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 842, Tue Jun 19 21:24:35 2001 UTC revision 1356, Wed Sep 10 02:54:11 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@tti-c.org)
17    Date: 2003/09/09 22:00:00 CDT
18    Tag: Release_110_43
19    Description:
20    
21        New working release 110.43.  New bootfiles.
22    
23    ----------------------------------------------------------------------
24    Name: Matthias Blume (blume@tti-c.org)
25    Date: 2003/09/09 19:20:00 CDT
26    Tag: blume-20030909-installer
27    Description:
28    
29    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
30    Modified config/install.bat to take advantage of it.  Also modified
31    config/install.sh (and called it config/new-install.sh) to take advantage
32    of it on Unix systems.  (The SML code is (supposed to be) platform-
33    independent.)
34    
35    The installer can now install everything under Win32
36    as well as under *nix as long as it compiles.
37    
38    Other changes:
39    
40      - made CML compile again under Win32
41      - made eXene compile under Win32 (by providing a fake structure UnixSock
42        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
43      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
44        same as type int;  under Win32 it isn't
45      - fixed some slice-related problems in the win32-specific parts of CML
46      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
47        (under Win32, the volume name is case-insensitive, and the
48        OS.Path code compares volume names for equality)
49    
50    ----------------------------------------------------------------------
51    Name: Matthias Blume (blume@tti-c.org)
52    Date: 2003/09/08 11:55:00 CDT
53    Tag: blume-20030908-fullpath
54    Description:
55    
56    Made Win32 version of OS.FileSys.fullPath return current directory
57    when given an empty string.  This is what the spec says, and incidentally,
58    CM depends on it.  (CM otherwise goes into an infinite loop in certain
59    cases when presented with the name of a non-existing .cm file.)
60    
61    ----------------------------------------------------------------------
62    Name: Matthias Blume (blume@tti-c.org)
63    Date: 2003/09/04 16:30:00 CDT
64    Tag: blume-20030905-slices-etc
65    Description:
66    
67      1. Changed interface to vectors and arrays in Basis to match
68         (draft) Basis spec.
69      2. Added signatures and implementations of slices according to
70         Basis spec.
71      3. Edited source code throughout the system to make it compile again
72         under 1. and 2.  (In some cases code had to be added to have it
73         match the new signatures.)
74      4. MLRISC should be backward-compatible: the copies of the originals
75         of files that needed to change under 3. were retained, the .cm files
76         check the compiler version number and use old versions when
77         appropriate.
78      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
79         dirstream -> string option (in accordance with Basis spec).
80      6. When generating code that counts lines, ml-lex used function
81         CharVector.foldli, taking advantage of its old interface.
82         This has been replaced with the corresponding code from
83         CharVectorSlice.  (html-lex must be re-lexed!)
84      7. BitArray in smlnj-lib/Util has been extended/modified to match the
85         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
86      8. Removed temporary additions (fromInternal, toInternal) from the
87         (now obsolete) IntInf in smlnj-lib/Util.
88      9. Cleaned up structure Byte.
89     10. Added localOffset, scan, and fromString to Date (according to spec).
90         Cleaned/corrected implementation of Date.
91         (Still need to check for correctness; implement better canonicalizeDate.)
92     11. Added "scan" to signature IEEE_REAL.
93     12. Some improvements to IntInf [in particular: efficiency-hack for
94         mod and rem when second operand is 2 (for parity checks).]
95     13. Changed representation of type Time.time, using a single IntInf.int
96         value counting microseconds.  This considerably simplified the
97         implementation of structure Time.  We now support negative time
98         values; scan and fromString handle signs.
99     14. Functor PrimIO now takes two additional arguments (VectorSlice and
100         ArraySlice).
101    
102    ----------------------------------------------------------------------
103    Name: Matthias Blume (blume@tti-c.org)
104    Date: 2003/08/28 17:00:00 CDT
105    Tag: blume-20030828-intinf
106    Description:
107    
108    This is a major update which comes with a version number bump
109    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
110    and an implementation of IntInf in the Basis.
111    
112    There are a fairly large number of related changes and updates throughout
113    the system:
114    
115    Basis:
116            - Implemented IntInf.
117            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
118            - Added some missing Real64 operations, most notably Real.toLargeInt.
119            - Added FixedInt as a synonym for Int32.
120    
121    compiler:
122            * Added support for a built-in intinf type.
123              - literals
124              - pattern matching
125              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
126              - overloading on literals and operations
127    
128            This required adding a primitive type intinf, some additional
129            primops, and implementations for several non-trivial intinf
130            operations in Core.  (The intinf type is completely abstract
131            to the compiler; all operations get delegated back to the Core.)
132    
133            * Intinf equality is handled by polyequal.  However, the compiler
134            does not print its usual warning in this case (since polyequal
135            is the right thing to do there).
136    
137            * Improved the organization of structure InlineT.
138    
139            * A word about conversion primops:
140            If conversions involving intinf do not cancel out during
141            CPS contract, then the compiler must insert calls to Core functions.
142            Since all core access must be resolved already during the FLINT
143            translate phase, it would be too late a the time of CPS contract
144            to add new Core calls.  For this reason, conversion primops
145            for intinf carry two arguments:  1. the numeric argument that
146            they are supposed to convert, and 2. the Core function that
147            can help with this conversion if necessary.  If CPS contract
148            eliminates a primop, then the associated Core function becomes
149            dead and goes away.  Intinf conversion primops that do not get
150            eliminated by CPS contract get rewritten into calls of their
151            core functions by a separate, new phase.
