * Changed the return type of TypeUtil.getFunction accordingly.
2640    * Type equality ignores the argument names.
2641    * TypeUtil.composite tries to preserve argument names but gives up quickly
2642      if there is a mismatch.
2643    
2644    installation script:
2645    * attempts to use "curl" if available (unless "wget" is available as well)
2646    
2647    CM:
2648    * has an experimental implementation of "portable graphs" which I will
2649      soon propose as an implementation-independent library format
2650    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
2651    
2652    NLFFI-LIB:
2653    * some cleanup (all cosmetic)
2654    
2655    NLFFIGEN:
2656    * temporarily disabled the mechanism that suppresses ML output for
2657      C definitions whose identifiers start with an underscore character
2658    * generate val bindings for enum constants
2659    * user can request that only one style (light or heavy) is being used;
2660      default is to use both (command-line arguments: -heavy and -light)
2661    * fixed bug in handling of function types involving incomplete pointers
2662    * generate ML entry points that take record arguments (i.e., using
2663      named arguments) for C functions that have a prototype with named
2664      arguments
2665      (see changes to CKIT)
2666    
2667    ----------------------------------------------------------------------
2668    Name: Allen Leung
2669    Date: 2001/10/27 20:34:00 EDT
2670    Tag: leunga-20011027-x86-fast-fp-call
2671    Description:
2672    
2673       Fixed the bug described in blume-20010920-slowfp.
2674    
2675       The fix involves
2676          1. generating FCOPYs in FSTP in ia32-svid
2677          2. marking a CALL with the appropriate annotation
2678    
2679    ----------------------------------------------------------------------
2680    Name: Matthias Blume
2681    Date: 2001/10/16 11:32:00 EDT
2682    Tag: blume-20011016-netbsd
2683    Description:
2684    
2685    Underscore patch from Chris Richards (fixing problem with compiling
2686    runtime system under recent NetBSD).
2687    
2688    ----------------------------------------------------------------------
2689    Name: Allen Leung
2690    Date: 2001/10/12 17:18:32 EDT 2001
2691    Tag: leung-20011012-x86-printflowgraph
2692    Description:
2693    
2694    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
2695    
2696    ----------------------------------------------------------------------
2697    Name: Lal George
2698    Date: 2001/10/11 23:51:34 EDT
2699    Tag: george-20011011-too-many-instrs
2700    Description:
2701    
2702    The representation of a program point never expected to see more
2703    than 65536 instructions in a basic block!
2704    
2705    ----------------------------------------------------------------------
2706    Name: Lal George
2707    Date: 2001/10/09 09:41:37 EDT
2708    Tag: george-20011008-mlrisc-labels
2709    Description:
2710    
2711    Changed the machine description files to support printing of
2712    local and global labels in assembly code, based on host assembler
2713    conventions.
2714    
2715    ----------------------------------------------------------------------
2716    Name: Matthias Blume
2717    Date: 2001/09/25 15:25:00 EDT
2718    Tag: blume-20010925-exninfo
2719    Description:
2720    
2721    I provided a non-hook implementation of exnName (at the toplevel) and
2722    made the "dummy" implementation of exnMessage (at the toplevel) more
2723    useful: if nothing gets "hooked in", then at least you are going to
2724    see the exception name and a message indicating why you don't see more.
2725    
2726    [For the time being, programs that need exnMessage and want to use
2727    ml-build should either use General.exnMessage (strongly recommended) or
2728    refer to structure General at some other point so that CM sees a
2729    static dependency.]
2730    
2731    [Similar remarks go for "print" and "use":  If you want to use their
2732    functionality in stand-alone programs generated by ml-build, then use
2733    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
2734    
2735    ----------------------------------------------------------------------
2736    Name: Matthias Blume
2737    Date: 2001/09/20 17:28:00 EDT
2738    Tag: blume-20010920-slowfp
2739    Description:
2740    
2741    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
2742    
2743    ----------------------------------------------------------------------
2744    Name: Matthias Blume
2745    Date: 2001/09/20 17:20:00 EDT
2746    Tag: blume-20010920-canonicalpaths
2747    Description:
2748    
2749    0. Updated the BOOT file (something that I forgot to do earlier).
2750    
2751    1. Small internal change to CM so that it avoids "/../" in filenames
2752       as much as possible (but only where it is safe).
2753    
2754    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
2755       that arguments that contain delimiters are passed through correctly.
2756       This change also means that all "special" arguments of the form
2757       @SMLxxx... must come first.
2758    
2759    3. Changed install script to put relative anchor names for tool commands
2760       into pathconfig.
