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 826, Wed May 9 20:32:17 2001 UTC revision 1368, Sat Sep 13 04:44:43 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: 2003/09/12 11:45:00 CDT
18    Tag: blume-20030912-various
19    Description:
20    
21     - fixed bug in Real.toLargeInt
22     - fixed bug in Posix.ProcEnv.times
23     - changed inputLine functions to return an option
24     - minor installer improvements / bugfixes
25     - changed default @SMLalloc parameter for x86/celeron to 64k
26    
27    ----------------------------------------------------------------------
28    Name: Matthias Blume (blume (at) tti - c (dot) org)
29    Date: 2003/09/09 22:00:00 CDT
30    Tag: Release_110_43
31    Description:
32    
33        New working release 110.43.  New bootfiles.
34    
35    ----------------------------------------------------------------------
36    Name: Matthias Blume (blume (at) tti - c (dot) org)
37    Date: 2003/09/09 19:20:00 CDT
38    Tag: blume-20030909-installer
39    Description:
40    
41    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
42    Modified config/install.bat to take advantage of it.  Also modified
43    config/install.sh (and called it config/new-install.sh) to take advantage
44    of it on Unix systems.  (The SML code is (supposed to be) platform-
45    independent.)
46    
47    The installer can now install everything under Win32
48    as well as under *nix as long as it compiles.
49    
50    Other changes:
51    
52      - made CML compile again under Win32
53      - made eXene compile under Win32 (by providing a fake structure UnixSock
54        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
55      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
56        same as type int;  under Win32 it isn't
57      - fixed some slice-related problems in the win32-specific parts of CML
58      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
59        (under Win32, the volume name is case-insensitive, and the
60        OS.Path code compares volume names for equality)
61    
62    ----------------------------------------------------------------------
63    Name: Matthias Blume (blume (at) tti - c (dot) org)
64    Date: 2003/09/08 11:55:00 CDT
65    Tag: blume-20030908-fullpath
66    Description:
67    
68    Made Win32 version of OS.FileSys.fullPath return current directory
69    when given an empty string.  This is what the spec says, and incidentally,
70    CM depends on it.  (CM otherwise goes into an infinite loop in certain
71    cases when presented with the name of a non-existing .cm file.)
72    
73    ----------------------------------------------------------------------
74    Name: Matthias Blume (blume (at) tti - c (dot) org)
75    Date: 2003/09/04 16:30:00 CDT
76    Tag: blume-20030905-slices-etc
77    Description:
78    
79      1. Changed interface to vectors and arrays in Basis to match
80         (draft) Basis spec.
81      2. Added signatures and implementations of slices according to
82         Basis spec.
83      3. Edited source code throughout the system to make it compile again
84         under 1. and 2.  (In some cases code had to be added to have it
85         match the new signatures.)
86      4. MLRISC should be backward-compatible: the copies of the originals
87         of files that needed to change under 3. were retained, the .cm files
88         check the compiler version number and use old versions when
89         appropriate.
90      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
91         dirstream -> string option (in accordance with Basis spec).
92      6. When generating code that counts lines, ml-lex used function
93         CharVector.foldli, taking advantage of its old interface.
94         This has been replaced with the corresponding code from
95         CharVectorSlice.  (html-lex must be re-lexed!)
96      7. BitArray in smlnj-lib/Util has been extended/modified to match the
97         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
98      8. Removed temporary additions (fromInternal, toInternal) from the
99         (now obsolete) IntInf in smlnj-lib/Util.
100      9. Cleaned up structure Byte.
101     10. Added localOffset, scan, and fromString to Date (according to spec).
102         Cleaned/corrected implementation of Date.
103         (Still need to check for correctness; implement better canonicalizeDate.)
104     11. Added "scan" to signature IEEE_REAL.
105     12. Some improvements to IntInf [in particular: efficiency-hack for
106         mod and rem when second operand is 2 (for parity checks).]
107     13. Changed representation of type Time.time, using a single IntInf.int
108         value counting microseconds.  This considerably simplified the
109         implementation of structure Time.  We now support negative time
110         values; scan and fromString handle signs.
111     14. Functor PrimIO now takes two additional arguments (VectorSlice and
112         ArraySlice).
113    
114    ----------------------------------------------------------------------
115    Name: Matthias Blume (blume (at) tti - c (dot) org)
116    Date: 2003/08/28 17:00:00 CDT
117    Tag: blume-20030828-intinf
118    Description:
119    
120    This is a major update which comes with a version number bump
121    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
122    and an implementation of IntInf in the Basis.
123    
124    There are a fairly large number of related changes and updates throughout
125    the system:
126    
127    Basis:
128            - Implemented IntInf.
129            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
130            - Added some missing Real64 operations, most notably Real.toLargeInt.
131            - Added FixedInt as a synonym for Int32.
132    
133    compiler:
134            * Added support for a built-in intinf type.
135              - literals
136              - pattern matching
137              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
138              - overloading on literals and operations
139    
140            This required adding a primitive type intinf, some additional
141            primops, and implementations for several non-trivial intinf
142            operations in Core.  (The intinf type is completely abstract
143            to the compiler; all operations get delegated back to the Core.)
144    
145            * Intinf equality is handled by polyequal.  However, the compiler
146            does not print its usual warning in this case (since polyequal
147            is the right thing to do there).
148    
149            * Improved the organization of structure InlineT.
150    
151            * A word about conversion primops:
152            If conversions involving intinf do not cancel out during
153            CPS contract, then the compiler must insert calls to Core functions.
154            Since all core access must be resolved already during the FLINT
155            translate phase, it would be too late a the time of CPS contract
156            to add new Core calls.  For this reason, conversion primops
157            for intinf carry two arguments:  1. the numeric argument that
158            they are supposed to convert, and 2. the Core function that
159            can help with this conversion if necessary.  If CPS contract
160            eliminates a primop, then the associated Core function becomes
161            dead and goes away.  Intinf conversion primops that do not get
162            eliminated by CPS contract get rewritten into calls of their
163            core functions by a separate, new phase.
164    
165    interactive system:
166            - Control.Print.intinfDepth controls max length of intinf constants
167            being printed. (Analogous to Control.Print.stringDepth.)
168            - Cleanup in printutil and pputil: got rid of unused stuff and
169            duplicates; replaced some of the code with code that makes better
170            use of library functionality.
171    
172    CM:
173            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
174            are no longer silent.
175    
176    CKIT:
177            Fixed mismatched uses of Int32 and LargeInt.  I always decided
178            in favor of LargeInt -- which is now the same as IntInf.
179            CKIT-knowledgable people should check whether this is what's
180            intended and otherwise change things back to using Int32 or
181            FixedInt.
182    
183    Throughout the code:
184            Started using IntInf.int literals and built-in operations
185            (e.g., comparison with 0) where this seems appropriate.
186    
187    
188    ----------------------------------------------------------------------
189    Name: Dave MacQueen (dbm@cs.uchicago.edu)
190    Date: 2003/08/13 11:36:00 CDT
191    Tag: dbm-20030813-mcz-merge1
192    Description:
193    
194    Merging changes from the mcz-branch development branch into trunk.
195    These changes involve replacement of the emulated old prettyprinter
196    interface with direct use of the SML/NJ Lib PP library, and fixing
197    of a couple of bugs (895, 1186) relating to error messages.  A new
198    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
199    has been added.
200    
201    ----------------------------------------------------------------------
202    Name: Matthias Blume (blume (at) tti - c (dot) org)
203    Date: 2003/08/11 15:45:00 CDT
204    Tag: blume-20030811-windows
205    Description:
206    
207    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
208    
209        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
210    
211    This patch restores SML/NJ's ability to run under win32.  There are a
212    number of changes, including fixes for several bugs that had gone
213    unnoticed until now:
214    
215      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
216        trying to run the win32 version from within cygwin.)
217    
218      - There are a number of simple .bat scripts that substitute for their
219        corresponding Unix shell-scripts.  (See below.)
220    
221      - The internals of ml-build have been modified slightly.  The main
222        difference is that instead of calling ".link-sml" (or link-sml.bat)
223        using OS.Process.system, the ML process delegates this task back
224        to the script.  Otherwise problems arise in mixed environments such
225        as Cygwin where scripts look and work like Unix scripts, but
226        where OS.Process.system cannot run them.
227    
228      - In CM, the srcpath pickler used native pathname syntax -- which
229        is incorrect in the case of cross-compilation.  The new pickle format
230        is independent of platform-specific naming conventions.
231    
232      - Path configuration files (such as lib/pathconfig) can now choose
233        between native and standard syntax.  Placing a line of the form
234    
235            standard!
236    
237        into the file causes all subsequent paths to be interpreted using
238        CM standard pathname syntax (= Unix conventions); a line
239    
240            native!
241    
242        switches back to native style.  This was needed so that
243        path config files can be written portably, see src/system/pathconfig.
244    
245      - Runtime system:
246    
247          - win32-filesys.c:  get_file_time and set_file_time now
248            access modification time, not creation time.
249    
250          - I/O code made aware of new array representation.
251    
252          - Bug fixes in X86.prim.masm.
253    
254       - src/system/makeml made aware of win32. (For use under cygwin
255            and other Unix-environments for windows.)
256    
257       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
258         which caused BinIO.inputAll to fail consistently.
259    
260    .bat scripts:
261    
262       Windows .bat scripts assume that SMLNJ_HOME is defined.
263    
264       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
265         applications (sml, ml-yacc, ml-lex).
266       - ml-build.bat: analogous to ml-build.
267       - config\install.bat:  Analogous to config/install.sh.  This requires
268         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
269         (nmake etc. must be on the path, and vcvars32 must have been run.)
270         Moreover, sources for ml-lex and ml-yacc need to exist under src,
271         and the bootfile hierarchy must have been unpacked under
272         sml.boot.x86-win32.
273         The script is very primitive and does a poor job at error checking.
274         It only installs the base system, ml-lex, and ml-yacc.  No other
275         libraries are being installed (i.e., you get only those that
276         are part of the compiler.)
277       - link-sml.bat:  analogous to .link-sml, but not currently used
278    
279    Unrelated bug fixes:
280    
281      - ml-nlffigen now exports structures ST_* corresponding to incomplete
282        types.
283      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
284        otherwise.)
285    
286    ----------------------------------------------------------------------
287    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
288    Date: 2003/06/17
289    Tag: macqueen-20030617-bug895
290    Description:
291    
292    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
293    Tag will be used for new development branch (mcz-branch) for
294    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
295    
296    ----------------------------------------------------------------------
297    Name: Matthias Blume (blume (at) tti - c (dot) org)
298    Date: 2003/05/27 16:55:00 CDT
299    Tag: blume-20030527-polyeq
300    Description:
301    
302    Tried to eliminated most cases of polymorphic equality.
303    
304    ----------------------------------------------------------------------
305    Name: Matthias Blume (blume (at) tti - c (dot) org)
306    Date: 2003/05/21 17:45:00 CDT
307    Tag: blume-20030517-complete
308    Description:
309    
310    Two changes:
311    
312      1. Added a flag for controlling whether non-exhaustive bindings will
313         be treated as errors (default is false).
314      2. Cleaned up the *entire* source tree so that CMB.make goes through
315         without a single non-exhaustive match- or bind warning.
316    
317    ----------------------------------------------------------------------
318    Name: Matthias Blume (blume (at) tti - c (dot) org)
319    Date: 2003/05/17 10:20:00 CDT
320    Tag: blume-20030517-absyn
321    Description:
322    
323    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
324    
325       This mainly affects the quality of error messages.  However, some
326       of the code is now more straightforward than before. (Treatment of
327       the above four constructs in translate.sml is much simpler than
328       the "macro-expansion" that was going on before.  Plus, the mach-
329       compiler no longer gets invoked just to be able to compile an
330       if-expression.)
331    
332    2. The ErrorMsg.Error exception is now caught and absorbed by the
333       interactive loop.
334    
335    ----------------------------------------------------------------------
336    Name: Allen Leung
337    Date: 2003/05/16 13:05:00 CDT
338    Tag: leunga-20030516-cygwin-runtime
339    Description:
340    
341        Ported the runtime system to cygwin, which uses the unix
342    x86-unix bin files.  Missing/buggy features:
343    
344       o getnetbyname, getnetbyaddr: these functions seem to be missing in
345         the Cygwin library.
346       o Ctrl-C handling may be flaky.
347       o Windows system calls and Windows I/O are not supported.
348    
349       A new set of binfiles is located at:
350    
351         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
352    
353       This is only needed for bootstrapping the cygwin version of smlnj.
354       Other x86 versions can use the existing binfiles.
355    
356    ----------------------------------------------------------------------
357    Name: Matthias Blume
358    Date: 2003/04/08 15:42:00 CDT
359    Tag: blume-20030408-listpair
360    Description:
361    
362    1. Added a target 'mlrisc' to installer.
363    
364    2. Added missing elements to structure ListPair.
365    
366    ----------------------------------------------------------------------
367    Name: Allen Leung
368    Date: 2003/01/07 10:40:00 EST
369    Tag: leunga-20030107-int-rem
370    Description:
371    
372    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
373    The arguments to the SUBL instruction were swapped.
374    
375    ----------------------------------------------------------------------
376    Name: Matthias Blume
377    Date: 2002/12/12 16:25:00 EST
378    Tag: blume-20021212-risc-ra
379    Description:
380    
381    Fixed a serious bug in the rewrite code for FP spilling/reloading that
382    sent the RA into an infinite loop when floating point registers get
383    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
384    110.37 and 110.38.)
385    There was another set of potential problems related to the handling of
386    MLRISC annotations (but those did not yet cause real problems, apparently).
387    
388    ----------------------------------------------------------------------
389    Name: Matthias Blume
390    Date: 2002/12/06 22:40:00 EST
391    Tag: blume-20021206-cm-fileid
392    Description:
393    
394    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
395    This fixes the problem of CM getting confused by files that suddenly
396    change their identity (e.g., by getting unlinked and recreated by some
397    text editor such as vi).  There might be a better/cheaper/cleaner way
398    of doing this, but for now this will have to do.
399    
400    ----------------------------------------------------------------------
401    Name: Matthias Blume
402    Date: 2002/10/28 09:50:00 EST
403    Tag: blume-20021028-typecheck
404    Description:
405    
406    Exported structure Typecheck from $smlnj/viscomp/core.cm.
407    
408    ----------------------------------------------------------------------
409    Name: Matthias Blume
410    Date: 2002/10/17 09:10:00 EDT
411    Tag: Release_110_42
412    Description:
413    
414    In good old tradition, there has been a slight hiccup so that we have
415    to patch 110.42 after the fact.   The old release tag has been replaced
416    (see below).
417    
418    The change solves a problem with two competing approaches the
419    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
420    each other's way.
421    
422    This change only affects the runtime system code and the installer script.
423    (No new bootfiles.)
424    
425    ----------------------------------------------------------------------
426    Name: Matthias Blume
427    Date: 2002/10/16 12:00:00 EDT
428    Tag: Release_110_42_removed
429    Description:
430    
431    New working release.  New bootfiles.
432    
433    ----------------------------------------------------------------------
434    Name: Matthias Blume
435    Date: 2002/10/10 13:10:00 EDT
436    Tag: blume-20021010-ppc-divs
437    Description:
438    
439    The mltree operator DIVS must be implemented with an overflow check on
440    the PPC because the hardware indicates divide-by-zero using "overflow" as
441    well.