152    
153    interactive system:
154            - Control.Print.intinfDepth controls max length of intinf constants
155            being printed. (Analogous to Control.Print.stringDepth.)
156            - Cleanup in printutil and pputil: got rid of unused stuff and
157            duplicates; replaced some of the code with code that makes better
158            use of library functionality.
159    
160    CM:
161            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
162            are no longer silent.
163    
164    CKIT:
165            Fixed mismatched uses of Int32 and LargeInt.  I always decided
166            in favor of LargeInt -- which is now the same as IntInf.
167            CKIT-knowledgable people should check whether this is what's
168            intended and otherwise change things back to using Int32 or
169            FixedInt.
170    
171    Throughout the code:
172            Started using IntInf.int literals and built-in operations
173            (e.g., comparison with 0) where this seems appropriate.
174    
175    
176    ----------------------------------------------------------------------
177    Name: Dave MacQueen (dbm@cs.uchicago.edu)
178    Date: 2003/08/13 11:36:00 CDT
179    Tag: dbm-20030813-mcz-merge1
180    Description:
181    
182    Merging changes from the mcz-branch development branch into trunk.
183    These changes involve replacement of the emulated old prettyprinter
184    interface with direct use of the SML/NJ Lib PP library, and fixing
185    of a couple of bugs (895, 1186) relating to error messages.  A new
186    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
187    has been added.
188    
189    ----------------------------------------------------------------------
190    Name: Matthias Blume (blume@tti-c.org)
191    Date: 2003/08/11 15:45:00 CDT
192    Tag: blume-20030811-windows
193    Description:
194    
195    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
196    
197        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
198    
199    This patch restores SML/NJ's ability to run under win32.  There are a
200    number of changes, including fixes for several bugs that had gone
201    unnoticed until now:
202    
203      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
204        trying to run the win32 version from within cygwin.)
205    
206      - There are a number of simple .bat scripts that substitute for their
207        corresponding Unix shell-scripts.  (See below.)
208    
209      - The internals of ml-build have been modified slightly.  The main
210        difference is that instead of calling ".link-sml" (or link-sml.bat)
211        using OS.Process.system, the ML process delegates this task back
212        to the script.  Otherwise problems arise in mixed environments such
213        as Cygwin where scripts look and work like Unix scripts, but
214        where OS.Process.system cannot run them.
215    
216      - In CM, the srcpath pickler used native pathname syntax -- which
217        is incorrect in the case of cross-compilation.  The new pickle format
218        is independent of platform-specific naming conventions.
219    
220      - Path configuration files (such as lib/pathconfig) can now choose
221        between native and standard syntax.  Placing a line of the form
222    
223            standard!
224    
225        into the file causes all subsequent paths to be interpreted using
226        CM standard pathname syntax (= Unix conventions); a line
227    
228            native!
229    
230        switches back to native style.  This was needed so that
231        path config files can be written portably, see src/system/pathconfig.
232    
233      - Runtime system:
234    
235          - win32-filesys.c:  get_file_time and set_file_time now
236            access modification time, not creation time.
237    
238          - I/O code made aware of new array representation.
239    
240          - Bug fixes in X86.prim.masm.
241    
242       - src/system/makeml made aware of win32. (For use under cygwin
243            and other Unix-environments for windows.)
244    
245       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
246         which caused BinIO.inputAll to fail consistently.
247    
248    .bat scripts:
249    
250       Windows .bat scripts assume that SMLNJ_HOME is defined.
251    
252       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
253         applications (sml, ml-yacc, ml-lex).
254       - ml-build.bat: analogous to ml-build.
255       - config\install.bat:  Analogous to config/install.sh.  This requires
256         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
257         (nmake etc. must be on the path, and vcvars32 must have been run.)
258         Moreover, sources for ml-lex and ml-yacc need to exist under src,
259         and the bootfile hierarchy must have been unpacked under
260         sml.boot.x86-win32.
261         The script is very primitive and does a poor job at error checking.
262         It only installs the base system, ml-lex, and ml-yacc.  No other
263         libraries are being installed (i.e., you get only those that
264         are part of the compiler.)
265       - link-sml.bat:  analogous to .link-sml, but not currently used
266    
267    Unrelated bug fixes:
268    
269      - ml-nlffigen now exports structures ST_* corresponding to incomplete
270        types.
271      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
272        otherwise.)
273    
274    ----------------------------------------------------------------------
275    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
276    Date: 2003/06/17
277    Tag: macqueen-20030617-bug895
278    Description:
279    
280    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
281    Tag will be used for new development branch (mcz-branch) for
282    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
283    
284    ----------------------------------------------------------------------
285    Name: Matthias Blume (blume@tti-c.org)
286    Date: 2003/05/27 16:55:00 CDT
287    Tag: blume-20030527-polyeq
288    Description:
289    
290    Tried to eliminated most cases of polymorphic equality.
291    
292    ----------------------------------------------------------------------
293    Name: Matthias Blume (blume@tti-c.org)
294    Date: 2003/05/21 17:45:00 CDT
295    Tag: blume-20030517-complete
296    Description:
297    
298    Two changes:
299    
300      1. Added a flag for controlling whether non-exhaustive bindings will
301         be treated as errors (default is false).
302      2. Cleaned up the *entire* source tree so that CMB.make goes through
303         without a single non-exhaustive match- or bind warning.