2761    
2762    ----------------------------------------------------------------------
2763    Name: Matthias Blume
2764    Date: 2001/09/18 15:35:00 EDT
2765    Tag: blume-20010918-readme11036
2766    Description:
2767    
2768    Added README files.
2769    
2770    ----------------------------------------------------------------------
2771    Name: Matthias Blume
2772    Date: 2001/09/18 11:45:00 EDT
2773    Tag: Release_110_36 (retag)
2774    Description:
2775    
2776    Fixed mistake in config/preloads. Retagged as 110.36.
2777    
2778    ----------------------------------------------------------------------
2779    Name: Matthias Blume
2780    Date: 2001/09/18 09:40:00 EDT
2781    Tag: Release_110_36_orig (tag changed)
2782    Description:
2783    
2784    New version (110.36).  New bootfiles.
2785    
2786    ----------------------------------------------------------------------
2787    Name: Matthias Blume
2788    Date: 2001/09/14 16:15:00 EDT
2789    Tag: blume-20010914-x86fastfp
2790    Description:
2791    
2792    John committed some changes that Allen made, in particular a (hopefully)
2793    correctly working version of the x86-fp module.
2794    
2795    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
2796    flag to "true".  Everything seems to compile to a fixpoint ok, and
2797    "mandelbrot" speeds up by about 15%.
2798    
2799    ----------------------------------------------------------------------
2800    Name: Matthias Blume
2801    Date: 2001/09/13 11:20:00 EDT
2802    Tag: blume-20010913-minimal
2803    Description:
2804    
2805    1. Stefan Monnier's patch to fix a miscompilation problem that
2806       was brought to light by John Reppy's work on Moby.
2807    
2808    2. Implemented a minimal "structure Compiler" that contains just
2809       "version" and "architecture".  The minimal version will be
2810       available when the full version is not.  This is for backward-
2811       compatibility with code that wants to test Compiler.version.
2812    
2813    ----------------------------------------------------------------------
2814    Name: Matthias Blume
2815    Date: 2001/08/28 14:03:00 EDT
2816    Tag: blume-20010828-ml-lex
2817    Description:
2818    
2819    Fix for bug 1581, received from Neophytos Michael.
2820    
2821    ----------------------------------------------------------------------
2822    Name: Matthias Blume
2823    Date: 2001/08/27 11:20:00 EDT
2824    Tag: blume-20010827-readme11035
2825    Description:
2826    
2827    Fleshed out the README file for 110.35.
2828    
2829    ----------------------------------------------------------------------
2830    Name: Matthias Blume
2831    Date: 2001/08/24 17:10:00 EDT
2832    Tag: Release_110_35
2833    Description:
2834    
2835    New version number (110.35).  New bootfiles.
2836    
2837    ----------------------------------------------------------------------
2838    Name: Lal George
2839    Date: 2001/08/24 13:47:18 EDT 2001
2840    Tag: george-20010824-MLRISC-graphs
2841    Description:
2842    
2843     removed clusters from MLRISC completely and replaced with graphs.
2844    
2845    ----------------------------------------------------------------------
2846    Name: Matthias Blume
2847    Date: 2001/08/23 17:50:00 EDT
2848    Tag: blume-20010823-toplevel
2849    Description:
2850    
2851    - some reorganization of the code that implements various kinds of
2852      environments in the compiler (static, dynamic, symbolic, combined)
2853    - re-implemented the EnvRef module so that evalStream works properly
2854      (if the stream contains references to "use", "CM.make", etc.)
2855    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
2856    
2857    ----------------------------------------------------------------------
2858    Name: Matthias Blume
2859    Date: 2001/08/20 15:50 EDT
2860    Tag: blume20010820-slipup
2861    Description:
2862    
2863    I forgot to commit a few files.  Here they are...
2864    
2865    ----------------------------------------------------------------------
2866    Name: Matthias Blume
2867    Date: 2001/08/20 15:35:00 EDT
2868    Tag: blume-20010820-debugprof
2869    Description:
2870    
2871    !!!! NEW BOOTFILES !!!!
2872    
2873    This is another round of reorganizing the compiler sources.  This
2874    time the main goal was to factor out all the "instrumentation"
2875    passes (for profiling and backtracing) into their own library.
2876    The difficulty was to do it in such a way that it does not depend
2877    on elaborate.cm but only on elabdata.cm.
2878    
2879    Therefore there have been further changes to both elaborate.cm and
2880    elabdata.cm -- more "generic" things have been moved from the former
2881    to the latter.  As a result, I was forced to split the assignment
2882    of numbers indicating "primtyc"s into two portions: SML-generic and
2883    SML/NJ-specific.  Since it would have been awkward to maintain,
2884    I bit the bullet and actually _changed_ the mapping between these
2885    numbers and primtycs.  The bottom line of this is that you need
2886    a new set of bin- and bootfiles.