442    
443    ----------------------------------------------------------------------
444    Name: Matthias Blume
445    Date: 2002/07/23 11:45:00
446    Tag: blume-20020723-smlnj-home
447    Description:
448    
449    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
450    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
451    CM_PATHCONFIG is also set), the path configuration file is assumed
452    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
453    the entire tree to some other place and everything will "just work".
454    
455    (Companion commands such as ml-build and ml-makedepend also sense this
456    variable.)
457    
458    ----------------------------------------------------------------------
459    Name: Matthias Blume
460    Date: 2002/07/12 21:19:00 EDT
461    Tag: blume-20020712-liveness
462    Description:
463    
464    Exported two useful "step" functions from liveness module (MLRISC).
465    
466    ----------------------------------------------------------------------
467    Name: Matthias Blume
468    Date: 2002/07/05 16:00 EDT
469    Tag: Release_110_41
470    Description:
471    
472    New working release.  New bootfiles.
473    
474    ----------------------------------------------------------------------
475    Name: Matthias Blume
476    Date: 2002/07/05 10:25:00 EDT
477    Tag: blume-20020705-btimp
478    Description:
479    
480    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
481    other clients can set up backtracing support.
482    
483    ----------------------------------------------------------------------
484    Name: Matthias Blume
485    Date: 2002/06/25 17:23:00 EDT
486    Tag: blume-20020625-fpmax
487    Description:
488    
489    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
490    case.  (The sense of the isNaN test was reversed -- which made min and
491    max always return their first argument.)
492    
493    ----------------------------------------------------------------------
494    Name: Matthias Blume
495    Date: 2002/06/11
496    Tag: blume-20020611-unixpath
497    Description:
498    
499    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
500    
501    ----------------------------------------------------------------------
502    Name: Matthias Blume
503    Date: 2002/06/10 16:35:00 EDT
504    Tag: blume-20020610-ieeereal
505    Description:
506    
507    I back-ported my implementation of IEEEReal.fromString from the
508    idlbasis-devel branch so that we can test it.
509    
510    Another small change is that ppDec tries to give more information
511    than just "<sig>" in the case of functors.  However, this code is
512    broken in some mysterious way if the functor's body's signature
513    has not been declared by ascription but gets inferred from the
514    implementation.  This needs fixing...
515    
516    ----------------------------------------------------------------------
517    Name: Matthias Blume
518    Date: 2002/05/31
519    Tag: blume-20020531-btrace-mode
520    Description:
521    
522    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
523    the wayside when I switched over to using Controls everywhere.)
524    
525    ----------------------------------------------------------------------
526    Name: Lal George
527    Date: 2002/05/23 12:21:40 EDT
528    Tag: george-20020523-visual-labels
529    Description:
530    
531    Labels are now displayed in the graphical output to make
532    the fall-through and target blocks obvious.
533    
534    ----------------------------------------------------------------------
535    Name: Matthias Blume
536    Date: 2002/05/22 11:03:00 EDT
537    Tag: blume-20020522-shrink
538    Description:
539    
540    John tweaked yesterday's fix for 1131 to handle an out-of-memory
541    situation that comes up when allocating huge arrays.
542    
543    ----------------------------------------------------------------------
544    Name: Matthias Blume
545    Date: 2002/05/21 16:00:00 EDT
546    Tag: Release_110_40
547    Description:
548    
549    New working release (110.40).  New bootfiles.
550    
551    [Also: John Reppy fixed GC bug 1131.]
552    
553    ----------------------------------------------------------------------
554    Name: Matthias Blume
555    Date: 2002/05/21 12:35:00 EDT
556    Tag: blume-20020521-cmdoc
557    Description:
558    
559    CM documentation update.
560    
561    ----------------------------------------------------------------------
562    Name: Matthias Blume
563    Date: 2002/05/21 10:55:00 EDT
564    Tag: blume-20020521-misc
565    Description:
566    
567    - John tweaked runtime to be silent on heap export (except when
568      GC messages are on).
569    - I added a few more things (cross-compiling versions of CMB) to
570      config/preloads (as suggestions).
571    
572    ----------------------------------------------------------------------
573    Name: Matthias Blume
574    Date: 2002/05/20 22:25:00 EDT
575    Tag: blume-20020520-controls
576    Description:
577    
578    - Added ControlUtil structure to control-lib.cm.
579    - Use it throughout.
580    - Used Controls facility to define MLRISC controls (as opposed to
581      registering MLRISC control ref cells with Controls after the fact)
582    - Fixed messed-up controls priorities.
583    
584    * Removed again all the stuff from config/preloads that one wouldn't
585      be able to preload at the time the initial heap image is built.
586      (Many libraries, e.g., CML, do not exist yet at this time.  The
587       only libraries that can be preloaded via config/preloads are those
588       that come bundled with the bootfiles.)
589    
590    ----------------------------------------------------------------------
591    Name: Matthias Blume
592    Date: 2002/05/20 10:59:00 EDT
593    Tag: blume-20020520-preloads
594    Description:
595    
596    Added a lot of commented-out suggestions for things to be included
597    in config/preloads.
598    
599    ----------------------------------------------------------------------
600    Name: Allen Leung
601    Date: 2002/05/18 14:20:00 EDT
602    Tag: leunga-20020518-mdl
603    Description:
604    
605    o Made the mdl tool stuff compile and run again.
606    o I've disabled all the stuff that depends on RTL specifications; they
607      are all badly broken anyway.
608    
609    ----------------------------------------------------------------------
610    Name: Matthias Blume
611    Date: 2002/05/17 16:49:00 EDT
612    Tag: blume-20020517-controls
613    Description:
614    
615    0. John Reppy made several modifications to the SML/NJ library.
616       In particular, there is a shiny new controls-lib.cm.
617    
618    1. Pushed new controls interface through compiler so that everything
619       compiles again.
620    
621    2. Added FormatComb and FORMAT_COMB to the CML version of the
622       SML/NJ library (so that CML compiles again).
623    
624    3. Modified init scripts because XXX_DEFAULT environment variables
625       are no longer with us.  (Boot-time initialization is now done
626       using the same environment variables that are also used for
627       startup-time initialization of controls.)
628    
629    ----------------------------------------------------------------------
630    Name: Lal George
631    Date: 2002/05/15 09:20:10 EDT
632    Tag: george-20020515-pseudo-op-decls
633    Description:
634    
635            All pseudo-ops emitted before the first segment declaration
636    such as TEXT, DATA, and BSS directives are assumed to be global
637    declarations and are emitted first in the assembly file. This is
638    useful in a number of situations where one has pseudo-ops that are not
639    specific to any segment, and also works around the constraint that one
640    cannot have client pseudo-ops in the TEXT segment.
641    
642            Because no segment is associated with these declarations it is
643    an error to allocate any space or objects before the first segment
644    directive and an exception will be raised. However, we cannot make
645    this check for client pseudo-ops.
646    
647            These top level declarations are a field in the CFG graph_info.
648    In theory you can continue to add to this field after the CFG has been
649    built -- provided you know what you are doing;-)
650    
651    ----------------------------------------------------------------------
652    Name: Matthias Blume
653    Date: 2002/05/13 16:40:00 EDT
654    Tag: blume-20020513-pp-etc
655    Description:
656    
657    A few minor bugfixes:
658    
659      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
660        (Bogus printouts for redefined bindings still occur.  Compiler
661        bug should no longer occur now.  We need to redo the prettyprinter
662        from scratch.)
663    
664      - CM pathname printer now also adds escape sequences for ( and )
665    
666      - commend and docu fixes for ml-nlffi
667    
668    ----------------------------------------------------------------------
669    Name: Matthias Blume
670    Date: 2002/05/10 16:40:00 EDT
671    Tag: blume-20020510-erg-textio
672    Description:
673    
674    Applied the following bugfix provided by Emden Gansner:
675    
676        Output is corrupted when outputSubstr is used rather than output.
677    
678        The problem occurs when a substring
679    
680            ss = (s, dataStart, dataLen)
681    
682        where dataStart > 0, fills a stream buffer with avail bytes left.
683        avail bytes of s, starting at index dataStart, are copied into the
684        buffer, the buffer is flushed, and then the remaining dataLen-avail
685        bytes of ss are copied into the beginning of the buffer. Instead of
686        starting this copy at index dataStart+avail in s, the current code
687        starts the copy at index avail.
688    
689        Fix:
690        In text-io-fn.sml, change line 695 from
691             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
692        to
693             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
694    
695    ----------------------------------------------------------------------
696    Name: Matthias Blume
697    Date: 2002/04/12 13:55:00 EDT
698    Tag: blume-20020412-assyntax
699    Description:
700    
701    1. Grabbed newer assyntax.h from the XFree86 project.
702    2. Fiddled with how to compile X86.prim.asm without warnings.
703    3. (Very) Minor cleanup in CM.
704    
705    ----------------------------------------------------------------------
706    Name: Matthias Blume
707    Date: 2002/04/01 (no joke!) 17:07:00 EST
708    Tag: blume-20020401-x86div
709    Description:
710    
711    Added full support for div/mod/rem/quot on the x86, using the machine
712    instruction's two results (without clumsily recomputing the remainder)
713    directly where appropriate.
714    
715    Some more extensive power-of-two support was added to the x86 instruction
716    selector (avoiding expensive divs, mods, and muls where they can be
717    replaced with cheaper shifts and masks).  However, this sort of thing
718    ought to be done earlier, e.g., within the CPS optimizer so that
719    all architectures benefit from it.
720    
721    The compiler compiles to a fixed point, but changes might be somewhat
722    fragile nevertheless.  Please, report any strange things that you might
723    see wrt. div/mod/quot/rem...
724    
725    ----------------------------------------------------------------------
726    Name: Matthias Blume
727    Date: 2002/03/29 17:22:00
728    Tag: blume-20020329-div
729    Description:
730    
731    Fixed my broken div/mod logic.  Unfortunately, this means that the
732    inline code for div/mod now has one more comparison than before.
733    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
734    The problem was with quotient = 0, because that alone does not tell
735    us which way the rounding went.  One then has to look at whether
736    remainder and divisor have the same sign...  :(
737    
738    Anyway, I replaced the bootfiles with fresh ones...
739    
740    ----------------------------------------------------------------------
741    Name: Matthias Blume
742    Date: 2002/03/29 14:10:00 EST
743    Tag: blume-20020329-inlprims
744    Description:
745    
746    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
747    
748    Primops have changed. This means that the bin/boot-file formats have
749    changed as well.
750    
751    To make sure that there is no confusion, I made a new version.
752    
753    
754    CHANGES:
755    
756    * removed REMT from mltree (remainder should never overflow).
757    
758    * added primops to deal with divisions of all flavors to the frontend
759    
760    * handled these primops all the way through so they map to their respective
761      MLRISC support
762    
763    * used these primops in the implementation of Int, Int32, Word, Word32
764    
765    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
766    
767    * parameterized INLMIN, INLMAX, and INLABS by a numkind
768    
769    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
770      floating point
771    
772    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
773      and Real (but Real.abs maps to a separate floating-point-only primop)
774    
775    
776    TODO items:
777    
778    * Hacked Alpha32 instruction selection, disabling the selection of REMx
779      instructions because the machine instruction encoder cannot handle
780      them.  (Hppa, PPC, and Sparc instruction selection did not handle
781      REM in the first place, and REM is supported by the x86 machine coder.)
782    
783    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
784      selection phase.  (The two can be streamlined because the hardware
785      delivers both quotient and remainder at the same time anyway.)
786    
787    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
788      (Currently the behavior is inconsistent both across architectures and
789      wrt. the draft Basis spec.)
790    
791    * Word8 should eventually be handled natively, too.
792    
793    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
794      as if there currently is no execution path that could trigger it in
795      SML/NJ.  (The assumptions underlying functions arith and promotable do not
796      hold for things like multiplication and division.)
797    
798    ----------------------------------------------------------------------
799    Name: Matthias Blume
800    Date: 2002/03/27 16:27:00 EST
801    Tag: blume-20020327-mlrisc-divisions
802    Description:
803    
804    Added support for all four division operations (ML's div, mod, quot,
805    and rem) to MLRISC.  In the course of doing so, I also rationalized
806    the naming (no more annoying switch-around of DIV and QUOT), by
807    parameterizing the operation by div_rounding_mode (which can be either
808    DIV_TO_ZERO or DIV_TO_NEGINF).
809    
810    The generic MLTreeGen functor takes care of compiling all four
811    operations down to only round-to-zero div.
812    
813    Missing pieces:
814    
815      * Doing something smarter than relying on MLTreeGen on architectures
816        like, e.g., the x86 where hardware division delivers both quotient and
817        remainder at the same time.  With this, the implementation of the
818        round-to-neginf operations could be further streamlined.
819    
820      * Remove inlining support for div/mod/rem from the frontend and replace it
821        with primops that get carried through to the backend.  Do this for all
822        int and word types.
823    
824    ----------------------------------------------------------------------
825    Name: Matthias Blume
826    Date: 2002/03/25 17:25:00 EST
827    Tag: blume-20020325-divmod
828    Description:
829    
830    I improved (hopefully without breaking them) the implementation of Int.div,
831    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
832    advantage of the following observations:
833    
834      Let  q = x quot y      r = x rem y
835           d = x div  y      m = x mod y
836    
837    where "quot" is the round-to-zero version of integer division that
838    hardware usually provides.  Then we have:
839    
840         r = x - q * y        where neither the * nor the - will overflow
841         d = if q >= 0 orelse x = q * y then q else q - 1
842                              where neither the * nor the - will overflow
843         m = if q >= 0 orelse r = 0 then r else r + y
844                              where the + will not overflow
845    
846    This results in substantial simplification of the generated code.
847    The following table shows the number of CFG nodes and edges generated
848    for
849            fun f (x, y) = x OPER y
850            (* with OPER \in div, mod, quot, rem *)
851    
852    
853        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
854        --------------------------------------------------------
855         div |         24 |         39 |         12 |         16
856         mod |         41 |         71 |         12 |         16
857        quot |          8 |         10 |          8 |         10
858         rem |         10 |         14 |          8 |         10
859    
860    
861    ----------------------------------------------------------------------
862    Name: Matthias Blume
863    Date: 2002/03/25 22:06:00 EST
864    Tag: blume-20020325-cprotobug
865    Description:
866    
867    Fixed a bug in cproto (c prototype decoder).
868    
869    ----------------------------------------------------------------------
870    Name: Matthias Blume
871    Date: 2002/03/25 16:00:00 EST
872    Tag: blume-20020325-raw-primops
873    Description:
874    
875    I did some cleanup to Allen's new primop code and
876    replaced yesterday's bootfiles with new ones.
877    (But they are stored in the same place.)
878    
879    ----------------------------------------------------------------------
880    Name: Matthias Blume
881    Date: 2002/03/24 22:40:00 EST
882    Tag: blume-20020324-bootfiles
883    Description:
884    
885    Made the bootfiles that Allen asked for.
886    
887    ----------------------------------------------------------------------
888    Name: Allen Leung
889    Date: 2002/03/23 15:50:00 EST
890    Tag: leunga-20020323-flint-cps-rcc-primops
891    Description:
892    
893      1. Changes to FLINT primops:
894    
895        (* make a call to a C-function;
896         * The primop carries C function prototype information and specifies
897         * which of its (ML-) arguments are floating point. C prototype
898         * information is for use by the backend, ML information is for
899         * use by the CPS converter. *)
900      | RAW_CCALL of { c_proto: CTypes.c_proto,
901                       ml_args: ccall_type list,
902                       ml_res_opt: ccall_type option,
903                       reentrant : bool
904                     } option
905       (* Allocate uninitialized storage on the heap.