304    
305    ----------------------------------------------------------------------
306    Name: Matthias Blume (blume@tti-c.org)
307    Date: 2003/05/17 10:20:00 CDT
308    Tag: blume-20030517-absyn
309    Description:
310    
311    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
312    
313       This mainly affects the quality of error messages.  However, some
314       of the code is now more straightforward than before. (Treatment of
315       the above four constructs in translate.sml is much simpler than
316       the "macro-expansion" that was going on before.  Plus, the mach-
317       compiler no longer gets invoked just to be able to compile an
318       if-expression.)
319    
320    2. The ErrorMsg.Error exception is now caught and absorbed by the
321       interactive loop.
322    
323    ----------------------------------------------------------------------
324    Name: Allen Leung
325    Date: 2003/05/16 13:05:00 CDT
326    Tag: leunga-20030516-cygwin-runtime
327    Description:
328    
329        Ported the runtime system to cygwin, which uses the unix
330    x86-unix bin files.  Missing/buggy features:
331    
332       o getnetbyname, getnetbyaddr: these functions seem to be missing in
333         the Cygwin library.
334       o Ctrl-C handling may be flaky.
335       o Windows system calls and Windows I/O are not supported.
336    
337       A new set of binfiles is located at:
338    
339         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
340    
341       This is only needed for bootstrapping the cygwin version of smlnj.
342       Other x86 versions can use the existing binfiles.
343    
344    ----------------------------------------------------------------------
345    Name: Matthias Blume
346    Date: 2003/04/08 15:42:00 CDT
347    Tag: blume-20030408-listpair
348    Description:
349    
350    1. Added a target 'mlrisc' to installer.
351    
352    2. Added missing elements to structure ListPair.
353    
354    ----------------------------------------------------------------------
355    Name: Allen Leung
356    Date: 2003/01/07 10:40:00 EST
357    Tag: leunga-20030107-int-rem
358    Description:
359    
360    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
361    The arguments to the SUBL instruction were swapped.
362    
363    ----------------------------------------------------------------------
364    Name: Matthias Blume
365    Date: 2002/12/12 16:25:00 EST
366    Tag: blume-20021212-risc-ra
367    Description:
368    
369    Fixed a serious bug in the rewrite code for FP spilling/reloading that
370    sent the RA into an infinite loop when floating point registers get
371    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
372    110.37 and 110.38.)
373    There was another set of potential problems related to the handling of
374    MLRISC annotations (but those did not yet cause real problems, apparently).
375    
376    ----------------------------------------------------------------------
377    Name: Matthias Blume
378    Date: 2002/12/06 22:40:00 EST
379    Tag: blume-20021206-cm-fileid
380    Description:
381    
382    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
383    This fixes the problem of CM getting confused by files that suddenly
384    change their identity (e.g., by getting unlinked and recreated by some
385    text editor such as vi).  There might be a better/cheaper/cleaner way
386    of doing this, but for now this will have to do.
387    
388    ----------------------------------------------------------------------
389    Name: Matthias Blume
390    Date: 2002/10/28 09:50:00 EST
391    Tag: blume-20021028-typecheck
392    Description:
393    
394    Exported structure Typecheck from $smlnj/viscomp/core.cm.
395    
396    ----------------------------------------------------------------------
397    Name: Matthias Blume
398    Date: 2002/10/17 09:10:00 EDT
399    Tag: Release_110_42
400    Description:
401    
402    In good old tradition, there has been a slight hiccup so that we have
403    to patch 110.42 after the fact.   The old release tag has been replaced
404    (see below).
405    
406    The change solves a problem with two competing approaches the
407    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
408    each other's way.
409    
410    This change only affects the runtime system code and the installer script.
411    (No new bootfiles.)
412    
413    ----------------------------------------------------------------------
414    Name: Matthias Blume
415    Date: 2002/10/16 12:00:00 EDT
416    Tag: Release_110_42_removed
417    Description:
418    
419    New working release.  New bootfiles.
420    
421    ----------------------------------------------------------------------
422    Name: Matthias Blume
423    Date: 2002/10/10 13:10:00 EDT
424    Tag: blume-20021010-ppc-divs
425    Description:
426    
427    The mltree operator DIVS must be implemented with an overflow check on
428    the PPC because the hardware indicates divide-by-zero using "overflow" as
429    well.
430    
431    ----------------------------------------------------------------------
432    Name: Matthias Blume
433    Date: 2002/07/23 11:45:00
434    Tag: blume-20020723-smlnj-home
435    Description:
436    
437    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
438    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
439    CM_PATHCONFIG is also set), the path configuration file is assumed
440    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
441    the entire tree to some other place and everything will "just work".
442    
443    (Companion commands such as ml-build and ml-makedepend also sense this
444    variable.)
445    
446    ----------------------------------------------------------------------
447    Name: Matthias Blume
448    Date: 2002/07/12 21:19:00 EDT
449    Tag: blume-20020712-liveness
450    Description:
451    
452    Exported two useful "step" functions from liveness module (MLRISC).
453    
454    ----------------------------------------------------------------------
455    Name: Matthias Blume
456    Date: 2002/07/05 16:00 EDT
457    Tag: Release_110_41
458    Description:
459    
460    New working release.  New bootfiles.
461    
462    ----------------------------------------------------------------------
463    Name: Matthias Blume
464    Date: 2002/07/05 10:25:00 EDT
465    Tag: blume-20020705-btimp
466    Description:
467    
468    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
469    other clients can set up backtracing support.