2887    
2888    I have built new bootfiles for all architectures, so doing a fresh
2889    checkout and config/install.sh should be all you need.
2890    
2891    The newly created library's name is
2892    
2893        $smlnj/viscomp/debugprof.cm
2894    
2895    and its sources live under
2896    
2897        src/compiler/DebugProf
2898    
2899    ----------------------------------------------------------------------
2900    Name: Matthias Blume
2901    Date: 2001/08/15 17:15:00 EDT
2902    Tag: blume-20010815-compreorg
2903    Description:
2904    
2905    This is a first cut at reorganizing the CM libraries that make up the
2906    core of the compiler.  The idea is to separate out pieces that could
2907    be used independently by tools, e.g., the parser, the typechecker, etc.
2908    
2909    The current status is a step in this direction, but it is not quite
2910    satisfactory yet.  Expect more changes in the future.
2911    
2912    Here is the current (new) organization...
2913    
2914        What used to be $smlnj/viscomp/core.cm is now divided into
2915        six CM libraries:
2916    
2917             $smlnj/viscomp/basics.cm
2918                           /parser.cm
2919                           /elabdata.cm
2920                           /elaborate.cm
2921                           /execute.cm
2922                           /core.cm
2923    
2924        The CM files for these libraries live under src/system/smlnj/viscomp.
2925        All these libraries are proxy libraries that contain precisely
2926        one CM library component.  Here are the locations of the components
2927        (all within the src/compiler tree):
2928    
2929             Basics/basics.cm
2930             Parse/parser.cm
2931             ElabData/elabdata.cm
2932             Elaborator/elaborate.cm
2933             Execution/execute.cm
2934             core.cm
2935    
2936         [This organization is the same that has been used already
2937         for a while for the architecture-specific parts of the visible
2938         compiler and for the old version of core.cm.]
2939    
2940         As you will notice, many source files have been moved from their
2941         respective original locations to a new home in one of the above
2942         subtrees.
2943    
2944         The division of labor between the new libraries is the following:
2945    
2946             basics.cm:
2947                - Simple, basic definitions that pertain to many (or all) of
2948                  the other libraries.
2949             parser.cm:
2950                - The SML parser, producing output of type Ast.dec.
2951                - The type family for Ast is also defined and exported here.
2952             elabdata.cm:
2953                - The datatypes that describe input and output of the elaborator.
2954                  This includes types, absyn, and static environments.
2955             elaborator.cm:
2956                - The SML/NJ type checker and elaborator.
2957                  This maps an Ast.dec (with a given static environment) to
2958                  an Absyn.dec (with a new static environment).
2959                - This libraries implements certain modules that used to be
2960                  structures as functors (to remove dependencies on FLINT).
2961             execute.cm:
2962                - Everything having to do with executing binary code objects.
2963                - Dynamic environments.
2964             core.cm:
2965                - SML/NJ-specific instantiations of the elaborator and MLRISC.
2966                - Top-level modules.
2967                - FLINT (this should eventually become its own library)
2968    
2969    Notes:
2970    
2971    I am not 100% happy with the way I separated the elaborator (and its
2972    data structures) from FLINT.  Two instances of the same problem:
2973    
2974        1. Data structures contain certain fields that carry FLINT-specific
2975           information.  I hacked around this using exn and the property list
2976           module from smlnj-lib.  But the fact that there are middle-end
2977           specific fields around at all is a bit annoying.
2978    
2979        2. The elaborator calculates certain FLINT-related information.  I tried
2980           to make this as abstract as I could using functorization, but, again,
2981           the fact that the elaborator has to perform calculations on behalf
2982           of the middle-end at all is not nice.
2983    
2984        3. Having to used exn and property lists is unfortunate because it
2985           weakens type checking.  The other alternative (parameterizing
2986           nearly *everything*) is not appealing, though.
2987    
2988    I removed the "rebinding =" warning hack because due to the new organization
2989    it was awkward to maintain it.  As a result, the compiler now issues some of
2990    these warnings when compiling init.cmi during bootstrap compilation. On
2991    the plus side, you also get a warning when you do, for example:
2992       val op = = Int32.+
2993    which was not the case up to now.
2994    
2995    I placed "assign" and "deref" into the _Core structure so that the
2996    code that deals with the "lazy" keyword can find them there.  This
2997    removes the need for having access to the primitive environment
2998    during elaboration.
2999    
3000    ----------------------------------------------------------------------
3001    Name: Matthias Blume
3002    Date: 2001/08/13
3003    Tag: blume-20010813-closures
3004    Description:
3005    
3006    This fix was sent to us by Zhong Shao.  It is supposed to improve the
3007    performance of certain loops by avoiding needless closure allocation.