906        * The record is meant to hold short-lived C objects, i.e., they
907        * are not ML pointers.  With the tag, the representation is
908        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
909        *)
910      | RAW_RECORD of {tag:bool,sz:int}
911      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
912    
913      2.  These CPS primops are now overloaded:
914    
915           rawload of {kind:numkind}
916           rawstore of {kind:numkind}
917    
918          The one argument form is:
919    
920             rawload {kind} address
921    
922          The two argument form is:
923    
924             rawload {kind} [ml object, byte-offset]
925    
926      3. RAW_CCALL/RCC now takes two extra arguments:
927    
928         a. The first is whether the C call is reentrant, i.e., whether
929            ML state should be saved and restored.
930         b. The second argument is a string argument specifying the name of
931            library and the C function.
932    
933         These things are currently not handled in the code generator, yet.
934    
935      4. In CProto,
936    
937         An encoding type of "bool" means "ml object" and is mapped into
938         C prototype of PTR.  Note that "bool" is different than "string",
939         even though "string" is also mapped into PTR, because "bool"
940         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
941    
942      5. Pickler/unpicker
943    
944         Changed to handle RAW_RECORD and newest RAW_CCALL
945    
946      6. MLRiscGen,
947    
948         1. Changed to handle the new rawload/rawstore/rawrecord operators.
949         2. Code for handling C Calls has been moved to a new module CPSCCalls,
950            in the file CodeGen/cpscompile/cps-c-calls.sml
951    
952      7. Added the conditional move operator
953    
954             condmove of branch
955    
956         to cps.  Generation of this is still buggy so it is currently
957         disabled.
958    
959    ----------------------------------------------------------------------
960    Name: Lal George
961    Date: 2002/03/22 14:18:25 EST
962    Tag: george-20020322-cps-branch-prob
963    Description:
964    
965    Implemented the Ball-Larus branch prediction-heuristics, and
966    incorporated graphical viewers for control flow graphs.
967    
968    Ball-Larus Heuristics:
969    ---------------------
970    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
971    
972    By design it uses the Dempster-Shafer theory for combining
973    probabilities.  For example, in the function:
974    
975        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
976    
977    the ball-larus heuristics predicts that the n=0 is unlikely
978    (OH-heuristic), and the 'then' branch is unlikely because of the
979    RH-heuristic -- giving the 'then' branch an even lower combined
980    probability using the Dempster-Shafer theory.
981    
982    Finally, John Reppy's loop analysis in MLRISC, further lowers the
983    probability of the 'then' branch because of the loop in the else
984    branch.
985    
986    
987    Graphical Viewing:
988    ------------------
989    I merely plugged in Allen's graphical viewers into the compiler. The
990    additional code is not much. At the top level, saying:
991    
992            Control.MLRISC.getFlag "cfg-graphical-view" := true;
993    
994    will display the graphical view of the control flow graph just before
995    back-patching.  daVinci must be in your path for this to work. If
996    daVinci is not available, then the default viewer can be changed
997    using:
998    
999            Control.MLRISC.getString "viewer"
1000    
1001    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1002    course, these viewers must be in your path.
1003    
1004    The above will display the compilation unit at the level of clusters,
1005    many of which are small, boring, and un-interesting. Also setting:
1006    
1007            Control.MLRISC.getInt "cfg-graphical-view_size"
1008    
1009    will display clusters that are larger than the value set by the above.
1010    
1011    
1012    ----------------------------------------------------------------------
1013    Name: Matthias Blume
1014    Date: 2002/03/21 22:20:00 EST
1015    Tag: blume-20020321-kmp-bugfix
1016    Description:
1017    
1018    Changed the interface to the KMP routine in PreString and fixed
1019    a minor bug in one place where it was used.
1020    
1021    ----------------------------------------------------------------------
1022    Name: Allen Leung
1023    Date: 2002/03/21 20:30:00 EST
1024    Tag: leunga-20020321-cfg
1025    Description:
1026    
1027      Fixed a potential problem in cfg edge splitting.
1028    
1029    ----------------------------------------------------------------------
1030    Name: Allen Leung
1031    Date: 2002/03/21 17:15:00 EST
1032    Tag: leunga-20020321-x86-fp-cfg
1033    Description:
1034    
1035      1. Recoded the buggy parts of x86-fp.
1036    
1037         a. All the block reordering code has been removed.
1038            We now depend on the block placement phases to do this work.
1039    
1040         b. Critical edge splitting code has been simplified and moved into the
1041            CFG modules, as where they belong.
1042    
1043         Both of these were quite buggy and complex.  The code is now much, much
1044         simpler.
1045    
1046      2. X86 backend.
1047    
1048         a. Added instructions for 64-bit support.  Instruction selection for
1049            64-bit has not been committed, however, since that
1050            requires changes to MLTREE which haven't been approved by
1051            Lal and John.
1052    
1053         b. Added support for FUCOMI and FUCOMIP when generating code for
1054            PentiumPro and above.  We only generate these instructions in
1055            the fast-fp mode.
1056    
1057         c. Added cases for JP and JNP in X86FreqProps.
1058    
1059      3. CFG
1060    
1061         CFG now has a bunch of methods for edge splitting and merging.
1062    
1063      4. Machine description.
1064    
1065         John's simplification of MLTREE_BASIS.fcond broke a few machine
1066         description things:
1067    
1068         rtl-build.{sig,sml} and hppa.mdl fixed.
1069    
1070         NOTE: the machine description stuff in the repository is still broken.
1071               Again, I can't put my fixes in because that involves
1072               changes to MLTREE.
1073    
1074    ----------------------------------------------------------------------
1075    Name: Matthias Blume
1076    Date: 2002/03/20 15:55:00 EST
1077    Tag: blume-20020320-kmp
1078    Description:
1079    
1080    Implemented Knuth-Morris-Pratt string matching in PreString and used
1081    it for String.isSubstring, Substring.isSubstring, and
1082    Substring.position.
1083    
1084    (Might need some stress-testing.  Simple examples worked fine.)
1085    
1086    ----------------------------------------------------------------------
1087    Name: Matthias Blume
1088    Date: 2002/03/19 16:37:00 EST
1089    Tag: blume-20020319-witnesses
1090    Description:
1091    
1092    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1093    
1094    This implements a generic mechanism for changing constness qualifiers
1095    anywhere within big C types without resorting to outright "casts".
1096    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1097    modify the constness at the outermost level.)
1098    The implementation of "convert" is based on the idea of "witness"
1099    values -- values that are not used by the operation but whose types
1100    "testify" to their applicability.  On the implementation side, "convert"
1101    is simply a projection (returning its second curried argument).  With
1102    cross-module inlining, it should not result in any machine code being
1103    generated.
1104    
1105    ----------------------------------------------------------------------
1106    Name: Matthias Blume
1107    Date: 2002/03/15 16:40:00 EST
1108    Tag: blume-20020315-basis
1109    Description:
1110    
1111    Provided (preliminary?) implementations for
1112    
1113      {String,Substring}.{concatWith,isSuffix,isSubstring}
1114    
1115    and
1116    
1117      Substring.full
1118    
1119    Those are in the Basis spec but they were missing in SML/NJ.
1120    
1121    ----------------------------------------------------------------------
1122    Name: Matthias Blume
1123    Date: 2002/03/14 21:30:00 EST
1124    Tag: blume-20020314-controls
1125    Description:
1126    
1127    Controls:
1128    ---------
1129    
1130    1. Factored out the recently-added Controls : CONTROLS stuff and put
1131       it into its own library $/controls-lib.cm.  The source tree for
1132       this is under src/smlnj-lib/Controls.
1133    
1134    2. Changed the names of types and functions in this interface, so they
1135       make a bit more "sense":
1136    
1137          module -> registry
1138          'a registry -> 'a group
1139    
1140    3. The interface now deals in ref cells only.  The getter/setter interface
1141       is (mostly) gone.
1142    
1143    4. Added a function that lets one register an already-existing ref cell.
1144    
1145    5. Made the corresponding modifications to the rest of the code so that
1146       everything compiles again.
1147    
1148    6. Changed the implementation of Controls.MLRISC back to something closer
1149       to the original.  In particular, this module (and therefore MLRISC)
1150       does not depend on Controls.  There now is some link-time code in
1151       int-sys.sml that registers the MLRISC controls with the Controls
1152       module.
1153    
1154    CM:
1155    ---
1156    
1157      * One can now specify the lambda-split aggressiveness in init.cmi.
1158    
1159    ----------------------------------------------------------------------
1160    Name: Allen Leung
1161    Date: 2002/03/13 17:30:00 EST
1162    Tag: leunga-20020313-x86-fp-unary
1163    Description:
1164    
1165    Bug fix for:
1166    
1167    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1168    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1169    > - fun f(x,(y,z)) = Real.~ y;
1170    > [autoloading]
1171    > [autoloading done]
1172    >       fchsl   (%eax), 184(%esp)
1173    > Error: MLRisc bug: X86MCEmitter.emitInstr
1174    >
1175    > uncaught exception Error
1176    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1177    
1178    The problem was that the code generator did not generate any fp registers
1179    in this case, and the ra didn't know that it needed to run the X86FP phase to
1180    translate the pseudo fp instruction.   This only happened with unary fp
1181    operators in certain situations.
1182    
1183    ----------------------------------------------------------------------
1184    Name: Matthias Blume
1185    Date: 2002/03/13 14:00:00 EST
1186    Tag: blume-20020313-overload-etc
1187    Description:
1188    
1189    1. Added _overload as a synonym for overload for backward compatibility.
1190       (Control.overloadKW must be true for either version to be accepted.)
1191    
1192    2. Fixed bug in install script that caused more things to be installed
1193       than what was requested in config/targets.
1194    
1195    3. Made CM aware of the (_)overload construct so that autoloading
1196       works.
1197    
1198    ----------------------------------------------------------------------
1199    Name: Matthias Blume
1200    Date: 2002/03/12 22:03:00 EST
1201    Tag: blume-20020312-url
1202    Description:
1203    
1204    Forgot to update BOOT and srcarchiveurl.
1205    
1206    ----------------------------------------------------------------------
1207    Name: Matthias Blume
1208    Date: 2002/03/12 17:30:00 EST
1209    Tag: blume-20020312-version110392
1210    Description:
1211    
1212    Yet another version number bump (because of small changes to the
1213    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1214    
1215    Changes:
1216    
1217      The new pid generation scheme described a few weeks ago was overly
1218      complicated.  I implemented a new mechanism that is simpler and
1219      provides a bit more "stability":  Once CM has seen a compilation
1220      unit, it keeps its identity constant (as long as you do not delete
1221      those crucial CM/GUID/* files).  This means that when you change
1222      an interface, compile, then go back to the old interface, and
1223      compile again, you arrive at the original pid.
1224    
1225      There now also is a mechanism that instructs CM to use the plain
1226      environment hash as a module's pid (effectively making its GUID
1227      the empty string).  For this, "noguid" must be specified as an
1228      option to the .sml file in question within its .cm file.
1229      This is most useful for code that is being generated by tools such
1230      as ml-nlffigen (because during development programmers tend to
1231      erase the tool's entire output directory tree including CM's cached
1232      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1233      revert to the old, broken behavior of SML/NJ, but in specific cases
1234      where there is no danger of interface confusion, its use is ok
1235      (I think).
1236    
1237      ml-nlffigen by default generates "noguid" annotations.  They can be
1238      turned off by specifying -guid in its command line.
1239    
1240    ----------------------------------------------------------------------
1241    Name: Lal George
1242    Date: 2002/03/12 12 14:42:36 EST
1243    Tag: george-20020312-frequency-computation
1244    Description:
1245    
1246    Integrated jump chaining and static block frequency into the
1247    compiler. More details and numbers later.
1248    
1249    ----------------------------------------------------------------------
1250    Name: Lal George
1251    Date: 2002/03/11 11 22:38:53 EST
1252    Tag: george-20020311-jump-chain-elim
1253    Description:
1254    
1255    Tested the jump chain elimination on all architectures (except the
1256    hppa).  This is on by default right now and is profitable for the
1257    alpha and x86, however, it may not be profitable for the sparc and ppc
1258    when compiling the compiler.
1259    
1260    The gc test will typically jump to a label at the end of the cluster,
1261    where there is another jump to an external cluster containing the actual
1262    code to invoke gc. This is to allow factoring of common gc invocation
1263    sequences. That is to say, we generate:
1264    
1265            f:
1266               testgc
1267               ja   L1      % jump if above to L1
1268    
1269            L1:
1270               jmp L2
1271    
1272    
1273    After jump chain elimination the 'ja L1' instructions is converted to
1274    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1275    up being implemented in their long form (if L2 is far away) using:
1276    
1277            jbe     L3      % jump if below or equal to L3
1278            jmp     L2
1279         L3:
1280            ...
1281    
1282    
1283    For large compilation units L2  may be far away.
1284    
1285    
1286    ----------------------------------------------------------------------
1287    Name: Matthias Blume
1288    Date: 2002/03/11 13:30:00 EST
1289    Tag: blume-20020311-mltreeeval
1290    Description:
1291    
1292    A functor parameter was missing.
1293    
1294    ----------------------------------------------------------------------
1295    Name: Allen Leung
1296    Date: 2002/03/11 10:30:00 EST
1297    Tag: leunga-20020311-runtime-string0
1298    Description:
1299    
1300       The representation of the empty string now points to a
1301    legal null terminated C string instead of unit.  It is now possible
1302    to convert an ML string into C string with InlineT.CharVector.getData.
1303    This compiles into one single machine instruction.
1304    
1305    ----------------------------------------------------------------------
1306    Name: Allen Leung
1307    Date: 2002/03/10 23:55:00 EST
1308    Tag: leunga-20020310-x86-call
1309    Description:
1310    
1311       Added machine generation for CALL instruction (relative displacement mode)
1312    
1313    ----------------------------------------------------------------------
1314    Name: Matthias Blume
1315    Date: 2002/03/08 16:05:00
1316    Tag: blume-20020308-entrypoints
1317    Description:
1318    
1319    Version number bumped to 110.39.1.  NEW BOOTFILES!
1320    
1321    Entrypoints: non-zero offset into a code object where execution should begin.
1322    
1323    - Added the notion of an entrypoint to CodeObj.
1324    - Added reading/writing of entrypoint info to Binfile.
1325    - Made runtime system bootloader aware of entrypoints.
1326    - Use the address of the label of the first function given to mlriscGen
1327      as the entrypoint.  This address is currently always 0, but it will
1328      not be 0 once we turn on block placement.
1329    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
1330      with entry points) from mlriscGen.
1331    
1332    ----------------------------------------------------------------------
1333    Name: Allen Leung
1334    Date: 2002/03/07 20:45:00 EST
1335    Tag: leunga-20020307-x86-cmov
1336    Description:
1337    
1338       Bug fixes for CMOVcc on x86.
1339    
1340       1. Added machine code generation for CMOVcc
1341       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1342       3. CMOVcc cannot have an immediate operand as argument.
1343    
1344    ----------------------------------------------------------------------
1345    Name: Matthias Blume
1346    Date: 2002/03/07 16:15:00 EST
1347    Tag: blume-20020307-controls
1348    Description:
1349    
1350    This is a very large but mostly boring patch which makes (almost)
1351    every tuneable compiler knob (i.e., pretty much everything under
1352    Control.* plus a few other things) configurable via both the command
1353    line and environment variables in the style CM did its configuration
1354    until now.