470    
471    ----------------------------------------------------------------------
472    Name: Matthias Blume
473    Date: 2002/06/25 17:23:00 EDT
474    Tag: blume-20020625-fpmax
475    Description:
476    
477    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
478    case.  (The sense of the isNaN test was reversed -- which made min and
479    max always return their first argument.)
480    
481    ----------------------------------------------------------------------
482    Name: Matthias Blume
483    Date: 2002/06/11
484    Tag: blume-20020611-unixpath
485    Description:
486    
487    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
488    
489    ----------------------------------------------------------------------
490    Name: Matthias Blume
491    Date: 2002/06/10 16:35:00 EDT
492    Tag: blume-20020610-ieeereal
493    Description:
494    
495    I back-ported my implementation of IEEEReal.fromString from the
496    idlbasis-devel branch so that we can test it.
497    
498    Another small change is that ppDec tries to give more information
499    than just "<sig>" in the case of functors.  However, this code is
500    broken in some mysterious way if the functor's body's signature
501    has not been declared by ascription but gets inferred from the
502    implementation.  This needs fixing...
503    
504    ----------------------------------------------------------------------
505    Name: Matthias Blume
506    Date: 2002/05/31
507    Tag: blume-20020531-btrace-mode
508    Description:
509    
510    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
511    the wayside when I switched over to using Controls everywhere.)
512    
513    ----------------------------------------------------------------------
514    Name: Lal George
515    Date: 2002/05/23 12:21:40 EDT
516    Tag: george-20020523-visual-labels
517    Description:
518    
519    Labels are now displayed in the graphical output to make
520    the fall-through and target blocks obvious.
521    
522    ----------------------------------------------------------------------
523    Name: Matthias Blume
524    Date: 2002/05/22 11:03:00 EDT
525    Tag: blume-20020522-shrink
526    Description:
527    
528    John tweaked yesterday's fix for 1131 to handle an out-of-memory
529    situation that comes up when allocating huge arrays.
530    
531    ----------------------------------------------------------------------
532    Name: Matthias Blume
533    Date: 2002/05/21 16:00:00 EDT
534    Tag: Release_110_40
535    Description:
536    
537    New working release (110.40).  New bootfiles.
538    
539    [Also: John Reppy fixed GC bug 1131.]
540    
541    ----------------------------------------------------------------------
542    Name: Matthias Blume
543    Date: 2002/05/21 12:35:00 EDT
544    Tag: blume-20020521-cmdoc
545    Description:
546    
547    CM documentation update.
548    
549    ----------------------------------------------------------------------
550    Name: Matthias Blume
551    Date: 2002/05/21 10:55:00 EDT
552    Tag: blume-20020521-misc
553    Description:
554    
555    - John tweaked runtime to be silent on heap export (except when
556      GC messages are on).
557    - I added a few more things (cross-compiling versions of CMB) to
558      config/preloads (as suggestions).
559    
560    ----------------------------------------------------------------------
561    Name: Matthias Blume
562    Date: 2002/05/20 22:25:00 EDT
563    Tag: blume-20020520-controls
564    Description:
565    
566    - Added ControlUtil structure to control-lib.cm.
567    - Use it throughout.
568    - Used Controls facility to define MLRISC controls (as opposed to
569      registering MLRISC control ref cells with Controls after the fact)
570    - Fixed messed-up controls priorities.
571    
572    * Removed again all the stuff from config/preloads that one wouldn't
573      be able to preload at the time the initial heap image is built.
574      (Many libraries, e.g., CML, do not exist yet at this time.  The
575       only libraries that can be preloaded via config/preloads are those
576       that come bundled with the bootfiles.)
577    
578    ----------------------------------------------------------------------
579    Name: Matthias Blume
580    Date: 2002/05/20 10:59:00 EDT
581    Tag: blume-20020520-preloads
582    Description:
583    
584    Added a lot of commented-out suggestions for things to be included
585    in config/preloads.
586    
587    ----------------------------------------------------------------------
588    Name: Allen Leung
589    Date: 2002/05/18 14:20:00 EDT
590    Tag: leunga-20020518-mdl
591    Description:
592    
593    o Made the mdl tool stuff compile and run again.
594    o I've disabled all the stuff that depends on RTL specifications; they
595      are all badly broken anyway.
596    
597    ----------------------------------------------------------------------
598    Name: Matthias Blume
599    Date: 2002/05/17 16:49:00 EDT
600    Tag: blume-20020517-controls
601    Description:
602    
603    0. John Reppy made several modifications to the SML/NJ library.
604       In particular, there is a shiny new controls-lib.cm.
605    
606    1. Pushed new controls interface through compiler so that everything
607       compiles again.
608    
609    2. Added FormatComb and FORMAT_COMB to the CML version of the
610       SML/NJ library (so that CML compiles again).
611    
612    3. Modified init scripts because XXX_DEFAULT environment variables
613       are no longer with us.  (Boot-time initialization is now done
614       using the same environment variables that are also used for
615       startup-time initialization of controls.)
616    
617    ----------------------------------------------------------------------
618    Name: Lal George
619    Date: 2002/05/15 09:20:10 EDT
620    Tag: george-20020515-pseudo-op-decls
621    Description:
622    
623            All pseudo-ops emitted before the first segment declaration
624    such as TEXT, DATA, and BSS directives are assumed to be global
625    declarations and are emitted first in the assembly file. This is
626    useful in a number of situations where one has pseudo-ops that are not
627    specific to any segment, and also works around the constraint that one
628    cannot have client pseudo-ops in the TEXT segment.