3008    
3009    ----------------------------------------------------------------------
3010    Name: Lal George
3011    Date: 2001/07/31 10:03:23 EDT 2001
3012    Tag: george-20010731-x86-fmalloc
3013    Description: Fixed bug in x86 calls
3014    
3015        There was a bug where call instructions would mysteriously
3016        vanish. The call instruction had to be one that returned
3017        a floating point value.
3018    
3019    ----------------------------------------------------------------------
3020    Name: Lal George
3021    Date: 2001/07/19 16:36:29 EDT 2001
3022    Tag: george-20010719-simple-cells
3023    Description:
3024    
3025    I have dramatically simplified the interface for CELLS in MLRISC.
3026    
3027    In summary, the cells interface is broken up into three parts:
3028    
3029      1. CellsBasis : CELLS_BASIS
3030    
3031            CellsBasis is a top level structure and common for all
3032            architectures.  it contains the definitions of basic datatypes
3033            and utility  functions over these types.
3034    
3035      2. functor Cells() : CELLS
3036    
3037            Cells generates an interface for CELLS that incorporates the
3038            specific resources on the target architecture, such as the
3039            presence of special register classes, their number and size,
3040            and various useful substructures.
3041    
3042      3. <ARCH>CELLS
3043    
3044            e.g. SparcCells: SPARCCELLS
3045    
3046            <ARCH>CELLS usually contains additional bindings for special
3047            registers  on the architecture, such as:
3048    
3049                    val r0 : cell           (* register zero *)
3050                    val y : cell            (* Y register *)
3051                    val psr : cell          (* processor status register *)
3052                    ...
3053    
3054            The structure returned by applying the Cells functor is opened
3055            in this interface.
3056    
3057    The main implication of all this is that the datatypes for cells is
3058    split between CellsBasis and CELLS -- a fairly simple change for user
3059    code.
3060    
3061    In the old scheme the CELLS interface had a definitional binding of
3062    the form:
3063    
3064            signature CELLS = sig
3065    
3066               structure CellsBasis = CellsBasis
3067    
3068               ...
3069    
3070            end
3071    
3072    With all the sharing constraints that goes on in MLRISC, this old
3073    design  quickly leads to errors such as:
3074    
3075            "structure definition spec inside of sharing ... "
3076    
3077    
3078    and appears to require an unacceptable amount of sharing and where
3079    constraint hackery.
3080    
3081    I think this error message (the interaction of definitional specs and
3082    sharing) requires more explanation on our web page.
3083    
3084    ----------------------------------------------------------------------
3085    Name: Matthias Blume
3086    Date: 2001/07/19 15:00:00 EDT
3087    Tag: blume-20010719-libreorg
3088    Description:
3089    
3090    This update puts together a fairly extensive but straightforward change
3091    to the way the libraries that implement the interactive system are
3092    organized:
3093    
3094       The biggest change is the elimination of structure Compiler.  As a
3095       replacement for this structure, there is now a CM library
3096       (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
3097       that exports all the substructures of the original structure Compiler
3098       directly.  So instead of saying Compiler.Foo.bar one now simply
3099       says Foo.bar.  (The CM libraries actually export a collection of
3100       structures that is richer than the collection of substructures of
3101       structure Compiler.)
3102    
3103       To make the transition smooth, there is a separate library called
3104       $smlnj/compiler/compiler.cm which puts together and exports the
3105       original structure Compiler (or at least something very close to it).
3106    
3107       There are five members of the original structure Compiler
3108       that are not exported directly but which instead became members
3109       of a new structure Backend (described by signature BACKEND).  These are:
3110       structure Profile (: PROFILE), structure Compile (: COMPILE), structure
3111       Interact (: INTERACT), structure Machine (: MACHINE), and val
3112       architecture (: string).
3113    
3114       Structure Compiler.Version has become structure CompilerVersion.
3115    
3116       Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
3117       by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
3118       or x86, respectively.
3119       Each of these exports the same frontend structures that
3120       $smlnj/compiler.cm exports.  But they do not have a structure Backend
3121       and instead export some structure <Arch>Backend where <Arch> is Alpha32,
3122       Hppa, PPC, Sparc, or X86, respectively.
3123    
3124       Library $smlnj/compiler/all.cm exports the union of the exports of
3125       $smlnj/compiler/<arch>.cm
3126    
3127       There are no structures <Arch>Compiler anymore, use
3128       $smlnj/compiler/<arch>.cm instead.
3129    
3130       Library host-compiler-0.cm is gone.  Instead, the internal library
3131       that instantiates CM is now called cm0.cm.  Selection of the host
3132       compiler (backend) is no longer done here but. (Responsibility for it
3133       now lies with $smlnj/compiler/current.cm.  This seems to be more
3134       logical.)
3135    
3136       Many individual files have been moved or renamed.  Some files have
3137       been split into multiple files, and some "dead" files have been deleted.