1355    
1356    Try starting sml with '-h' (or, if you are brave, '-H')
1357    
1358    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1359    implements the underlying generic mechanism.
1360    
1361    The interface to some of the existing such facilities has changed somewhat.
1362    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1363    (The getFoo interface is still there for backward-compatibility, but its
1364    use is deprecated.)
1365    
1366    The ml-build script passes -Cxxx=yyy command-line arguments through so
1367    that one can now twiddle the compiler settings when using this "batch"
1368    compiler.
1369    
1370    TODO items:
1371    
1372    We should go through and throw out all controls that are no longer
1373    connected to anything.  Moreover, we should go through and provide
1374    meaningful (and correct!) documentation strings for those controls
1375    that still are connected.
1376    
1377    Currently, multiple calls to Controls.new are accepted (only the first
1378    has any effect).  Eventually we should make sure that every control
1379    is being made (via Controls.new) exactly once.  Future access can then
1380    be done using Controls.acc.
1381    
1382    Finally, it would probably be a good idea to use the getter-setter
1383    interface to controls rather than ref cells.  For the time being, both
1384    styles are provided by the Controls module, but getter-setter pairs are
1385    better if thread-safety is of any concern because they can be wrapped.
1386    
1387    *****************************************
1388    
1389    One bug fix: The function blockPlacement in three of the MLRISC
1390    backpatch files used to be hard-wired to one of two possibilities at
1391    link time (according to the value of the placementFlag).  But (I
1392    think) it should rather sense the flag every time.
1393    
1394    *****************************************
1395    
1396    Other assorted changes (by other people who did not supply a HISTORY entry):
1397    
1398    1. the cross-module inliner now works much better (Monnier)
1399    2. representation of weights, frequencies, and probabilities in MLRISC
1400       changed in preparation of using those for weighted block placement
1401       (Reppy, George)
1402    
1403    ----------------------------------------------------------------------
1404    Name: Lal George
1405    Date: 2002/03/07 14:44:24 EST 2002
1406    Tag: george-20020307-weighted-block-placement
1407    
1408    Tested the weighted block placement optimization on all architectures
1409    (except the hppa) using AMPL to generate the block and edge frequencies.
1410    Changes were required in the machine properties to correctly
1411    categorize trap instructions. There is an MLRISC flag
1412    "weighted-block-placement" that can be used to enable weighted block
1413    placement, but this will be ineffective without block/edge
1414    frequencies (coming soon).
1415    
1416    
1417    ----------------------------------------------------------------------
1418    Name: Lal George
1419    Date: 2002/03/05 17:24:48 EST
1420    Tag: george-20020305-linkage-cluster
1421    
1422    In order to support the block placement optimization, a new cluster
1423    is generated as the very first cluster (called the linkage cluster).
1424    It contains a single jump to the 'real' entry point for the compilation
1425    unit. Block placement has no effect on the linkage cluster itself, but
1426    all the other clusters  have full freedom in the manner in which they
1427    reorder blocks or functions.
1428    
1429    On the x86 the typical linkage code that is generated is:
1430       ----------------------
1431            .align 2
1432       L0:
1433            addl    $L1-L0, 72(%esp)
1434            jmp     L1
1435    
1436    
1437            .align  2
1438       L1:
1439       ----------------------
1440    
1441    72(%esp) is the memory location for the stdlink register. This
1442    must contain the address of the CPS function being called. In the
1443    above example, it contains the address of  L0; before
1444    calling L1 (the real entry point for the compilation unit), it
1445    must contain the address for L1, and hence
1446    
1447            addl $L1-L0, 72(%esp)
1448    
1449    I have tested this on all architectures except the hppa.The increase
1450    in code size is of course negligible
1451    
1452    ----------------------------------------------------------------------
1453    Name: Allen Leung
1454    Date: 2002/03/03 13:20:00 EST
1455    Tag: leunga-20020303-mlrisc-tools
1456    
1457      Added #[ ... ] expressions to mlrisc tools
1458    
1459    ----------------------------------------------------------------------
1460    Name: Matthias Blume
1461    Date: 2002/02/27 12:29:00 EST
1462    Tag: blume-20020227-cdebug
1463    Description:
1464    
1465    - made types in structure C and C_Debug to be equal
1466    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1467    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1468    
1469    ----------------------------------------------------------------------
1470    Name: Matthias Blume
1471    Date: 2002/02/26 12:00:00 EST
1472    Tag: blume-20020226-ffi
1473    Description:
1474    
1475    1. Fixed a minor bug in CM's "noweb" tool:
1476       If numbering is turned off, then truly don't number (i.e., do not
1477       supply the -L option to noweb).  The previous behavior was to supply
1478       -L'' -- which caused noweb to use the "default" line numbering scheme.
1479       Thanks to Chris Richards for pointing this out (and supplying the fix).
1480    
1481    2. Once again, I reworked some aspects of the FFI:
1482    
1483       A. The incomplete/complete type business:
1484    
1485       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
1486         gone!
1487       - ML types representing an incomplete type are now *equal* to
1488         ML types representing their corresponding complete types (just like
1489         in C).  This is still safe because ml-nlffigen will not generate
1490         RTTI for incomplete types, nor will it generate functions that
1491         require access to such RTTI.   But when ML code generated from both
1492         incomplete and complete versions of the C type meet, the ML types
1493         are trivially interoperable.
1494    
1495         NOTE:  These changes restore the full generality of the translation
1496         (which was previously lost when I eliminated functorization)!
1497    
1498       B. Enum types:
1499    
1500       - Structure C now has a type constructor "enum" that is similar to
1501         how the "su" constructor works.  However, "enum" is not a phantom
1502         type because each "T enum" has values (and is isomorphic to
1503         MLRep.Signed.int).
1504       - There are generic access operations for enum objects (using
1505         MLRep.Signed.int).
1506       - ml-nlffigen will generate a structure E_foo for each "enum foo".
1507         * The structure contains the definition of type "mlrep" (the ML-side
1508         representation type of the enum).  Normally, mlrep is the same
1509         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
1510         then mlrep will be defined as a datatype -- thus facilitating
1511         pattern matching on mlrep values.
1512         ("-ec" will be suppressed if there are duplicate values in an
1513          enumeration.)
1514         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
1515         will be generated for each C enum constant xxx.
1516         * Conversion functions m2i and i2m convert between mlrep and
1517         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
1518         * Coversion functions c and ml convert between mlrep and "tag enum".
1519         * Access functions (get/set) fetch and store mlrep values.
1520       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
1521         enumerations are merged into one single enumeration represented by
1522         structure E_'.
1523    
1524    ----------------------------------------------------------------------
1525    Name: Allen Leung
1526    Date: 2002/02/25 04:45:00 EST
1527    Tag: leunga-20020225-cps-spill
1528    
1529    This is a new implementation of the CPS spill phase.
1530    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
1531    In case of problems, replace it with the old file spill.sml
1532    
1533    The current compiler runs into some serious performance problems when
1534    constructing a large record.  This can happen when we try to compile a
1535    structure with many items.  Even a very simple structure like the following
1536    makes the compiler slow down.
1537    
1538        structure Foo = struct
1539           val x_1 = 0w1 : Word32.int
1540           val x_2 = 0w2 : Word32.int
1541           val x_3 = 0w3 : Word32.int
1542           ...
1543           val x_N = 0wN : Word32.int
1544        end
1545    
1546    The following table shows the compile time, from N=1000 to N=4000,
1547    with the old compiler:
1548    
1549    N
1550    1000   CPS 100 spill                           0.04u  0.00s  0.00g
1551           MLRISC ra                               0.06u  0.00s  0.05g
1552              (spills = 0 reloads = 0)
1553           TOTAL                                   0.63u  0.07s  0.21g
1554    
1555    1100   CPS 100 spill                           8.25u  0.32s  0.64g
1556           MLRISC ra                               5.68u  0.59s  3.93g
1557              (spills = 0 reloads = 0)
1558           TOTAL                                   14.71u  0.99s  4.81g
1559    
1560    1500   CPS 100 spill                           58.55u  2.34s  1.74g
1561           MLRISC ra                               5.54u  0.65s  3.91g
1562              (spills = 543 reloads = 1082)
1563           TOTAL                                   65.40u  3.13s  6.00g
1564    
1565    2000   CPS 100 spill                           126.69u  4.84s  3.08g
1566           MLRISC ra                               0.80u  0.10s  0.55g
1567              (spills = 42 reloads = 84)
1568           TOTAL                                   129.42u  5.10s  4.13g
1569    
1570    3000   CPS 100 spill                           675.59u  19.03s  11.64g
1571           MLRISC ra                               2.69u  0.27s  1.38g
1572              (spills = 62 reloads = 124)
1573           TOTAL                                   682.48u  19.61s  13.99g
1574    
1575    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
1576           MLRISC ra                               4.96u  0.27s  2.72g
1577              (spills = 85 reloads = 170)
1578           TOTAL                                   2375.26u  57.21s  48.00g
1579    
1580    As you can see the old cps spill module suffers from some serious
1581    performance problem.  But since I cannot decipher the old code fully,
1582    instead of patching the problems up, I'm reimplementing it
1583    with a different algorithm.  The new code is more modular,
1584    smaller when compiled, and substantially faster
1585    (O(n log n) time and O(n) space).  Timing of the new spill module:
1586    
1587    4000  CPS 100 spill                           0.02u  0.00s  0.00g
1588          MLRISC ra                               0.25u  0.02s  0.15g
1589             (spills=1 reloads=3)
1590          TOTAL                                   7.74u  0.34s  1.62g
1591    
1592    Implementation details:
1593    
1594    As far as I can tell, the purpose of the CPS spill module is to make sure the
1595    number of live variables at any program point (the bandwidth)
1596    does not exceed a certain limit, which is determined by the
1597    size of the spill area.
1598    
1599    When the bandwidth is too large, we decrease the register pressure by
1600    packing live variables into spill records.  How we achieve this is
1601    completely different than what we did in the old code.
1602    
1603    First, there is something about the MLRiscGen code generator
1604    that we should be aware of:
1605    
1606    o MLRiscGen performs code motion!
1607    
1608       In particular, it will move floating point computations and
1609       address computations involving only the heap pointer to
1610       their use sites (if there is only a single use).
1611       What this means is that if we have a CPS record construction
1612       statement
1613    
1614           RECORD(k,vl,w,e)
1615    
1616       we should never count the new record address w as live if w
1617       has only one use (which is often the case).
1618    
1619       We should do something similar to floating point, but the transformation
1620       there is much more complex, so I won't deal with that.
1621    
1622    Secondly, there are now two new cps primops at our disposal:
1623    
1624     1. rawrecord of record_kind option
1625        This pure operator allocates some uninitialized storage from the heap.
1626        There are two forms:
1627    
1628         rawrecord NONE [INT n]  allocates a tagless record of length n
1629         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
1630                                     and initializes the tag.
1631    
1632     2. rawupdate of cty
1633          rawupdate cty (v,i,x)
1634          Assigns to x to the ith component of record v.
1635          The storelist is not updated.
1636    
1637    We use these new primops for both spilling and increment record construction.
1638    
1639     1. Spilling.
1640    
1641        This is implemented with a linear scan algorithm (but generalized
1642        to trees).  The algorithm will create a single spill record at the
1643        beginning of the cps function and use rawupdate to spill to it,
1644        and SELECT or SELp to reload from it.  So both spills and reloads
1645        are fine-grain operations.  In contrast, in the old algorithm
1646        "spills" have to be bundled together in records.
1647    
1648        Ideally, we should sink the spill record construction to where
1649        it is needed.  We can even split the spill record into multiple ones
1650        at the places where they are needed.  But CPS is not a good
1651        representation for global code motion, so I'll keep it simple and
1652        am not attempting this.
1653    
1654     2. Incremental record construction (aka record splitting).
1655    
1656        Long records with many component values which are simulatenously live
1657        (recall that single use record addresses are not considered to
1658         be live) are constructed with rawrecord and rawupdate.
1659        We allocate space on the heap with rawrecord first, then gradually
1660        fill it in with rawupdate.  This is the technique suggested to me
1661        by Matthias.
1662    
1663        Some restrictions on when this is applicable:
1664        1. It is not a VECTOR record.  The code generator currently does not handle
1665           this case. VECTOR record uses double indirection like arrays.
1666        2. All the record component values are defined in the same "basic block"
1667           as the record constructor.  This is to prevent speculative
1668           record construction.
1669    
1670    ----------------------------------------------------------------------
1671    Name: Allen Leung
1672    Date: 2002/02/22 01:02:00 EST
1673    Tag: leunga-20020222-mlrisc-tools
1674    
1675    Minor bug fixes in the parser and rewriter
1676    
1677    ----------------------------------------------------------------------
1678    Name: Allen Leung
1679    Date: 2002/02/21 20:20:00 EST
1680    Tag: leunga-20020221-peephole
1681    
1682    Regenerated the peephole files.  Some contained typos in the specification
1683    and some didn't compile because of pretty printing bugs in the old version
1684    of 'nowhere'.
1685    
1686    ----------------------------------------------------------------------
1687    Name: Allen Leung
1688    Date: 2002/02/19 20:20:00 EST
1689    Tag: leunga-20020219-mlrisc-tools
1690    Description:
1691    
1692       Minor bug fixes to the mlrisc-tools library:
1693    
1694       1.  Fixed up parsing colon suffixed keywords
1695       2.  Added the ability to shut the error messages up
1696       3.  Reimplemented the pretty printer and fixed up/improved
1697           the pretty printing of handle and -> types.
1698       4.  Fixed up generation of literal symbols in the nowhere tool.
1699       5.  Added some SML keywords to to sml.sty
1700    
1701    ----------------------------------------------------------------------
1702    Name: Matthias Blume
1703    Date: 2002/02/19 16:20:00 EST
1704    Tag: blume-20020219-cmffi
1705    Description:
1706    
1707    A wild mix of changes, some minor, some major:
1708    
1709    * All C FFI-related libraries are now anchored under $c:
1710        $/c.cm      --> $c/c.cm
1711        $/c-int.cm  --> $c/internals/c-int.cm
1712        $/memory.cm --> $c/memory/memory.cm
1713    
1714    * "make" tool (in CM) now treats its argument pathname slightly
1715      differently:
1716        1. If the native expansion is an absolute name, then before invoking
1717           the "make" command on it, CM will apply OS.Path.mkRelative
1718           (with relativeTo = OS.FileSys.getDir()) to it.
1719        2. The argument will be passed through to subsequent phases of CM
1720           processing without "going native".  In particular, if the argument
1721           was an anchored path, then "make" will not lose track of that anchor.
1722    
1723    * Compiler backends now "know" their respective C calling conventions
1724      instead of having to be told about it by ml-nlffigen.  This relieves
1725      ml-nlffigen from one of its burdens.
1726    
1727    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
1728    
1729    * Export C_DEBUG and C_Debug from $c/c.cm.
1730    
1731    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
1732      subtyping relationship between incomplete pointers and their complete
1733      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
1734      with the convention of instantiating 'o with ('t, 'c) obj whenever
1735      the pointer target type is complete.  In the incomplete case, 'o
1736      will be instantiated with some "'c iobj" -- a type obtained by
1737      using one of the functors PointerToIncompleteType or PointerToCompleteType.
1738    
1739      Operations that work on both incomplete and complete pointer types are
1740      typed as taking an 'o ptr while operations that require the target to
1741      be known are typed as taking some ('t, 'c) obj ptr.