629    
630            Because no segment is associated with these declarations it is
631    an error to allocate any space or objects before the first segment
632    directive and an exception will be raised. However, we cannot make
633    this check for client pseudo-ops.
634    
635            These top level declarations are a field in the CFG graph_info.
636    In theory you can continue to add to this field after the CFG has been
637    built -- provided you know what you are doing;-)
638    
639    ----------------------------------------------------------------------
640    Name: Matthias Blume
641    Date: 2002/05/13 16:40:00 EDT
642    Tag: blume-20020513-pp-etc
643    Description:
644    
645    A few minor bugfixes:
646    
647      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
648        (Bogus printouts for redefined bindings still occur.  Compiler
649        bug should no longer occur now.  We need to redo the prettyprinter
650        from scratch.)
651    
652      - CM pathname printer now also adds escape sequences for ( and )
653    
654      - commend and docu fixes for ml-nlffi
655    
656    ----------------------------------------------------------------------
657    Name: Matthias Blume
658    Date: 2002/05/10 16:40:00 EDT
659    Tag: blume-20020510-erg-textio
660    Description:
661    
662    Applied the following bugfix provided by Emden Gansner:
663    
664        Output is corrupted when outputSubstr is used rather than output.
665    
666        The problem occurs when a substring
667    
668            ss = (s, dataStart, dataLen)
669    
670        where dataStart > 0, fills a stream buffer with avail bytes left.
671        avail bytes of s, starting at index dataStart, are copied into the
672        buffer, the buffer is flushed, and then the remaining dataLen-avail
673        bytes of ss are copied into the beginning of the buffer. Instead of
674        starting this copy at index dataStart+avail in s, the current code
675        starts the copy at index avail.
676    
677        Fix:
678        In text-io-fn.sml, change line 695 from
679             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
680        to
681             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
682    
683    ----------------------------------------------------------------------
684    Name: Matthias Blume
685    Date: 2002/04/12 13:55:00 EDT
686    Tag: blume-20020412-assyntax
687    Description:
688    
689    1. Grabbed newer assyntax.h from the XFree86 project.
690    2. Fiddled with how to compile X86.prim.asm without warnings.
691    3. (Very) Minor cleanup in CM.
692    
693    ----------------------------------------------------------------------
694    Name: Matthias Blume
695    Date: 2002/04/01 (no joke!) 17:07:00 EST
696    Tag: blume-20020401-x86div
697    Description:
698    
699    Added full support for div/mod/rem/quot on the x86, using the machine
700    instruction's two results (without clumsily recomputing the remainder)
701    directly where appropriate.
702    
703    Some more extensive power-of-two support was added to the x86 instruction
704    selector (avoiding expensive divs, mods, and muls where they can be
705    replaced with cheaper shifts and masks).  However, this sort of thing
706    ought to be done earlier, e.g., within the CPS optimizer so that
707    all architectures benefit from it.
708    
709    The compiler compiles to a fixed point, but changes might be somewhat
710    fragile nevertheless.  Please, report any strange things that you might
711    see wrt. div/mod/quot/rem...
712    
713    ----------------------------------------------------------------------
714    Name: Matthias Blume
715    Date: 2002/03/29 17:22:00
716    Tag: blume-20020329-div
717    Description:
718    
719    Fixed my broken div/mod logic.  Unfortunately, this means that the
720    inline code for div/mod now has one more comparison than before.
721    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
722    The problem was with quotient = 0, because that alone does not tell
723    us which way the rounding went.  One then has to look at whether
724    remainder and divisor have the same sign...  :(
725    
726    Anyway, I replaced the bootfiles with fresh ones...
727    
728    ----------------------------------------------------------------------
729    Name: Matthias Blume
730    Date: 2002/03/29 14:10:00 EST
731    Tag: blume-20020329-inlprims
732    Description:
733    
734    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
735    
736    Primops have changed. This means that the bin/boot-file formats have
737    changed as well.
738    
739    To make sure that there is no confusion, I made a new version.
740    
741    
742    CHANGES:
743    
744    * removed REMT from mltree (remainder should never overflow).
745    
746    * added primops to deal with divisions of all flavors to the frontend
747    
748    * handled these primops all the way through so they map to their respective
749      MLRISC support
750    
751    * used these primops in the implementation of Int, Int32, Word, Word32
752    
753    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
754    
755    * parameterized INLMIN, INLMAX, and INLABS by a numkind
756    
757    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
758      floating point
759    
760    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
761      and Real (but Real.abs maps to a separate floating-point-only primop)
762    
763    
764    TODO items:
765    
766    * Hacked Alpha32 instruction selection, disabling the selection of REMx
767      instructions because the machine instruction encoder cannot handle
768      them.  (Hppa, PPC, and Sparc instruction selection did not handle
769      REM in the first place, and REM is supported by the x86 machine coder.)
770    
771    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
772      selection phase.  (The two can be streamlined because the hardware
773      delivers both quotient and remainder at the same time anyway.)
774    
775    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
776      (Currently the behavior is inconsistent both across architectures and
777      wrt. the draft Basis spec.)
778    
779    * Word8 should eventually be handled natively, too.
780    
781    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
782      as if there currently is no execution path that could trigger it in
783      SML/NJ.  (The assumptions underlying functions arith and promotable do not
784      hold for things like multiplication and division.)