3138    
3139    Aside from these changes to library organization, there are also changes
3140    to the way the code itself is organized:
3141    
3142       Structure Binfile has been re-implemented in such a way that it no
3143       longer needs any knowledge of the compiler.  It exclusively deals
3144       with the details of binfile layout.  It no longer invokes the
3145       compiler (for the purpose of creating new prospective binfile
3146       content), and it no longer has any knowledge of how to interpret
3147       pickles.
3148    
3149       Structure Compile (: COMPILE) has been stripped down to the bare
3150       essentials of compilation.  It no longer deals with linking/execution.
3151       The interface has been cleaned up considerably.
3152    
3153       Utility routines for dealing with linking and execution have been
3154       moved into their own substructures.
3155    
3156       (The ultimate goal of these changes is to provide a light-weight
3157       binfile loader/linker (at least for, e.g., stable libraries) that
3158       does not require CM or the compiler to be present.)
3159    
3160    CM documentation has been updated to reflect the changes to library
3161    organization.
3162    
3163    ----------------------------------------------------------------------
3164    Name: Matthias Blume
3165    Date: 2001/07/10 17:30:00 EDT
3166    Tag: Release_110_34
3167    Description:
3168    
3169    Minor tweak to 110.34 (re-tagged):
3170    
3171      - README.html file added to CVS repository
3172      - runtime compiles properly under FreeBSD 3.X and 4.X
3173    
3174    ----------------------------------------------------------------------
3175    Name: Matthias Blume
3176    Date: 2001/07/10 17:30:00 EDT
3177    Tag: Release_110_34
3178    Description:
3179    
3180    New version number (110.34). New bootfiles.
3181    
3182    ----------------------------------------------------------------------
3183    Name: Matthias Blume
3184    Date: 2001/07/09 16:00:00 EDT
3185    Tag: blume-20010709-more-varargs
3186    Description:
3187    
3188    I changed the handling of varargs in ml-nlffigen again:
3189    The ellipsis ... will now simply be ignored (with an accompanying warning).
3190    
3191    The immediate effect is that you can actually call a varargs function
3192    from ML -- but you can't actually supply any arguments beyond the ones
3193    specified explicitly.  (For example, you can call printf with its format
3194    string, but you cannot pass additional arguments.)
3195    
3196    This behavior is only marginally more useful than the one before, but
3197    it has the advantage that a function or, more importantly, a function
3198    type never gets dropped on the floor, thus avoiding follow-up problems with
3199    other types that refer to the offending one.
3200    
3201    ----------------------------------------------------------------------
3202    Name: Matthias Blume
3203    Date: 2001/07/09 11:25:00 EDT
3204    Tag: blume-20010709-varargs
3205    Description:
3206    
3207    1. ckit-lib.cm now exports structure Error
3208    2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
3209       with a warning accompanied by a source location.  Moreover, it
3210       merely skips the offending function or type and proceeds with the
3211       rest of its work.u  As a result, one can safely feed C code containing
3212       "..." to ml-nlffigen.
3213    3. There are some internal improvements to CM, providing slightly
3214       more general string substitutions in the tools subsystem.
3215    
3216    ----------------------------------------------------------------------
3217    Name: Matthias Blume
3218    Date: 2001/06/27 15:10:00 EDT
3219    Tag: blume-20010627-concur
3220    Description:
3221    
3222    Fixed a small bug in CM's handling of parallel compilation.
3223    (You could observe the bug by Control-C-interrupting an ordinary
3224    CMB.make or CM.stabilize and then attaching some compile servers.
3225    The result was that all of a sudden the previously interrupted
3226    compilation would continue on its own.  This was because of
3227    an over-optimization: CM did not bother to clean out certain queues
3228    when no servers were attached "anyway", resulting in the contents
3229    of these queues to grab control when new servers did get attached.)
3230    
3231    There is also another minor update to the CM manual.
3232    
3233    ----------------------------------------------------------------------
3234    Name: Matthias Blume
3235    Date: 2001/06/26 16:15:00 EDT
3236    Tag: blume-20010626-cmdoc
3237    Description:
3238    
3239    Minor typo fixed in CM manual (syntax diagram for libraries).
3240    
3241    ----------------------------------------------------------------------
3242    Name: Matthias Blume
3243    Date: 2001/06/25 22:55:00 EDT
3244    Tag: blume-20010625-x86pc
3245    Description:
3246    
3247    Fixed a nasty bug in the X86 assembly code that caused signal
3248    handlers to fail (crash) randomly.