1742    
1743      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
1744      where void is an eqtype without any values.  This makes it possible
1745      to work on voidptr values using functions meant to operate on light
1746      incomplete pointers.
1747    
1748    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
1749      been vastly simplified.
1750    
1751    ----------------------------------------------------------------------
1752    Name: Matthias Blume
1753    Date: 2002/02/19 10:48:00 EST
1754    Tag: blume-20020219-pqfix
1755    Description:
1756    
1757    Applied Chris Okasaki's bug fix for priority queues.
1758    
1759    ----------------------------------------------------------------------
1760    Name: Matthias Blume
1761    Date: 2002/02/15 17:05:00
1762    Tag: Release_110_39
1763    Description:
1764    
1765    Last-minute retagging is becoming a tradition... :-(
1766    
1767    This is the working release 110.39.
1768    
1769    ----------------------------------------------------------------------
1770    Name: Matthias Blume
1771    Date: 2002/02/15 16:00:00 EST
1772    Tag: Release_110_39-orig
1773    Description:
1774    
1775    Working release 110.39.  New bootfiles.
1776    
1777    (Update: There was a small bug in the installer so it wouldn't work
1778    with all shells.  So I retagged. -Matthias)
1779    
1780    ----------------------------------------------------------------------
1781    Name: Matthias Blume
1782    Date: 2002/02/15 14:17:00 EST
1783    Tag: blume-20020215-showbindings
1784    Description:
1785    
1786    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
1787    the latter can be useful for exploring what bindings are available at
1788    the interactive prompt.  (The first function returns only the list
1789    of symbols that are really bound, the second prints those but also the
1790    ones that CM's autoloading mechanism knows about.)
1791    
1792    ----------------------------------------------------------------------
1793    Name: Matthias Blume
1794    Date: 2002/02/15 12:08:00 EST
1795    Tag: blume-20020215-iptrs
1796    Description:
1797    
1798    Two improvements to ml-nlffigen:
1799    
1800      1. Write files only if they do not exist or if their current contents
1801         do not coincide with what's being written.  (That is, avoid messing
1802         with the time stamps unless absolutely necessary.)
1803    
1804      2. Implement a "repository" mechanism for generated files related
1805         to "incomplete pointer types".   See the README file for details.
1806    
1807    ----------------------------------------------------------------------
1808    Name: Matthias Blume
1809    Date: 2002/02/14 11:50:00 EST
1810    Tag: blume-20020214-quote
1811    Description:
1812    
1813    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
1814    because of the new and improved tag generation scheme.  (Thanks to Allen
1815    Leung for pointing it out.)
1816    
1817    ----------------------------------------------------------------------
1818    Name: Lal George
1819    Date: 2002/02/14 09:55:27 EST 2002
1820    Tag: george-20020214-isabelle-bug
1821    Description:
1822    
1823    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
1824    of Isabelle on the x86.
1825    
1826    From Allen:
1827    -----------
1828     I've found the problem:
1829    
1830         in ra-core.sml, I use the counter "blocked" to keep track of the
1831         true number of elements in the freeze queue.  When the counter goes
1832         to zero, I skip examining the queue.  But I've messed up the
1833         bookkeeping in combine():
1834    
1835             else ();
1836             case !ucol of
1837               PSEUDO => (if !cntv > 0 then
1838                     (if !cntu > 0 then blocked := !blocked - 1 else ();
1839                                        ^^^^^^^^^^^^^^^^^^^^^^^
1840                      moveu := mergeMoveList(!movev, !moveu)
1841                     )
1842                  else ();
1843    
1844         combine() is called to coalesce two nodes u and v.
1845         I think I was thinking that if the move counts of u and v are both
1846         greater than zero then after they are coalesced then one node is
1847         removed from the freeze queue.  Apparently I was thinking that
1848         both u and v are of low degree, but that's clearly not necessarily true.
1849    
1850    
1851    02/12/2002:
1852        Here's the patch.  HOL now compiles.
1853    
1854        I don't know how this impact on performance (compile
1855        time or runtime).  This bug caused the RA (especially on the x86)
1856        to go thru the potential spill phase when there are still nodes on the
1857        freeze queue.
1858    
1859    
1860    
1861    
1862    ----------------------------------------------------------------------
1863    Name: Matthias Blume
1864    Date: 2002/02/13 22:40:00 EST
1865    Tag: blume-20020213-fptr-rtti
1866    Description:
1867    
1868    Fixed a bug in ml-nlffigen that was introduced with one of the previous
1869    updates.
1870    
1871    ----------------------------------------------------------------------
1872    Name: Matthias Blume
1873    Date: 2002/02/13 16:41:00 EST
1874    Tag: blume-20020213-cmlpq
1875    Description:
1876    
1877    Added new priority queue export symbols (which have just been added to
1878    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
1879    not compile and the installer would choke.)
1880    
1881    ----------------------------------------------------------------------
1882    Name: Matthias Blume
1883    Date: 2002/02/13 16:15:00 EST
1884    Tag: blume-20020213-various
1885    Description:
1886    
1887    1. More tweaks to ml-nlffigen:
1888    
1889       - better internal datastructures (resulting in slight speedup)
1890       - "-match" option requires exact match
1891       - "localized" gensym counters (untagged structs/unions nested within
1892         other structs/unions or within typedefs get a fresh counter; their
1893         tag will be prefixed by a concatenation of their parents' tags)
1894       - bug fixes (related to calculation of transitive closure of types
1895         to be included in the output)
1896    
1897    2. Minor Basis updates:
1898    
1899       - added implementations for List.collate and Option.app
1900    
1901    ----------------------------------------------------------------------
1902    Name: Matthias Blume
1903    Date: 2002/02/11 15:55:00 EST
1904    Tag: blume-20020211-gensym
1905    Description:
1906    
1907    Added a "-gensym" option to command line of ml-nlffigen.  This can be
1908    used to specify a "stem" -- a string that is inserted in all "gensym'd"
1909    names (ML structure names that correspond to unnamed C structs, unions,
1910    and enums), so that separate runs of ml-nlffigen do not clash.
1911    
1912    ----------------------------------------------------------------------
1913    Name: Matthias Blume
1914    Date: 2002/02/11 12:05:00 EST
1915    Tag: blume-20020211-gensml
1916    Description:
1917    
1918    A quick fix for a problem with GenSML (in the pgraph-util library):
1919    Make generation of toplevel "local" optional.  (Strictly speaking,
1920    signature definitions within "local" are not legal SML.)
1921    
1922    Other than that: updates to INSTALL and cm/TODO.
1923    
1924    ----------------------------------------------------------------------
1925    Name: Matthias Blume
1926    Date: 2002/02/08 15:00:00 EST
1927    Tag: blume-20020208-uniquepid
1928    Description:
1929    
1930    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
1931    
1932    1. The installer (config/install.sh) has gotten smarter:
1933    
1934         - Configuration options are a bit easier to specify now
1935           (in config/targets).
1936         - Bug in recognizing .tar.bz2 files fixed.
1937         - Installer automatically resolves dependencies between
1938           configuration options (e.g., if you ask for eXene, you will
1939           also get cml -- regardless whether you asked for it or not).
1940         - Installer can run in "quieter mode" by setting the environment
1941           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
1942           "completely silent", though.
1943         - Build HashCons library as part of smlnj-lib.
1944    
1945    2. A new scheme for assigning persistent identifiers to compilation
1946       units (and, by extension, to types etc.) has been put into place.
1947       This fixes a long-standing bug where types and even dynamic values
1948       can get internally confused, thereby compromising type safety
1949       (abstraction) and dynamic correctness.  See
1950    
1951         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
1952    
1953       for an example of how things could go wrong until now.
1954    
1955       The downside of the new scheme is that pids are not quite as
1956       persistent as they used to be: CM will generate a fresh pid
1957       for every compilation unit that it thinks it sees for the first
1958       time.  That means that if you compile starting from a clean, fresh
1959       source tree at two different times, you end up with different
1960       binaries.
1961    
1962       Cutoff recompilation, however, has not been compromised because
1963       CM keeps pid information in special caches between runs.
1964    
1965    ----------------------------------------------------------------------
1966    Name: Lal George
1967    Date: 2002/02/07 15:34:13 EST 2002
1968    Tag: <none>
1969    Description:
1970    
1971    Compilers that generate assembly code may produce  global labels
1972    whose value is resolved at link time. The various peephole optimization
1973    modules did not take this in account.
1974    
1975    TODO. The Labels.addrOf function should really return an option
1976    type so that clients are forced to deal with this issue, rather
1977    than an exception being raised.
1978    
1979    ----------------------------------------------------------------------
1980    Name: Lal George
1981    Date: 2002/02/06 13:55:02 EST
1982    Tag: george-20020206-ra-breakup
1983    Description:
1984    
1985    1. A bug fix from Allen.
1986    
1987        A typo causes extra fstp %st(0)'s to be generated at compensation
1988        edges, which might cause stack underflow traps at runtime.  This
1989        occurs in fft where there are extraneous fstps right before the 'into'
1990        trap instruction (in this case they are harmless since none of the
1991        integers overflow.)
1992    
1993    2. Pulled out various utility modules that were embedded in the modules
1994       of the register allocator. I need these modules for other purposes, but
1995       they are not complete enough to put into a library (just yet).
1996    ----------------------------------------------------------------------
1997    Name: Matthias Blume
1998    Date: 2002/01/31 16:05:00 EST
1999    Tag: blume-20020131-sparc-ccalls
2000    Description:
2001    
2002    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2003       of extra stack space by mistake.  Fixed.
2004    
2005    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2006    
2007    ----------------------------------------------------------------------
2008    Name: Allen Leung
2009    Date: 2002/01/30
2010    Tag: leunga-20020130-nowhere-bug-fix
2011    Description:
2012    
2013       MLRISC bug fixes:
2014       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2015       2. MachineInt.fromString was negating its value.
2016    
2017    ----------------------------------------------------------------------
2018    Name: Matthias Blume
2019    Date: 2002/01/29
2020    Tag: blume-20020129-INSTALL
2021    Description:
2022    
2023    - Added somewhat detailed installation instructions (file INSTALL).
2024    - Fixed curl-detection bug in config/install.sh.
2025    - It is now possible to select the URL getter using the URLGETTER
2026      environment variable:
2027    
2028          not set / "unknown"      --> automatic detection (script tries wget,
2029                                       curl, and lynx)
2030          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2031                                       how to properly invoke them)
2032          other                    --> use $URLGETTER directly, it must take
2033                                       precisely two command-line arguments
2034                                       (source URL and destination file name)
2035    
2036    ----------------------------------------------------------------------
2037    Name: Matthias Blume
2038    Date: 2002/01/28
2039    Tag: blume-20020128-sparc-ccalls
2040    Description:
2041    
2042    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2043    - Make use of the allocParam argument in sparc-c-calls.
2044    
2045    ----------------------------------------------------------------------
2046    Name: Matthias Blume
2047    Date: 2002/01/28
2048    Tag: blume-20020128-allocParam
2049    Description:
2050    
2051    John Reppy:  Changes c-calls API to accept client-callback for
2052    allocating extra stack space.
2053    me: Corresponding changes to mlriscGen (using a dummy argument that
2054        does not change the current behavior).
2055    
2056    ----------------------------------------------------------------------
2057    Name: Matthias Blume
2058    Date: 2002/01/28 12:00:00
2059    Tag: Release_110_38
2060    Description:
2061    
2062    This time for real!!!
2063    
2064    ----------------------------------------------------------------------
2065    Name: Matthias Blume
2066    Date: 2002/01/28 10:56:00 EST
2067    Tag: blume-20020128-retraction
2068    Description:
2069    
2070    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
2071       with blume-Release_110_38-retracted.)
2072    
2073    1. Fixed a problem with incorrect rounding modes in real64.sml.
2074       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2075    
2076    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2077       and enums fixed.  The naming of corresponding ML identifiers should
2078       now be consistent again.
2079    
2080    ----------------------------------------------------------------------
2081    Name: Allen Leung
2082    Date: 2002/01/27
2083    Tag: leunga-20020127-nowhere
2084    Description:
2085    
2086       Added a target called nowhere in the configuration scripts.
2087       Enabling this will build the MLRISC 'nowhere' tool (for translating
2088       programs with where-clauses into legal SML code) during installation.
2089    
2090    ----------------------------------------------------------------------
2091    Name: Matthias Blume
2092    Date: 2002/01/25 21:27:00 EST
2093    Tag: blume-Release_110_38-retracted
2094    Description:
2095    
2096    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
2097    
2098    README will be added later.
2099    
2100    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
2101    (Release_110_38) removed.  Reason: Last-minute bug fixes.
2102    
2103    ----------------------------------------------------------------------
2104    Name: Matthias Blume
2105    Date: 2002/01/25
2106    Tag: blume-20020125-ffi
2107    Description:
2108    
2109    A large number of tweaks and improvements to ml-nlffi-lib and
2110    ml-nlffigen:
2111    
2112       - ML represenation types have been streamlined
2113       - getter and setter functions work with concrete values, not abstract
2114         ones where possible
2115       - ml-nlffigen command line more flexible (see README file there)
2116       - some bugs have been fixed (hopefully)
2117    
2118    ----------------------------------------------------------------------
2119    Name: Lal George
2120    Date: 2002/01/24
2121    Tag: george-20020124-risc-ra-interface
2122    Description:
2123    
2124       There is a dramatic simplification in the interface to the
2125       register allocator for RISC architectures as a result of making
2126       parallel copy instructions explicit.
2127    
2128    ----------------------------------------------------------------------
2129    Name: Matthias Blume
2130    Date: 2002/01/22
2131    Tag: blume-20020122-x86-ccalls
2132    Description:
2133    
2134    Bug fix for c-calls on x86 (having to do with how char- and
2135    short-arguments are being handled).
2136    
2137    ----------------------------------------------------------------------
2138    Name: Matthias Blume
2139    Date: 2002/01/21
2140    Tag: blume-20020121-ff
2141    Description:
2142    
2143    Another day of fiddling with the FFI...
2144    
2145    1. Bug fix/workaround:  CKIT does not complain about negative array
2146       dimensions, so ml-nlffigen has to guard itself against this possibility.
2147       (Otherwise a negative dimension would send it into an infinite loop.)
2148    
2149    2. Some of the abstract types (light objects, light pointers, most "base"
2150       types) in structure C are now eqtypes.
2151    
2152    3. Added constructors and test functions for NULL function pointers.
2153    
2154    ----------------------------------------------------------------------
2155    Name: Matthias Blume
2156    Date: 2002/01/18
2157    Tag: blume-20020118-ready-for-new-release
2158    Description:
2159    
2160    Made config/srcarchiveurl point to a new place.  (Will provide boot
2161    files shortly.)
2162    
2163    Maybe we christen this to be 110.38?
2164    
2165    ----------------------------------------------------------------------
2166    Name: Matthias Blume
2167    Date: 2002/01/18
2168    Tag: blume-20020118-more-ffifiddle
2169    Description:
2170    
2171    Today's FFI fiddling:
2172    
2173      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
2174        C.Set.* that use concrete (MLRep.*) arguments and results instead
2175        of abstract ones.
2176    
2177      - Provided word-style bit operations etc. for "int" representation
2178        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
2179        and Long.
2180    
2181    ----------------------------------------------------------------------
2182    Name: Matthias Blume
2183    Date: 2002/01/18
2184    Tag: blume-20020118-use-x86-fp
2185    Description:
2186    
2187    Now that x86-fast-fp seems to be working, I turned it back on again
2188    by default.  (Seems to work fine now, even with the FFI.)