785    
786    ----------------------------------------------------------------------
787    Name: Matthias Blume
788    Date: 2002/03/27 16:27:00 EST
789    Tag: blume-20020327-mlrisc-divisions
790    Description:
791    
792    Added support for all four division operations (ML's div, mod, quot,
793    and rem) to MLRISC.  In the course of doing so, I also rationalized
794    the naming (no more annoying switch-around of DIV and QUOT), by
795    parameterizing the operation by div_rounding_mode (which can be either
796    DIV_TO_ZERO or DIV_TO_NEGINF).
797    
798    The generic MLTreeGen functor takes care of compiling all four
799    operations down to only round-to-zero div.
800    
801    Missing pieces:
802    
803      * Doing something smarter than relying on MLTreeGen on architectures
804        like, e.g., the x86 where hardware division delivers both quotient and
805        remainder at the same time.  With this, the implementation of the
806        round-to-neginf operations could be further streamlined.
807    
808      * Remove inlining support for div/mod/rem from the frontend and replace it
809        with primops that get carried through to the backend.  Do this for all
810        int and word types.
811    
812    ----------------------------------------------------------------------
813    Name: Matthias Blume
814    Date: 2002/03/25 17:25:00 EST
815    Tag: blume-20020325-divmod
816    Description:
817    
818    I improved (hopefully without breaking them) the implementation of Int.div,
819    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
820    advantage of the following observations:
821    
822      Let  q = x quot y      r = x rem y
823           d = x div  y      m = x mod y
824    
825    where "quot" is the round-to-zero version of integer division that
826    hardware usually provides.  Then we have:
827    
828         r = x - q * y        where neither the * nor the - will overflow
829         d = if q >= 0 orelse x = q * y then q else q - 1
830                              where neither the * nor the - will overflow
831         m = if q >= 0 orelse r = 0 then r else r + y
832                              where the + will not overflow
833    
834    This results in substantial simplification of the generated code.
835    The following table shows the number of CFG nodes and edges generated
836    for
837            fun f (x, y) = x OPER y
838            (* with OPER \in div, mod, quot, rem *)
839    
840    
841        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
842        --------------------------------------------------------
843         div |         24 |         39 |         12 |         16
844         mod |         41 |         71 |         12 |         16
845        quot |          8 |         10 |          8 |         10
846         rem |         10 |         14 |          8 |         10
847    
848    
849    ----------------------------------------------------------------------
850    Name: Matthias Blume
851    Date: 2002/03/25 22:06:00 EST
852    Tag: blume-20020325-cprotobug
853    Description:
854    
855    Fixed a bug in cproto (c prototype decoder).
856    
857    ----------------------------------------------------------------------
858    Name: Matthias Blume
859    Date: 2002/03/25 16:00:00 EST
860    Tag: blume-20020325-raw-primops
861    Description:
862    
863    I did some cleanup to Allen's new primop code and
864    replaced yesterday's bootfiles with new ones.
865    (But they are stored in the same place.)
866    
867    ----------------------------------------------------------------------
868    Name: Matthias Blume
869    Date: 2002/03/24 22:40:00 EST
870    Tag: blume-20020324-bootfiles
871    Description:
872    
873    Made the bootfiles that Allen asked for.
874    
875    ----------------------------------------------------------------------
876    Name: Allen Leung
877    Date: 2002/03/23 15:50:00 EST
878    Tag: leunga-20020323-flint-cps-rcc-primops
879    Description:
880    
881      1. Changes to FLINT primops:
882    
883        (* make a call to a C-function;
884         * The primop carries C function prototype information and specifies
885         * which of its (ML-) arguments are floating point. C prototype
886         * information is for use by the backend, ML information is for
887         * use by the CPS converter. *)
888      | RAW_CCALL of { c_proto: CTypes.c_proto,
889                       ml_args: ccall_type list,
890                       ml_res_opt: ccall_type option,
891                       reentrant : bool
892                     } option
893       (* Allocate uninitialized storage on the heap.
894        * The record is meant to hold short-lived C objects, i.e., they
895        * are not ML pointers.  With the tag, the representation is
896        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
897        *)
898      | RAW_RECORD of {tag:bool,sz:int}
899      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
900    
901      2.  These CPS primops are now overloaded:
902    
903           rawload of {kind:numkind}
904           rawstore of {kind:numkind}
905    
906          The one argument form is:
907    
908             rawload {kind} address
909    
910          The two argument form is:
911    
912             rawload {kind} [ml object, byte-offset]
913    
914      3. RAW_CCALL/RCC now takes two extra arguments:
915    
916         a. The first is whether the C call is reentrant, i.e., whether
917            ML state should be saved and restored.
918         b. The second argument is a string argument specifying the name of
919            library and the C function.
920    
921         These things are currently not handled in the code generator, yet.
922    
923      4. In CProto,
924    
925         An encoding type of "bool" means "ml object" and is mapped into
926         C prototype of PTR.  Note that "bool" is different than "string",
927         even though "string" is also mapped into PTR, because "bool"
928         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
929    
930      5. Pickler/unpicker
931    
932         Changed to handle RAW_RECORD and newest RAW_CCALL
933    
934      6. MLRiscGen,
935    
936         1. Changed to handle the new rawload/rawstore/rawrecord operators.
937         2. Code for handling C Calls has been moved to a new module CPSCCalls,
938            in the file CodeGen/cpscompile/cps-c-calls.sml
939    
940      7. Added the conditional move operator
941    
942             condmove of branch
943    
944         to cps.  Generation of this is still buggy so it is currently
945         disabled.