3249    
3250    ----------------------------------------------------------------------
3251    Name: Matthias Blume
3252    Date: 2001/06/25 12:05:00 EDT
3253    Tag: blume-20010625-nlffigen
3254    Description:
3255    
3256    This update fixes a number of minor bugs in ml-nlffigen as reported by
3257    Nick Carter <nbc@andrew.cmu.edu>.
3258    
3259      1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
3260      2. Default names for generated files are now derived from the name of
3261         the C file *without its directory*.  In particular, this causes generated
3262         files to be placed locally even if the C file is in some system directory.
3263      3. Default names for generated signatures and structures are also derived
3264         from the C file name without its directory.  This avoids silly things
3265         like "structure GL/GL".
3266         (Other silly names are still possible because ml-nlffigen does not do
3267          a thorough check of whether generated names are legal ML identifiers.
3268          When in doubt, use command line arguments to force particular names.)
3269    
3270    ----------------------------------------------------------------------
3271    Name: Matthias Blume
3272    Date: 2001/06/21 12:25:00 EDT
3273    Tag: blume-20010621-eXene
3274    Description:
3275    
3276    eXene now compiles and (sort of) works again.
3277    
3278    The library name (for version > 110.33) is $/eXene.cm.
3279    
3280    I also added an new example in src/eXene/examples/nbody.  See the
3281    README file there for details.
3282    
3283    ----------------------------------------------------------------------
3284    Name: Matthias Blume
3285    Date: 2001/06/20 16:40:00 EDT
3286    Tag: blume-20010620-cml
3287    Description:
3288    
3289    CML now compiles and works again.
3290    
3291    Libraries (for version > 110.33):
3292    
3293      $cml/cml.cm            Main CML library.
3294      $cml/basis.cm          CML's version of $/basis.cm.
3295      $cml/cml-internal.cm   Internal helper library.
3296      $cml/core-cml.cm       Internal helper library.
3297      $cml-lib/trace-cml.cm  Tracing facility.
3298      $cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
3299    
3300    The installer (config/install.sh) has been taught how to properly
3301    install this stuff.
3302    
3303    ----------------------------------------------------------------------
3304    Name: Matthias Blume
3305    Date: 2001/06/19 17:55:00 EDT
3306    Tag: blume-20010619-instantiate
3307    Description:
3308    
3309    This un-breaks the fix for bug 1432.
3310    (The bug was originally fixed in 110.9 but I broke it again some
3311    time after that.)
3312    
3313    ----------------------------------------------------------------------
3314    Name: Matthias Blume
3315    Date: 2001/06/19 17:25:00 EDT
3316    Tag: blume-20010619-signals
3317    Description:
3318    
3319    This should (hopefully) fix the long-standing signal handling bug.
3320    (The runtime system was constructing a continuation record with an
3321    incorrect descriptor which would cause the GC to drop data on the floor...)
3322    
3323    ----------------------------------------------------------------------
3324    Name: Matthias Blume
3325    Date: 2001/06/15 15:05:00 EDT
3326    Tag: blume-20010615-moresparc
3327    Description:
3328    
3329    Here is a short late-hour update related to Sparc c-calls:
3330    
3331     -- made handling of double-word arguments a bit smarter
3332    
3333     -- instruction selection phase tries to collapse certain clumsily
3334        constructed ML-Trees; typical example:
3335    
3336            ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
3337    
3338        This currently has no further impact on SML/NJ since mlriscGen does
3339        not seem to generate such patterns in the first place, and c-calls
3340        (which did generate them in the beginning) has meanwhile been fixed
3341        so as to avoid them as well.
3342    
3343    ----------------------------------------------------------------------
3344    Name: Matthias Blume
3345    Date: 2001/06/15 15:05:00 EDT
3346    Tag: blume-20010615-sparc
3347    Description:
3348    
3349    The purpose of this update is to provide an implementation of NLFFI
3350    on Sparc machines.
3351    
3352    Here are the changes in detail:
3353    
3354       * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
3355       the Sparc implementation of the c-calls API.
3356       * The Sparc backend of SML/NJ has been modified to uniformely use %fp
3357       for accessing the ML frame.  Thus, we have a real frame pointer and
3358       can freely modify %sp without need for an omit-frame-ptr phase.
3359       The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
3360       this case.
3361       * ml-nlffigen has been taught to produce code for different architectures
3362       and calling conventions.
3363       * In a way similar to what was done in the x86 case, the Sparc
3364       backend uses its own specific extension to mltree.  (For example,
3365       it needs to be able to generate UNIMP instructions which are part
3366       of the calling convention.)
3367       * ml-nlffi-lib was reorganized to make it more modular (in particular,
3368       to make it easier to plug in new machine- and os-dependent parts).
3369    
3370    There are some other fairly unrelated bug fixes and cleanups as well:
3371    
3372       * I further hacked the .cm files for MLRISC tools (like MDLGen) so
3373       that they properly share their libraries with existing SML/NJ libraries.