2189    
2190    Other than that, I added some documentation about the FFI to
2191    src/ml-nlffigen/README and updated the FFI test examples in
2192    src/ml-nlffi-lib/Tests/*.
2193    
2194    ----------------------------------------------------------------------
2195    Name: Allen Leung
2196    Date: 2002/01/17
2197    Tag: leunga-20020117-x86-fast-fp-call
2198    Description:
2199    
2200       1. Fixed a problem with handling return fp values when x86's fast fp
2201          mode is turned on.
2202    
2203       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
2204          of %f32.
2205    
2206       3. Added a constructor INT32lit to the ast of MLRISC tools.
2207    
2208    ----------------------------------------------------------------------
2209    Name: Matthias Blume
2210    Date: 2002/01/16
2211    Tag: blume-20020116-ffifiddle
2212    Description:
2213    
2214    More fiddling with the FFI interface:
2215    
2216     - Make constness 'c instead of rw wherever possible.  This eliminates
2217       the need for certain explicit coercions.  (However, due to ML's
2218       value polymorphism, there will still be many cases where explicit
2219       coercions are necessary.  Phantom types are not the whole answer
2220       to modeling a subtyping relationship in ML.)
2221    
2222     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
2223    
2224     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
2225       clumsy workaround for varargs problem.)
2226    
2227    ----------------------------------------------------------------------
2228    Name: Lal George
2229    Date: 2002/01/15
2230    Tag: <none>
2231    Description:
2232    
2233    1. Since COPY instructions are no longer native to the architecture,
2234       a generic functor can be used to implement the expandCopies function.
2235    
2236    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
2237       TEXT segment.
2238    
2239    ----------------------------------------------------------------------
2240    Name: Matthias Blume
2241    Date: 2002/01/15
2242    Tag: blume-20020115-ffiupdates
2243    Description:
2244    
2245    1. Fix for bug resulting in single-precision float values being returned
2246       incorrectly from FFI calls.
2247    
2248    2. Small modifications to C FFI API:
2249    
2250        - memory-allocation routines return straight objects (no options)
2251          and raise an exception in out-of-memory situations
2252        - unsafe extensions to cast between function pointers and pointers
2253          from/to ints
2254        - added structure C_Debug as an alternative to structure C where
2255          pointer-dereferencing (|*| and |*!) always check for null-pointers
2256        - added open_lib' to DynLinkage;  open_lib' works like open_lib
2257          but also takes a (possibly empty) list of existing library handles
2258          that the current library depends on
2259    
2260    ----------------------------------------------------------------------
2261    Name: Matthias Blume
2262    Date: 2002/01/10
2263    Tag: blume-20020110-newffigen
2264    Description:
2265    
2266    1. Updates to portable graph code.
2267    
2268    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
2269       more scalable now so that even huge interfaces such as the one
2270       for GTK compile in finite time and space. :-)
2271       See src/ml-nlffigen/README for details on what's new.
2272    
2273    ----------------------------------------------------------------------
2274    Name: Lal George
2275    Date: 2001/01/09 14:31:35 EST 2002
2276    Tag: george-20011206-rm-native-copy
2277    Description:
2278    
2279            Removed the native COPY and FCOPY instructions
2280            from all the architectures and replaced it with the
2281            explicit COPY instruction from the previous commit.
2282    
2283            It is now possible to simplify many of the optimizations
2284            modules that manipulate copies. This has not been
2285            done in this change.
2286    
2287    ----------------------------------------------------------------------
2288    Name: Lal George
2289    Date: 2001/12/06 16:50:13 EST 2001
2290    Tag: george-20011206-mlrisc-instruction
2291    Description:
2292    
2293    Changed the representation of instructions from being fully abstract
2294    to being partially concrete. That is to say:
2295    
2296      from
2297            type instruction
2298    
2299      to
2300            type instr                              (* machine instruction *)
2301    
2302            datatype instruction =
2303                LIVE of {regs: C.cellset, spilled: C.cellset}
2304              | KILL of {regs: C.cellset, spilled: C.cellset}
2305              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
2306              | ANNOTATION of {i: instruction, a: Annotations.annotation}
2307              | INSTR of instr
2308    
2309    This makes the handling of certain special instructions that appear on
2310    all architectures easier and uniform.
2311    
2312    LIVE and KILL say that a list of registers are live or killed at the
2313    program point where they appear. No spill code is generated when an
2314    element of the 'regs' field is spilled, but the register is moved to
2315    the 'spilled' (which is present, more for debugging than anything else).
2316    
2317    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
2318    We used to generate:
2319    
2320            DEFFREG f1
2321            f1 := f2 + f3
2322            trapb
2323    
2324    but now generate:
2325    
2326            f1 := f2 + f3
2327            trapb
2328            LIVE {regs=[f1,f2,f3], spilled=[]}
2329    
2330    Furthermore, the DEFFREG (hack) required that all floating point instruction
2331    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
2332    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
2333    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
2334    this architecture).
2335    
2336    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
2337    all the architectures. This will result in further simplification of the
2338    register allocator that must be aware of them for coalescing purposes, and
2339    will also simplify certain aspects of the machine description that provides
2340    callbacks related to parallel copies.
2341    
2342    ANNOTATION should be obvious, and now INSTR represents the honest to God
2343    machine instruction set!
2344    
2345    The <arch>/instructions/<arch>Instr.sml files define certain utility
2346    functions for making porting easier -- essentially converting upper case
2347    to lower case. All machine instructions (of type instr) are in upper case,
2348    and the lower case form generates an MLRISC instruction. For example on
2349    the alpha we have:
2350    
2351      datatype instr =
2352         LDA of {r:cell, b:cell, d:operand}
2353       | ...
2354    
2355      val lda : {r:cell, b:cell, d:operand} -> instruction
2356        ...
2357    
2358    where lda is just (INSTR o LDA), etc.
2359    
2360    ----------------------------------------------------------------------
2361    Name: Matthias Blume
2362    Date: 2001/11/22 21:40:00 EST
2363    Tag: Release_110_37
2364    Description:
2365    
2366    Release 110.37.  This time for real.
2367    
2368    ----------------------------------------------------------------------
2369    Name: Matthias Blume
2370    Date: 2001/11/21 16:35:00 EST
2371    Tag: blume-20011121-foot-in-mouth
2372    Description:
2373    
2374    Removed the "Release_110_37" tag because of a serious bug.
2375    This will be re-tagged once the bug is fixed.
2376    
2377    ----------------------------------------------------------------------
2378    Name: Matthias Blume
2379    Date: 2001/11/21 16:14:00 EST
2380    Tag: blume-20011121-forgottenfile
2381    Description:
2382    
2383    Forgot to add a file.  (Just a .tex-file -- part of
2384    the CM manual source.)
2385    
2386    ----------------------------------------------------------------------
2387    Name: Matthias Blume
2388    Date: 2001/11/21 16:10:00 EST
2389    Tag: blume-20011121-invalid_110_37
2390    Description:
2391    
2392    Note: I removed the original tag "Release_110_37" from this commit
2393          because we found a serious bug in all non-x86 backends.
2394          - Matthias
2395    
2396    1. Modifications to the SML/NJ code generator and to the runtime system
2397       so that code object name strings are directly inserted into code
2398       objects at code generation time.  The only business the runtime system
2399       has with this is now to read the name strings on occasions.
2400       (The encoding of the name string has also changed somewhat.)
2401    
2402    2. CM now implements a simple "set calculus" for specifying export lists.
2403       In particular, it is now possible to refer to the export lists of
2404       other libraries/groups/sources and form unions as well as differences.
2405       See the latest CM manual for details.
2406    
2407    3. An separate notion of "proxy" libraries has again be eliminated from
2408       CM's model.  (Proxy libraries are now simply a special case of using
2409       the export list calculus.)
2410    
2411    4. Some of the existing libraries now take advantage of the new set
2412       calculus.
2413       (Notice that not all libraries have been converted because some
2414       of the existing .cm-files are supposed to be backward compatible
2415       with 110.0.x.)
2416    
2417    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
2418       "General.exnMessage"!  The former relies on a certain hook to be
2419       initialized, and that often does not happen in the stand-alone case.)
2420    
2421    ----------------------------------------------------------------------
2422    Name: Lal George
2423    Date: 2001/11/21  13:56:18 EST
2424    Tag: george-2001121-pseudo-ops
2425    Description:
2426    
2427      Implemented a complete redesign of MLRISC pseudo-ops. Now there
2428      ought to never be any question of incompatabilities with
2429      pseudo-op syntax expected by host assemblers.
2430    
2431      For now, only modules supporting GAS syntax are implemented
2432      but more should follow, such as MASM, and vendor assembler
2433      syntax, e.g. IBM as, Sun as, etc.
2434    
2435    ----------------------------------------------------------------------
2436    Name: Matthias Blume
2437    Date: 2001/11/14 11:52:00 EST
2438    Tag: blume-20011114-srcname
2439    Description:
2440    
2441    1. Routed the name of the current source file to mlriscgen where it
2442       should be directly emitted into the code object.  (This last part
2443       is yet to be done.)
2444    
2445    2. Some cleanup of the pgraph code to make it match the proposal that
2446       I put out the other day.  (The proposal notwithstanding, things are
2447       still in flux here.)
2448    
2449    ----------------------------------------------------------------------
2450    Name: Lal George
2451    Date: 2001/11/14 09:44:04 EST
2452    Tag:
2453    Description:
2454    
2455      Fix for a backpatching bug reported by Allen.
2456    
2457      Because the boundary between short and long span-dependent
2458      instructions is +/- 128, there are an astounding number of
2459      span-dependent instructions whose size is over estimated.
2460    
2461      Allen came up with the idea of letting the size of span
2462      dependent instructions be non-monotonic, for a maxIter
2463      number of times, after which the size must be monotonically
2464      increasing.
2465    
2466      This table shows the number of span-dependent instructions
2467      whose size was over-estimated as a function of maxIter, for the
2468      file Parse/parse/ml.grm.sml:
2469    
2470         maxIter            # of instructions:
2471            10                      687
2472            20                      438
2473            30                      198
2474            40                        0
2475    
2476      In compiling the compiler, there is no significant difference in
2477      compilation speed between maxIter=10 and maxIter=40. Actually,
2478      my measurements showed that maxIter=40 was a tad faster than
2479      maxIter=10! Also 96% of the  files in the compiler reach a fix
2480      point within 13 iterations, so fixing maxIter at 40, while high,
2481      is okay.
2482    
2483    ----------------------------------------------------------------------
2484    Name: Matthias Blume
2485    Date: 2001/10/31 15:25:00 EST
2486    Tag: blume-20011031-pgraph
2487    Description:
2488    
2489    CKIT:
2490    * Changed the "Function" constructor of type Ast.ctype to carry optional
2491      argument identifiers.
2492    * Changed the return type of TypeUtil.getFunction accordingly.
2493    * Type equality ignores the argument names.
2494    * TypeUtil.composite tries to preserve argument names but gives up quickly
2495      if there is a mismatch.
2496    
2497    installation script:
2498    * attempts to use "curl" if available (unless "wget" is available as well)
2499    
2500    CM:
2501    * has an experimental implementation of "portable graphs" which I will
2502      soon propose as an implementation-independent library format
2503    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
2504    
2505    NLFFI-LIB:
2506    * some cleanup (all cosmetic)
2507    
2508    NLFFIGEN:
2509    * temporarily disabled the mechanism that suppresses ML output for
2510      C definitions whose identifiers start with an underscore character
2511    * generate val bindings for enum constants
2512    * user can request that only one style (light or heavy) is being used;
2513      default is to use both (command-line arguments: -heavy and -light)
2514    * fixed bug in handling of function types involving incomplete pointers
2515    * generate ML entry points that take record arguments (i.e., using
2516      named arguments) for C functions that have a prototype with named
2517      arguments
2518      (see changes to CKIT)
2519    
2520    ----------------------------------------------------------------------
2521    Name: Allen Leung
2522    Date: 2001/10/27 20:34:00 EDT
2523    Tag: leunga-20011027-x86-fast-fp-call
2524    Description:
2525    
2526       Fixed the bug described in blume-20010920-slowfp.
2527    
2528       The fix involves
2529          1. generating FCOPYs in FSTP in ia32-svid
2530          2. marking a CALL with the appropriate annotation
2531    
2532    ----------------------------------------------------------------------
2533    Name: Matthias Blume
2534    Date: 2001/10/16 11:32:00 EDT
2535    Tag: blume-20011016-netbsd
2536    Description:
2537    
2538    Underscore patch from Chris Richards (fixing problem with compiling
2539    runtime system under recent NetBSD).
2540    
2541    ----------------------------------------------------------------------
2542    Name: Allen Leung
2543    Date: 2001/10/12 17:18:32 EDT 2001
2544    Tag: leung-20011012-x86-printflowgraph
2545    Description:
2546    
2547    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
2548    
2549    ----------------------------------------------------------------------
2550    Name: Lal George
2551    Date: 2001/10/11 23:51:34 EDT
2552    Tag: george-20011011-too-many-instrs
2553    Description:
2554    
2555    The representation of a program point never expected to see more
2556    than 65536 instructions in a basic block!
2557    
2558    ----------------------------------------------------------------------
2559    Name: Lal George
2560    Date: 2001/10/09 09:41:37 EDT
2561    Tag: george-20011008-mlrisc-labels
2562    Description:
2563    
2564    Changed the machine description files to support printing of
2565    local and global labels in assembly code, based on host assembler
2566    conventions.
2567    
2568    ----------------------------------------------------------------------
2569    Name: Matthias Blume
2570    Date: 2001/09/25 15:25:00 EDT
2571    Tag: blume-20010925-exninfo
2572    Description:
2573    
2574    I provided a non-hook implementation of exnName (at the toplevel) and
2575    made the "dummy" implementation of exnMessage (at the toplevel) more
2576    useful: if nothing gets "hooked in", then at least you are going to
2577    see the exception name and a message indicating why you don't see more.
2578    
2579    [For the time being, programs that need exnMessage and want to use
2580    ml-build should either use General.exnMessage (strongly recommended) or
2581    refer to structure General at some other point so that CM sees a
2582    static dependency.]
2583    
2584    [Similar remarks go for "print" and "use":  If you want to use their
2585    functionality in stand-alone programs generated by ml-build, then use
2586    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
2587    
2588    ----------------------------------------------------------------------
2589    Name: Matthias Blume
2590    Date: 2001/09/20 17:28:00 EDT
2591    Tag: blume-20010920-slowfp
2592    Description:
2593    
2594    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
2595    
2596    ----------------------------------------------------------------------
2597    Name: Matthias Blume
2598    Date: 2001/09/20 17:20:00 EDT
2599    Tag: blume-20010920-canonicalpaths
2600    Description:
2601    
2602    0. Updated the BOOT file (something that I forgot to do earlier).
2603    
2604    1. Small internal change to CM so that it avoids "/../" in filenames
2605       as much as possible (but only where it is safe).
2606    
2607    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
2608       that arguments that contain delimiters are passed through correctly.
2609       This change also means that all "special" arguments of the form
2610       @SMLxxx... must come first.
2611    
2612    3. Changed install script to put relative anchor names for tool commands
2613       into pathconfig.