946    
947    ----------------------------------------------------------------------
948    Name: Lal George
949    Date: 2002/03/22 14:18:25 EST
950    Tag: george-20020322-cps-branch-prob
951    Description:
952    
953    Implemented the Ball-Larus branch prediction-heuristics, and
954    incorporated graphical viewers for control flow graphs.
955    
956    Ball-Larus Heuristics:
957    ---------------------
958    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
959    
960    By design it uses the Dempster-Shafer theory for combining
961    probabilities.  For example, in the function:
962    
963        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
964    
965    the ball-larus heuristics predicts that the n=0 is unlikely
966    (OH-heuristic), and the 'then' branch is unlikely because of the
967    RH-heuristic -- giving the 'then' branch an even lower combined
968    probability using the Dempster-Shafer theory.
969    
970    Finally, John Reppy's loop analysis in MLRISC, further lowers the
971    probability of the 'then' branch because of the loop in the else
972    branch.
973    
974    
975    Graphical Viewing:
976    ------------------
977    I merely plugged in Allen's graphical viewers into the compiler. The
978    additional code is not much. At the top level, saying:
979    
980            Control.MLRISC.getFlag "cfg-graphical-view" := true;
981    
982    will display the graphical view of the control flow graph just before
983    back-patching.  daVinci must be in your path for this to work. If
984    daVinci is not available, then the default viewer can be changed
985    using:
986    
987            Control.MLRISC.getString "viewer"
988    
989    which can be set to "dot" or "vcg" for the corresponding viewers. Of
990    course, these viewers must be in your path.
991    
992    The above will display the compilation unit at the level of clusters,
993    many of which are small, boring, and un-interesting. Also setting:
994    
995            Control.MLRISC.getInt "cfg-graphical-view_size"
996    
997    will display clusters that are larger than the value set by the above.
998    
999    
1000    ----------------------------------------------------------------------
1001    Name: Matthias Blume
1002    Date: 2002/03/21 22:20:00 EST
1003    Tag: blume-20020321-kmp-bugfix
1004    Description:
1005    
1006    Changed the interface to the KMP routine in PreString and fixed
1007    a minor bug in one place where it was used.
1008    
1009    ----------------------------------------------------------------------
1010    Name: Allen Leung
1011    Date: 2002/03/21 20:30:00 EST
1012    Tag: leunga-20020321-cfg
1013    Description:
1014    
1015      Fixed a potential problem in cfg edge splitting.
1016    
1017    ----------------------------------------------------------------------
1018    Name: Allen Leung
1019    Date: 2002/03/21 17:15:00 EST
1020    Tag: leunga-20020321-x86-fp-cfg
1021    Description:
1022    
1023      1. Recoded the buggy parts of x86-fp.
1024    
1025         a. All the block reordering code has been removed.
1026            We now depend on the block placement phases to do this work.
1027    
1028         b. Critical edge splitting code has been simplified and moved into the
1029            CFG modules, as where they belong.
1030    
1031         Both of these were quite buggy and complex.  The code is now much, much
1032         simpler.
1033    
1034      2. X86 backend.
1035    
1036         a. Added instructions for 64-bit support.  Instruction selection for
1037            64-bit has not been committed, however, since that
1038            requires changes to MLTREE which haven't been approved by
1039            Lal and John.
1040    
1041         b. Added support for FUCOMI and FUCOMIP when generating code for
1042            PentiumPro and above.  We only generate these instructions in
1043            the fast-fp mode.
1044    
1045         c. Added cases for JP and JNP in X86FreqProps.
1046    
1047      3. CFG
1048    
1049         CFG now has a bunch of methods for edge splitting and merging.
1050    
1051      4. Machine description.
1052    
1053         John's simplification of MLTREE_BASIS.fcond broke a few machine
1054         description things:
1055    
1056         rtl-build.{sig,sml} and hppa.mdl fixed.
1057    
1058         NOTE: the machine description stuff in the repository is still broken.
1059               Again, I can't put my fixes in because that involves
1060               changes to MLTREE.
1061    
1062    ----------------------------------------------------------------------
1063    Name: Matthias Blume
1064    Date: 2002/03/20 15:55:00 EST
1065    Tag: blume-20020320-kmp
1066    Description:
1067    
1068    Implemented Knuth-Morris-Pratt string matching in PreString and used
1069    it for String.isSubstring, Substring.isSubstring, and
1070    Substring.position.
1071    
1072    (Might need some stress-testing.  Simple examples worked fine.)
1073    
1074    ----------------------------------------------------------------------
1075    Name: Matthias Blume
1076    Date: 2002/03/19 16:37:00 EST
1077    Tag: blume-20020319-witnesses
1078    Description:
1079    
1080    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1081    
1082    This implements a generic mechanism for changing constness qualifiers
1083    anywhere within big C types without resorting to outright "casts".
1084    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1085    modify the constness at the outermost level.)
1086    The implementation of "convert" is based on the idea of "witness"
1087    values -- values that are not used by the operation but whose types
1088    "testify" to their applicability.  On the implementation side, "convert"
1089    is simply a projection (returning its second curried argument).  With
1090    cross-module inlining, it should not result in any machine code being
1091    generated.