3374       * I fixed a minor cosmetic bug in CM, supressing certain spurious
3375       follow-up error messages.
3376       * Updates to CM/CMB documentation.
3377    
3378    TODO items:
3379    
3380       * MLRISC should use a different register as its asmTemp on the Sparc.
3381         (The current %o2 is a really bad choice because it is part of the
3382         calling conventions, so things might interfere in unexpected ways.)
3383    
3384    ----------------------------------------------------------------------
3385    Name: Matthias Blume
3386    Date: 2001/06/07
3387    Tag: blume-20010607-calls
3388    Description:
3389    
3390    A number of internal changes related to C calls and calling conventions:
3391    
3392    1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
3393       number of bytes popped implicitly (by the callee).  In most cases
3394       this field is 0 but on x86/win32 it is some non-zero value.  This
3395       is information provided for the benefit of the "omit-frameptr" pass.
3396    2. The CALL instruction on the x86 carries a similar "pops" field.
3397       The instruction selection phase copies its value from the ML-Tree
3398       CALL statement.
3399    3. On all other architectures, the instruction selection phase checks
3400       whether "pops=0" and complains if not.
3401    4. The c-calls implementation for x86 now accepts two calling conventions:
3402       "ccall" and "stdcall".  When "ccall" is selected, the caller cleans
3403       up after the call and pops is set to 0.  For "stdcall", the caller
3404       does nothing, leaving the cleanup to the callee; pops is set to
3405       the number of bytes that were pushed onto the stack.
3406    5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
3407       distinguish between "ccall" and "stdcall".
3408    6. The UNIMP instruction has been added to the supported Sparc instruction
3409       set. (This is needed for implementing the official C calling convention
3410       on this architecture.)
3411    7. I fixed some of the .cm files under src/MLRISC/Tools to make them
3412       work with the latest CM.
3413    
3414    ----------------------------------------------------------------------
3415    Name: Matthias Blume
3416    Date: 2001/06/05 15:10:00 EDT
3417    Tag: blume-20010605-cm-index
3418    Description:
3419    
3420    0. The "lambdasplit" parameter for class "sml" in CM has been documented.
3421    
3422    1. CM can now generate "index files".  These are human-readable files
3423       that list on a per-.cm-file basis each toplevel symbol defined or
3424       imported.  The location of the index file for
3425       <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
3426       To enable index-file generation, set CM.Control.generate_index to true
3427       or export an environment-symbol: export CM_GENERATE_INDEX=true.
3428    
3429       The CM manual has been updated accordingly.
3430    
3431    2. I made some slight modifications to the c-calls API in MLRISC.
3432    
3433         a) There is now a callback to support saving/restoring of
3434            dedicated but caller-save registers around the actual call
3435            instruction.
3436         b) One can optionally specify a comment-annotation for the
3437            call instruction.
3438    
3439    3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
3440       (For example, the comment annotation shows the C prototype of
3441        the function being called.)
3442    
3443    ----------------------------------------------------------------------
3444    Name: Matthias Blume
3445    Date: 2001/06/01 13:30:00 EDT
3446    Tag: blume-20010601-nlffi-cleanup
3447    Description:
3448    
3449    This is mostly a cleanup of MLFFI stuff:
3450    
3451       - some signature files have been put into a more exposed place
3452       - the ugly 'f type parameter is gone (simplifies types tremendously!)
3453       - ml-nlffigen changed accordingly
3454       - tutorial updated
3455    
3456    Other changes:
3457    
3458       - author's affiliation in CM manual(s) updated
3459       - some more recognized keywords added to Allen's sml.sty
3460    
3461    ----------------------------------------------------------------------
3462    Name: Matthias Blume
3463    Date: 2001/05/25 15:30:00 EDT
3464    Tag: blume-20010525-iptr
3465    Description:
3466    
3467      - put the official 110.33-README (as it appears on the ftp server) under
3468        CVS
3469      - fixed a small bug related to incomplete pointer types in
3470        ml-nlffigen
3471      - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
3472        (it does not need the 'f type parameter)
3473    
3474    ----------------------------------------------------------------------
3475    Name: Matthias Blume
3476    Date: 2001/05/23 14:30:00 EDT
3477    Tag: Release_110_33
3478    Description:
3479    
3480    New version number (110.33).  New bootfiles.
3481    
3482    ----------------------------------------------------------------------
3483    Name: Matthias Blume
3484    Date: 2001/05/22 18:06:00 EDT
3485    Tag: blume-20010522-targets
3486    Description:
3487    
3488    Made install.sh use file config/targets.customized if it exists, falling
3489    back to config/targets if it doesn't.  This way one can have a customized
3490    version of the targets file without touching the "real thing", thus
3491    eliminating the constant fear of accidentally checking something bogus
3492    back into the CVS repository...  (File config/targets.customized must
3493    not be added to the repository!)