2614    
2615    ----------------------------------------------------------------------
2616    Name: Matthias Blume
2617    Date: 2001/09/18 15:35:00 EDT
2618    Tag: blume-20010918-readme11036
2619    Description:
2620    
2621    Added README files.
2622    
2623    ----------------------------------------------------------------------
2624    Name: Matthias Blume
2625    Date: 2001/09/18 11:45:00 EDT
2626    Tag: Release_110_36 (retag)
2627    Description:
2628    
2629    Fixed mistake in config/preloads. Retagged as 110.36.
2630    
2631    ----------------------------------------------------------------------
2632    Name: Matthias Blume
2633    Date: 2001/09/18 09:40:00 EDT
2634    Tag: Release_110_36_orig (tag changed)
2635    Description:
2636    
2637    New version (110.36).  New bootfiles.
2638    
2639    ----------------------------------------------------------------------
2640    Name: Matthias Blume
2641    Date: 2001/09/14 16:15:00 EDT
2642    Tag: blume-20010914-x86fastfp
2643    Description:
2644    
2645    John committed some changes that Allen made, in particular a (hopefully)
2646    correctly working version of the x86-fp module.
2647    
2648    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
2649    flag to "true".  Everything seems to compile to a fixpoint ok, and
2650    "mandelbrot" speeds up by about 15%.
2651    
2652    ----------------------------------------------------------------------
2653    Name: Matthias Blume
2654    Date: 2001/09/13 11:20:00 EDT
2655    Tag: blume-20010913-minimal
2656    Description:
2657    
2658    1. Stefan Monnier's patch to fix a miscompilation problem that
2659       was brought to light by John Reppy's work on Moby.
2660    
2661    2. Implemented a minimal "structure Compiler" that contains just
2662       "version" and "architecture".  The minimal version will be
2663       available when the full version is not.  This is for backward-
2664       compatibility with code that wants to test Compiler.version.
2665    
2666    ----------------------------------------------------------------------
2667    Name: Matthias Blume
2668    Date: 2001/08/28 14:03:00 EDT
2669    Tag: blume-20010828-ml-lex
2670    Description:
2671    
2672    Fix for bug 1581, received from Neophytos Michael.
2673    
2674    ----------------------------------------------------------------------
2675    Name: Matthias Blume
2676    Date: 2001/08/27 11:20:00 EDT
2677    Tag: blume-20010827-readme11035
2678    Description:
2679    
2680    Fleshed out the README file for 110.35.
2681    
2682    ----------------------------------------------------------------------
2683    Name: Matthias Blume
2684    Date: 2001/08/24 17:10:00 EDT
2685    Tag: Release_110_35
2686    Description:
2687    
2688    New version number (110.35).  New bootfiles.
2689    
2690    ----------------------------------------------------------------------
2691    Name: Lal George
2692    Date: 2001/08/24 13:47:18 EDT 2001
2693    Tag: george-20010824-MLRISC-graphs
2694    Description:
2695    
2696     removed clusters from MLRISC completely and replaced with graphs.
2697    
2698    ----------------------------------------------------------------------
2699    Name: Matthias Blume
2700    Date: 2001/08/23 17:50:00 EDT
2701    Tag: blume-20010823-toplevel
2702    Description:
2703    
2704    - some reorganization of the code that implements various kinds of
2705      environments in the compiler (static, dynamic, symbolic, combined)
2706    - re-implemented the EnvRef module so that evalStream works properly
2707      (if the stream contains references to "use", "CM.make", etc.)
2708    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
2709    
2710    ----------------------------------------------------------------------
2711    Name: Matthias Blume
2712    Date: 2001/08/20 15:50 EDT
2713    Tag: blume20010820-slipup
2714    Description:
2715    
2716    I forgot to commit a few files.  Here they are...
2717    
2718    ----------------------------------------------------------------------
2719    Name: Matthias Blume
2720    Date: 2001/08/20 15:35:00 EDT
2721    Tag: blume-20010820-debugprof
2722    Description:
2723    
2724    !!!! NEW BOOTFILES !!!!
2725    
2726    This is another round of reorganizing the compiler sources.  This
2727    time the main goal was to factor out all the "instrumentation"
2728    passes (for profiling and backtracing) into their own library.
2729    The difficulty was to do it in such a way that it does not depend
2730    on elaborate.cm but only on elabdata.cm.
2731    
2732    Therefore there have been further changes to both elaborate.cm and
2733    elabdata.cm -- more "generic" things have been moved from the former
2734    to the latter.  As a result, I was forced to split the assignment
2735    of numbers indicating "primtyc"s into two portions: SML-generic and
2736    SML/NJ-specific.  Since it would have been awkward to maintain,
2737    I bit the bullet and actually _changed_ the mapping between these
2738    numbers and primtycs.  The bottom line of this is that you need
2739    a new set of bin- and bootfiles.
2740    
2741    I have built new bootfiles for all architectures, so doing a fresh
2742    checkout and config/install.sh should be all you need.
2743    
2744    The newly created library's name is
2745    
2746        $smlnj/viscomp/debugprof.cm
2747    
2748    and its sources live under
2749    
2750        src/compiler/DebugProf
2751    
2752    ----------------------------------------------------------------------
2753    Name: Matthias Blume
2754    Date: 2001/08/15 17:15:00 EDT
2755    Tag: blume-20010815-compreorg
2756    Description:
2757    
2758    This is a first cut at reorganizing the CM libraries that make up the
2759    core of the compiler.  The idea is to separate out pieces that could
2760    be used independently by tools, e.g., the parser, the typechecker, etc.
2761    
2762    The current status is a step in this direction, but it is not quite
2763    satisfactory yet.  Expect more changes in the future.
2764    
2765    Here is the current (new) organization...
2766    
2767        What used to be $smlnj/viscomp/core.cm is now divided into
2768        six CM libraries:
2769    
2770             $smlnj/viscomp/basics.cm
2771                           /parser.cm
2772                           /elabdata.cm
2773                           /elaborate.cm
2774                           /execute.cm
2775                           /core.cm
2776    
2777        The CM files for these libraries live under src/system/smlnj/viscomp.
2778        All these libraries are proxy libraries that contain precisely
2779        one CM library component.  Here are the locations of the components
2780        (all within the src/compiler tree):
2781    
2782             Basics/basics.cm
2783             Parse/parser.cm
2784             ElabData/elabdata.cm
2785             Elaborator/elaborate.cm
2786             Execution/execute.cm
2787             core.cm
2788    
2789         [This organization is the same that has been used already
2790         for a while for the architecture-specific parts of the visible
2791         compiler and for the old version of core.cm.]
2792    
2793         As you will notice, many source files have been moved from their
2794         respective original locations to a new home in one of the above
2795         subtrees.
2796    
2797         The division of labor between the new libraries is the following:
2798    
2799             basics.cm:
2800                - Simple, basic definitions that pertain to many (or all) of
2801                  the other libraries.
2802             parser.cm:
2803                - The SML parser, producing output of type Ast.dec.
2804                - The type family for Ast is also defined and exported here.
2805             elabdata.cm:
2806                - The datatypes that describe input and output of the elaborator.
2807                  This includes types, absyn, and static environments.
2808             elaborator.cm:
2809                - The SML/NJ type checker and elaborator.
2810                  This maps an Ast.dec (with a given static environment) to
2811                  an Absyn.dec (with a new static environment).
2812                - This libraries implements certain modules that used to be
2813                  structures as functors (to remove dependencies on FLINT).
2814             execute.cm:
2815                - Everything having to do with executing binary code objects.
2816                - Dynamic environments.
2817             core.cm:
2818                - SML/NJ-specific instantiations of the elaborator and MLRISC.
2819                - Top-level modules.
2820                - FLINT (this should eventually become its own library)
2821    
2822    Notes:
2823    
2824    I am not 100% happy with the way I separated the elaborator (and its
2825    data structures) from FLINT.  Two instances of the same problem:
2826    
2827        1. Data structures contain certain fields that carry FLINT-specific
2828           information.  I hacked around this using exn and the property list
2829           module from smlnj-lib.  But the fact that there are middle-end
2830           specific fields around at all is a bit annoying.
2831    
2832        2. The elaborator calculates certain FLINT-related information.  I tried
2833           to make this as abstract as I could using functorization, but, again,
2834           the fact that the elaborator has to perform calculations on behalf
2835           of the middle-end at all is not nice.
2836    
2837        3. Having to used exn and property lists is unfortunate because it
2838           weakens type checking.  The other alternative (parameterizing
2839           nearly *everything*) is not appealing, though.
2840    
2841    I removed the "rebinding =" warning hack because due to the new organization
2842    it was awkward to maintain it.  As a result, the compiler now issues some of
2843    these warnings when compiling init.cmi during bootstrap compilation. On
2844    the plus side, you also get a warning when you do, for example:
2845       val op = = Int32.+
2846    which was not the case up to now.
2847    
2848    I placed "assign" and "deref" into the _Core structure so that the
2849    code that deals with the "lazy" keyword can find them there.  This
2850    removes the need for having access to the primitive environment
2851    during elaboration.
2852    
2853    ----------------------------------------------------------------------
2854    Name: Matthias Blume
2855    Date: 2001/08/13
2856    Tag: blume-20010813-closures
2857    Description:
2858    
2859    This fix was sent to us by Zhong Shao.  It is supposed to improve the
2860    performance of certain loops by avoiding needless closure allocation.
2861    
2862    ----------------------------------------------------------------------
2863    Name: Lal George
2864    Date: 2001/07/31 10:03:23 EDT 2001
2865    Tag: george-20010731-x86-fmalloc
2866    Description: Fixed bug in x86 calls
2867    
2868        There was a bug where call instructions would mysteriously
2869        vanish. The call instruction had to be one that returned
2870        a floating point value.
2871    
2872    ----------------------------------------------------------------------
2873    Name: Lal George
2874    Date: 2001/07/19 16:36:29 EDT 2001
2875    Tag: george-20010719-simple-cells
2876    Description:
2877    
2878    I have dramatically simplified the interface for CELLS in MLRISC.
2879    
2880    In summary, the cells interface is broken up into three parts:
2881    
2882      1. CellsBasis : CELLS_BASIS
2883    
2884            CellsBasis is a top level structure and common for all
2885            architectures.  it contains the definitions of basic datatypes
2886            and utility  functions over these types.
2887    
2888      2. functor Cells() : CELLS
2889    
2890            Cells generates an interface for CELLS that incorporates the
2891            specific resources on the target architecture, such as the
2892            presence of special register classes, their number and size,
2893            and various useful substructures.
2894    
2895      3. <ARCH>CELLS
2896    
2897            e.g. SparcCells: SPARCCELLS
2898    
2899            <ARCH>CELLS usually contains additional bindings for special
2900            registers  on the architecture, such as:
2901    
2902                    val r0 : cell           (* register zero *)
2903                    val y : cell            (* Y register *)
2904                    val psr : cell          (* processor status register *)
2905                    ...
2906    
2907            The structure returned by applying the Cells functor is opened
2908            in this interface.
2909    
2910    The main implication of all this is that the datatypes for cells is
2911    split between CellsBasis and CELLS -- a fairly simple change for user
2912    code.
2913    
2914    In the old scheme the CELLS interface had a definitional binding of
2915    the form:
2916    
2917            signature CELLS = sig
2918    
2919               structure CellsBasis = CellsBasis
2920    
2921               ...
2922    
2923            end
2924    
2925    With all the sharing constraints that goes on in MLRISC, this old
2926    design  quickly leads to errors such as:
2927    
2928            "structure definition spec inside of sharing ... "
2929    
2930    
2931    and appears to require an unacceptable amount of sharing and where
2932    constraint hackery.
2933    
2934    I think this error message (the interaction of definitional specs and
2935    sharing) requires more explanation on our web page.
2936    
2937    ----------------------------------------------------------------------
2938    Name: Matthias Blume
2939    Date: 2001/07/19 15:00:00 EDT
2940    Tag: blume-20010719-libreorg
2941    Description:
2942    
2943    This update puts together a fairly extensive but straightforward change
2944    to the way the libraries that implement the interactive system are
2945    organized:
2946    
2947       The biggest change is the elimination of structure Compiler.  As a
2948       replacement for this structure, there is now a CM library
2949       (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
2950       that exports all the substructures of the original structure Compiler
2951       directly.  So instead of saying Compiler.Foo.bar one now simply
2952       says Foo.bar.  (The CM libraries actually export a collection of
2953       structures that is richer than the collection of substructures of
2954       structure Compiler.)
2955    
2956       To make the transition smooth, there is a separate library called
2957       $smlnj/compiler/compiler.cm which puts together and exports the
2958       original structure Compiler (or at least something very close to it).
2959    
2960       There are five members of the original structure Compiler
2961       that are not exported directly but which instead became members
2962       of a new structure Backend (described by signature BACKEND).  These are:
2963       structure Profile (: PROFILE), structure Compile (: COMPILE), structure
2964       Interact (: INTERACT), structure Machine (: MACHINE), and val
2965       architecture (: string).
2966    
2967       Structure Compiler.Version has become structure CompilerVersion.
2968    
2969       Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
2970       by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
2971       or x86, respectively.
2972       Each of these exports the same frontend structures that
2973       $smlnj/compiler.cm exports.  But they do not have a structure Backend
2974       and instead export some structure <Arch>Backend where <Arch> is Alpha32,
2975       Hppa, PPC, Sparc, or X86, respectively.
2976    
2977       Library $smlnj/compiler/all.cm exports the union of the exports of
2978       $smlnj/compiler/<arch>.cm
2979    
2980       There are no structures <Arch>Compiler anymore, use
2981       $smlnj/compiler/<arch>.cm instead.
2982    
2983       Library host-compiler-0.cm is gone.  Instead, the internal library
2984       that instantiates CM is now called cm0.cm.  Selection of the host
2985       compiler (backend) is no longer done here but. (Responsibility for it
2986       now lies with $smlnj/compiler/current.cm.  This seems to be more
2987       logical.)
2988    
2989       Many individual files have been moved or renamed.  Some files have
2990       been split into multiple files, and some "dead" files have been deleted.
2991    
2992    Aside from these changes to library organization, there are also changes
2993    to the way the code itself is organized:
2994    
2995       Structure Binfile has been re-implemented in such a way that it no
2996       longer needs any knowledge of the compiler.  It exclusively deals
2997       with the details of binfile layout.  It no longer invokes the
2998       compiler (for the purpose of creating new prospective binfile
2999       content), and it no longer has any knowledge of how to interpret
3000       pickles.
3001    
3002       Structure Compile (: COMPILE) has been stripped down to the bare
3003       essentials of compilation.  It no longer deals with linking/execution.
3004       The interface has been cleaned up considerably.
3005    
3006       Utility routines for dealing with linking and execution have been
3007       moved into their own substructures.
3008    
3009       (The ultimate goal of these changes is to provide a light-weight
3010       binfile loader/linker (at least for, e.g., stable libraries) that
3011       does not require CM or the compiler to be present.)
3012    
3013    CM documentation has been updated to reflect the changes to library
3014    organization.
3015    
3016    ----------------------------------------------------------------------
3017    Name: Matthias Blume
3018    Date: 2001/07/10 17:30:00 EDT
3019    Tag: Release_110_34
3020    Description:
3021    
3022    Minor tweak to 110.34 (re-tagged):
3023    
3024      - README.html file added to CVS repository
3025      - runtime compiles properly under FreeBSD 3.X and 4.X
3026    
3027    ----------------------------------------------------------------------
3028    Name: Matthias Blume
3029    Date: 2001/07/10 17:30:00 EDT
3030    Tag: Release_110_34
3031    Description:
3032    
3033    New version number (110.34). New bootfiles.
3034    
3035    ----------------------------------------------------------------------
3036    Name: Matthias Blume
3037    Date: 2001/07/09 16:00:00 EDT
3038    Tag: blume-20010709-more-varargs
3039    Description:
3040    
3041    I changed the handling of varargs in ml-nlffigen again:
3042    The ellipsis ... will now simply be ignored (with an accompanying warning).
3043    
3044    The immediate effect is that you can actually call a varargs function
3045    from ML -- but you can't actually supply any arguments beyond the ones
3046    specified explicitly.  (For example, you can call printf with its format
3047    string, but you cannot pass additional arguments.)