3494    
3495    ----------------------------------------------------------------------
3496    Name: Matthias Blume
3497    Date: 2001/05/22 16:30:00 EDT
3498    Tag: blume-20010522-minitut
3499    Description:
3500    
3501    1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
3502       struct returns.
3503    2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
3504       incomplete, preliminary documentation for NLFFI.
3505    
3506    ----------------------------------------------------------------------
3507    Name: Matthias Blume
3508    Date: 2001/05/14 11:30:00 EDT
3509    Tag: blume-20010514-script
3510    Description:
3511    
3512    Some bugs in install script fixed.
3513    
3514    In addition to that I also made a slight change to the NLFFI API:
3515    Functors generated by ml-nlffigen now take the dynamic library as a
3516    straight functor argument, not as a suspended one.  (The original
3517    functor code used to force the suspension right away anyway, so there
3518    was nothing gained by this complication of the interface.)
3519    
3520    ----------------------------------------------------------------------
3521    Name: Matthias Blume
3522    Date: 2001/05/11 14:35:00 EDT
3523    Tag: blume-20010511-ml-nlffi
3524    Description:
3525    
3526    I finally took the plunge and added my new FFI code to the main
3527    repository.  For x86-linux it is now ready for prime-time.
3528    
3529    There are two new subdirectories of "src":
3530    
3531      - ml-nlffi-lib:
3532           The utility library for programs using the FFI interface.
3533           Here is the implementation of $/c.cm and its associated low-level
3534           partners $/c-int.cm and $/memory.cm.
3535      - ml-nlffigen:
3536           A stand-alone program for generating ML glue code from C source
3537           code.
3538    
3539    Building ml-nlffigen requires $/ckit-lib.cm.
3540    
3541    The config/install.sh script has been updates to do the Right Thing
3542    (hopefully).
3543    
3544    Notice that the source tree for the C-Kit will not be put under "src"
3545    but directly under the installation root directory.  (This is the
3546    structure that currently exists on the CVS server when you check out
3547    module "sml".)  Fortunately, config/install.sh knows about this oddity.
3548    
3549    Bugs: No documentation yet.
3550    
3551    ----------------------------------------------------------------------
3552    Name: Matthias Blume
3553    Date: 2001/05/09 16:35:00 EDT
3554    Tag: blume-20010509-cpscontract
3555    Description:
3556    
3557    Fixed a bug in the accounting code in cpsopt/contract.sml.  (The
3558    wrapper/unwrapper elimination did not decrement usage counts and some
3559    dead variables got overlooked by the dead-up logic.)
3560    
3561    ----------------------------------------------------------------------
3562    Name: Lal George
3563    Date: 2001/05/08  17:26:09 EDT
3564    Tag: george-20010508-omit-frameptr
3565    Description:
3566    
3567    Changes to implement the omit-frame-pointer optimization to support
3568    raw C calls. For now, there is only support on the Intel x86, but
3569    other architectures will follow as more experience is gained with this.
3570    
3571    
3572    ----------------------------------------------------------------------
3573    Name: Matthias Blume
3574    Date: 2001/05/07 14:40:00 EDT
3575    Tag: blume-20010507-proxies
3576    Description:
3577    
3578    I made into "proxy libraries" all libraries that qualify for such a
3579    change.  (A qualifying library is a library that has another library or
3580    groups as its sole member and repeats that member's export list
3581    verbatim.  A proxy library avoids this repetition by omitting its export
3582    list, effectively inheriting the list that its (only) member exports.
3583    See the CM manual for more explanation.)
3584    The main effect is that explicit export lists for these libraries
3585    do not have to be kepts in sync, making maintenance a bit easier.
3586    
3587    I also added copyright notices to many .cm-files.
3588    
3589    Last but not least, I made a new set of bootfiles.
3590    
3591    ----------------------------------------------------------------------
3592    Name: Matthias Blume
3593    Date: 2001/05/04 17:00:00 EDT
3594    Tag: blume-20010504-cm-lsplit
3595    Description:
3596    
3597    0. John merged pending changes to $/smlnj-lib.cm
3598    
3599    1. Allen's previous change accidentally backed out of one of Lal's
3600       earlier changes.  I undid this mistake (re-introducing Lal's change).
3601    
3602    2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
3603       within the compiler where applicable.  There is some code simplification
3604       because of that.
3605    
3606    3. The "split" phase (in FLINT) is now part of the default list of phases.
3607       Compiler.Control.LambdaSplitting.* can be used to globally control the
3608       lambda-splitting (cross-module-inlining) engine.  In addition to that,
3609       it can now also be controlled on a per-source basis: CM has been taught