3048    
3049    This behavior is only marginally more useful than the one before, but
3050    it has the advantage that a function or, more importantly, a function
3051    type never gets dropped on the floor, thus avoiding follow-up problems with
3052    other types that refer to the offending one.
3053    
3054    ----------------------------------------------------------------------
3055    Name: Matthias Blume
3056    Date: 2001/07/09 11:25:00 EDT
3057    Tag: blume-20010709-varargs
3058    Description:
3059    
3060    1. ckit-lib.cm now exports structure Error
3061    2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
3062       with a warning accompanied by a source location.  Moreover, it
3063       merely skips the offending function or type and proceeds with the
3064       rest of its work.u  As a result, one can safely feed C code containing
3065       "..." to ml-nlffigen.
3066    3. There are some internal improvements to CM, providing slightly
3067       more general string substitutions in the tools subsystem.
3068    
3069    ----------------------------------------------------------------------
3070    Name: Matthias Blume
3071    Date: 2001/06/27 15:10:00 EDT
3072    Tag: blume-20010627-concur
3073    Description:
3074    
3075    Fixed a small bug in CM's handling of parallel compilation.
3076    (You could observe the bug by Control-C-interrupting an ordinary
3077    CMB.make or CM.stabilize and then attaching some compile servers.
3078    The result was that all of a sudden the previously interrupted
3079    compilation would continue on its own.  This was because of
3080    an over-optimization: CM did not bother to clean out certain queues
3081    when no servers were attached "anyway", resulting in the contents
3082    of these queues to grab control when new servers did get attached.)
3083    
3084    There is also another minor update to the CM manual.
3085    
3086    ----------------------------------------------------------------------
3087    Name: Matthias Blume
3088    Date: 2001/06/26 16:15:00 EDT
3089    Tag: blume-20010626-cmdoc
3090    Description:
3091    
3092    Minor typo fixed in CM manual (syntax diagram for libraries).
3093    
3094    ----------------------------------------------------------------------
3095    Name: Matthias Blume
3096    Date: 2001/06/25 22:55:00 EDT
3097    Tag: blume-20010625-x86pc
3098    Description:
3099    
3100    Fixed a nasty bug in the X86 assembly code that caused signal
3101    handlers to fail (crash) randomly.
3102    
3103    ----------------------------------------------------------------------
3104    Name: Matthias Blume
3105    Date: 2001/06/25 12:05:00 EDT
3106    Tag: blume-20010625-nlffigen
3107    Description:
3108    
3109    This update fixes a number of minor bugs in ml-nlffigen as reported by
3110    Nick Carter <nbc@andrew.cmu.edu>.
3111    
3112      1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
3113      2. Default names for generated files are now derived from the name of
3114         the C file *without its directory*.  In particular, this causes generated
3115         files to be placed locally even if the C file is in some system directory.
3116      3. Default names for generated signatures and structures are also derived
3117         from the C file name without its directory.  This avoids silly things
3118         like "structure GL/GL".
3119         (Other silly names are still possible because ml-nlffigen does not do
3120          a thorough check of whether generated names are legal ML identifiers.
3121          When in doubt, use command line arguments to force particular names.)
3122    
3123    ----------------------------------------------------------------------
3124    Name: Matthias Blume
3125    Date: 2001/06/21 12:25:00 EDT
3126    Tag: blume-20010621-eXene
3127    Description:
3128    
3129    eXene now compiles and (sort of) works again.
3130    
3131    The library name (for version > 110.33) is $/eXene.cm.
3132    
3133    I also added an new example in src/eXene/examples/nbody.  See the
3134    README file there for details.
3135    
3136    ----------------------------------------------------------------------
3137    Name: Matthias Blume
3138    Date: 2001/06/20 16:40:00 EDT
3139    Tag: blume-20010620-cml
3140    Description:
3141    
3142    CML now compiles and works again.
3143    
3144    Libraries (for version > 110.33):
3145    
3146      $cml/cml.cm            Main CML library.
3147      $cml/basis.cm          CML's version of $/basis.cm.
3148      $cml/cml-internal.cm   Internal helper library.
3149      $cml/core-cml.cm       Internal helper library.
3150      $cml-lib/trace-cml.cm  Tracing facility.
3151      $cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
3152    
3153    The installer (config/install.sh) has been taught how to properly
3154    install this stuff.
3155    
3156    ----------------------------------------------------------------------
3157    Name: Matthias Blume
3158    Date: 2001/06/19 17:55:00 EDT
3159    Tag: blume-20010619-instantiate
3160    Description:
3161    
3162    This un-breaks the fix for bug 1432.
3163    (The bug was originally fixed in 110.9 but I broke it again some
3164    time after that.)
3165    
3166    ----------------------------------------------------------------------
3167    Name: Matthias Blume
3168    Date: 2001/06/19 17:25:00 EDT
3169    Tag: blume-20010619-signals
3170    Description:
3171    
3172    This should (hopefully) fix the long-standing signal handling bug.
3173    (The runtime system was constructing a continuation record with an
3174    incorrect descriptor which would cause the GC to drop data on the floor...)
3175    
3176    ----------------------------------------------------------------------
3177    Name: Matthias Blume
3178    Date: 2001/06/15 15:05:00 EDT
3179    Tag: blume-20010615-moresparc
3180    Description:
3181    
3182    Here is a short late-hour update related to Sparc c-calls:
3183    
3184     -- made handling of double-word arguments a bit smarter
3185    
3186     -- instruction selection phase tries to collapse certain clumsily
3187        constructed ML-Trees; typical example:
3188    
3189            ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
3190    
3191        This currently has no further impact on SML/NJ since mlriscGen does
3192        not seem to generate such patterns in the first place, and c-calls
3193        (which did generate them in the beginning) has meanwhile been fixed
3194        so as to avoid them as well.
3195    
3196    ----------------------------------------------------------------------
3197    Name: Matthias Blume
3198    Date: 2001/06/15 15:05:00 EDT
3199    Tag: blume-20010615-sparc
3200    Description:
3201    
3202    The purpose of this update is to provide an implementation of NLFFI
3203    on Sparc machines.
3204    
3205    Here are the changes in detail:
3206    
3207       * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
3208       the Sparc implementation of the c-calls API.
3209       * The Sparc backend of SML/NJ has been modified to uniformely use %fp
3210       for accessing the ML frame.  Thus, we have a real frame pointer and
3211       can freely modify %sp without need for an omit-frame-ptr phase.
3212       The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
3213       this case.
3214       * ml-nlffigen has been taught to produce code for different architectures
3215       and calling conventions.
3216       * In a way similar to what was done in the x86 case, the Sparc
3217       backend uses its own specific extension to mltree.  (For example,
3218       it needs to be able to generate UNIMP instructions which are part
3219       of the calling convention.)
3220       * ml-nlffi-lib was reorganized to make it more modular (in particular,
3221       to make it easier to plug in new machine- and os-dependent parts).
3222    
3223    There are some other fairly unrelated bug fixes and cleanups as well:
3224    
3225       * I further hacked the .cm files for MLRISC tools (like MDLGen) so
3226       that they properly share their libraries with existing SML/NJ libraries.
3227       * I fixed a minor cosmetic bug in CM, supressing certain spurious
3228       follow-up error messages.
3229       * Updates to CM/CMB documentation.
3230    
3231    TODO items:
3232    
3233       * MLRISC should use a different register as its asmTemp on the Sparc.
3234         (The current %o2 is a really bad choice because it is part of the
3235         calling conventions, so things might interfere in unexpected ways.)
3236    
3237    ----------------------------------------------------------------------
3238    Name: Matthias Blume
3239    Date: 2001/06/07
3240    Tag: blume-20010607-calls
3241    Description:
3242    
3243    A number of internal changes related to C calls and calling conventions:
3244    
3245    1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
3246       number of bytes popped implicitly (by the callee).  In most cases
3247       this field is 0 but on x86/win32 it is some non-zero value.  This
3248       is information provided for the benefit of the "omit-frameptr" pass.
3249    2. The CALL instruction on the x86 carries a similar "pops" field.
3250       The instruction selection phase copies its value from the ML-Tree
3251       CALL statement.
3252    3. On all other architectures, the instruction selection phase checks
3253       whether "pops=0" and complains if not.
3254    4. The c-calls implementation for x86 now accepts two calling conventions:
3255       "ccall" and "stdcall".  When "ccall" is selected, the caller cleans
3256       up after the call and pops is set to 0.  For "stdcall", the caller
3257       does nothing, leaving the cleanup to the callee; pops is set to
3258       the number of bytes that were pushed onto the stack.
3259    5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
3260       distinguish between "ccall" and "stdcall".
3261    6. The UNIMP instruction has been added to the supported Sparc instruction
3262       set. (This is needed for implementing the official C calling convention
3263       on this architecture.)
3264    7. I fixed some of the .cm files under src/MLRISC/Tools to make them
3265       work with the latest CM.
3266    
3267    ----------------------------------------------------------------------
3268    Name: Matthias Blume
3269    Date: 2001/06/05 15:10:00 EDT
3270    Tag: blume-20010605-cm-index
3271    Description:
3272    
3273    0. The "lambdasplit" parameter for class "sml" in CM has been documented.
3274    
3275    1. CM can now generate "index files".  These are human-readable files
3276       that list on a per-.cm-file basis each toplevel symbol defined or
3277       imported.  The location of the index file for
3278       <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
3279       To enable index-file generation, set CM.Control.generate_index to true
3280       or export an environment-symbol: export CM_GENERATE_INDEX=true.
3281    
3282       The CM manual has been updated accordingly.
3283    
3284    2. I made some slight modifications to the c-calls API in MLRISC.
3285    
3286         a) There is now a callback to support saving/restoring of
3287            dedicated but caller-save registers around the actual call
3288            instruction.
3289         b) One can optionally specify a comment-annotation for the
3290            call instruction.
3291    
3292    3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
3293       (For example, the comment annotation shows the C prototype of
3294        the function being called.)
3295    
3296    ----------------------------------------------------------------------
3297    Name: Matthias Blume
3298    Date: 2001/06/01 13:30:00 EDT
3299    Tag: blume-20010601-nlffi-cleanup
3300    Description:
3301    
3302    This is mostly a cleanup of MLFFI stuff:
3303    
3304       - some signature files have been put into a more exposed place
3305       - the ugly 'f type parameter is gone (simplifies types tremendously!)
3306       - ml-nlffigen changed accordingly
3307       - tutorial updated
3308    
3309    Other changes:
3310    
3311       - author's affiliation in CM manual(s) updated
3312       - some more recognized keywords added to Allen's sml.sty
3313    
3314    ----------------------------------------------------------------------
3315    Name: Matthias Blume
3316    Date: 2001/05/25 15:30:00 EDT
3317    Tag: blume-20010525-iptr
3318    Description:
3319    
3320      - put the official 110.33-README (as it appears on the ftp server) under
3321        CVS
3322      - fixed a small bug related to incomplete pointer types in
3323        ml-nlffigen
3324      - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
3325        (it does not need the 'f type parameter)
3326    
3327    ----------------------------------------------------------------------
3328    Name: Matthias Blume
3329    Date: 2001/05/23 14:30:00 EDT
3330    Tag: Release_110_33
3331    Description:
3332    
3333    New version number (110.33).  New bootfiles.
3334    
3335    ----------------------------------------------------------------------
3336    Name: Matthias Blume
3337    Date: 2001/05/22 18:06:00 EDT
3338    Tag: blume-20010522-targets
3339    Description:
3340    
3341    Made install.sh use file config/targets.customized if it exists, falling
3342    back to config/targets if it doesn't.  This way one can have a customized
3343    version of the targets file without touching the "real thing", thus
3344    eliminating the constant fear of accidentally checking something bogus
3345    back into the CVS repository...  (File config/targets.customized must
3346    not be added to the repository!)
3347    
3348    ----------------------------------------------------------------------
3349    Name: Matthias Blume
3350    Date: 2001/05/22 16:30:00 EDT
3351    Tag: blume-20010522-minitut
3352    Description:
3353    
3354    1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
3355       struct returns.
3356    2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
3357       incomplete, preliminary documentation for NLFFI.
3358    
3359    ----------------------------------------------------------------------
3360    Name: Matthias Blume
3361    Date: 2001/05/14 11:30:00 EDT
3362    Tag: blume-20010514-script
3363    Description:
3364    
3365    Some bugs in install script fixed.
3366    
3367    In addition to that I also made a slight change to the NLFFI API:
3368    Functors generated by ml-nlffigen now take the dynamic library as a
3369    straight functor argument, not as a suspended one.  (The original
3370    functor code used to force the suspension right away anyway, so there
3371    was nothing gained by this complication of the interface.)
3372    
3373    ----------------------------------------------------------------------
3374    Name: Matthias Blume
3375    Date: 2001/05/11 14:35:00 EDT
3376    Tag: blume-20010511-ml-nlffi
3377    Description:
3378    
3379    I finally took the plunge and added my new FFI code to the main
3380    repository.  For x86-linux it is now ready for prime-time.
3381    
3382    There are two new subdirectories of "src":
3383    
3384      - ml-nlffi-lib:
3385           The utility library for programs using the FFI interface.
3386           Here is the implementation of $/c.cm and its associated low-level
3387           partners $/c-int.cm and $/memory.cm.
3388      - ml-nlffigen:
3389           A stand-alone program for generating ML glue code from C source
3390           code.
3391    
3392    Building ml-nlffigen requires $/ckit-lib.cm.
3393    
3394    The config/install.sh script has been updates to do the Right Thing
3395    (hopefully).
3396    
3397    Notice that the source tree for the C-Kit will not be put under "src"
3398    but directly under the installation root directory.  (This is the
3399    structure that currently exists on the CVS server when you check out
3400    module "sml".)  Fortunately, config/install.sh knows about this oddity.
3401    
3402    Bugs: No documentation yet.
3403    
3404    ----------------------------------------------------------------------
3405  Name: Matthias Blume  Name: Matthias Blume
3406  Date: 2001/05/09 16:35:00 EDT  Date: 2001/05/09 16:35:00 EDT
3407  Tag: blume-20010509-cpscontract  Tag: blume-20010509-cpscontract
# Line 2201  Line 5590 
5590       elaborator). There were a lot of changes during my "linkpath" trials       elaborator). There were a lot of changes during my "linkpath" trials
5591       that could have been reverted to their original state but weren't.       that could have been reverted to their original state but weren't.
5592       Please, don't be too harsh on me for messing with this code a bit more       Please, don't be too harsh on me for messing with this code a bit more
5593       than what was strictly necessary...  (I _did_ resist the tempation       than what was strictly necessary...  (I _did_ resist the temptation
5594       of doing any "global reformatting" to avoid an untimely death at       of doing any "global reformatting" to avoid an untimely death at
5595       Dave's hands. :)       Dave's hands. :)
5596    

Legend:
Removed from v.826  
changed lines
  Added in v.1368

root@smlnj-gforge.cs.uchicago.edu
ViewVC Help
Powered by ViewVC 1.0.0