1092    
1093    ----------------------------------------------------------------------
1094    Name: Matthias Blume
1095    Date: 2002/03/15 16:40:00 EST
1096    Tag: blume-20020315-basis
1097    Description:
1098    
1099    Provided (preliminary?) implementations for
1100    
1101      {String,Substring}.{concatWith,isSuffix,isSubstring}
1102    
1103    and
1104    
1105      Substring.full
1106    
1107    Those are in the Basis spec but they were missing in SML/NJ.
1108    
1109    ----------------------------------------------------------------------
1110    Name: Matthias Blume
1111    Date: 2002/03/14 21:30:00 EST
1112    Tag: blume-20020314-controls
1113    Description:
1114    
1115    Controls:
1116    ---------
1117    
1118    1. Factored out the recently-added Controls : CONTROLS stuff and put
1119       it into its own library $/controls-lib.cm.  The source tree for
1120       this is under src/smlnj-lib/Controls.
1121    
1122    2. Changed the names of types and functions in this interface, so they
1123       make a bit more "sense":
1124    
1125          module -> registry
1126          'a registry -> 'a group
1127    
1128    3. The interface now deals in ref cells only.  The getter/setter interface
1129       is (mostly) gone.
1130    
1131    4. Added a function that lets one register an already-existing ref cell.
1132    
1133    5. Made the corresponding modifications to the rest of the code so that
1134       everything compiles again.
1135    
1136    6. Changed the implementation of Controls.MLRISC back to something closer
1137       to the original.  In particular, this module (and therefore MLRISC)
1138       does not depend on Controls.  There now is some link-time code in
1139       int-sys.sml that registers the MLRISC controls with the Controls
1140       module.
1141    
1142    CM:
1143    ---
1144    
1145      * One can now specify the lambda-split aggressiveness in init.cmi.
1146    
1147    ----------------------------------------------------------------------
1148    Name: Allen Leung
1149    Date: 2002/03/13 17:30:00 EST
1150    Tag: leunga-20020313-x86-fp-unary
1151    Description:
1152    
1153    Bug fix for:
1154    
1155    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1156    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1157    > - fun f(x,(y,z)) = Real.~ y;
1158    > [autoloading]
1159    > [autoloading done]
1160    >       fchsl   (%eax), 184(%esp)
1161    > Error: MLRisc bug: X86MCEmitter.emitInstr
1162    >
1163    > uncaught exception Error
1164    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1165    
1166    The problem was that the code generator did not generate any fp registers
1167    in this case, and the ra didn't know that it needed to run the X86FP phase to
1168    translate the pseudo fp instruction.   This only happened with unary fp
1169    operators in certain situations.
1170    
1171    ----------------------------------------------------------------------
1172    Name: Matthias Blume
1173    Date: 2002/03/13 14:00:00 EST
1174    Tag: blume-20020313-overload-etc
1175    Description:
1176    
1177    1. Added _overload as a synonym for overload for backward compatibility.
1178       (Control.overloadKW must be true for either version to be accepted.)
1179    
1180    2. Fixed bug in install script that caused more things to be installed
1181       than what was requested in config/targets.
1182    
1183    3. Made CM aware of the (_)overload construct so that autoloading
1184       works.
1185    
1186    ----------------------------------------------------------------------
1187    Name: Matthias Blume
1188    Date: 2002/03/12 22:03:00 EST
1189    Tag: blume-20020312-url
1190    Description:
1191    
1192    Forgot to update BOOT and srcarchiveurl.
1193    
1194    ----------------------------------------------------------------------
1195    Name: Matthias Blume
1196    Date: 2002/03/12 17:30:00 EST
1197    Tag: blume-20020312-version110392
1198    Description:
1199    
1200    Yet another version number bump (because of small changes to the
1201    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1202    
1203    Changes:
1204    
1205      The new pid generation scheme described a few weeks ago was overly
1206      complicated.  I implemented a new mechanism that is simpler and
1207      provides a bit more "stability":  Once CM has seen a compilation
1208      unit, it keeps its identity constant (as long as you do not delete
1209      those crucial CM/GUID/* files).  This means that when you change
1210      an interface, compile, then go back to the old interface, and
1211      compile again, you arrive at the original pid.
1212    
1213      There now also is a mechanism that instructs CM to use the plain
1214      environment hash as a module's pid (effectively making its GUID
1215      the empty string).  For this, "noguid" must be specified as an
1216      option to the .sml file in question within its .cm file.
1217      This is most useful for code that is being generated by tools such
1218      as ml-nlffigen (because during development programmers tend to
1219      erase the tool's entire output directory tree including CM's cached
1220      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1221      revert to the old, broken behavior of SML/NJ, but in specific cases
1222      where there is no danger of interface confusion, its use is ok
1223      (I think).
1224    
1225      ml-nlffigen by default generates "noguid" annotations.  They can be
1226      turned off by specifying -guid in its command line.
1227    
1228    ----------------------------------------------------------------------
1229    Name: Lal George
1230    Date: 2002/03/12 12 14:42:36 EST
1231    Tag: george-20020312-frequency-computation
1232    Description:
1233    
1234    Integrated jump chaining and static block frequency into the
1235    compiler. More details and numbers later.
1236    
1237    ----------------------------------------------------------------------
1238    Name: Lal George
1239    Date: 2002/03/11 11 22:38:53 EST
1240    Tag: george-20020311-jump-chain-elim
1241    Description:
1242    
1243    Tested the jump chain elimination on all architectures (except the
1244    hppa).  This is on by default right now and is profitable for the
1245    alpha and x86, however, it may not be profitable for the sparc and ppc
1246    when compiling the compiler.
1247    
1248    The gc test will typically jump to a label at the end of the cluster,
1249    where there is another jump to an external cluster containing the actual
1250    code to invoke gc. This is to allow factoring of common gc invocation
1251    sequences. That is to say, we generate:
1252    
1253            f:
1254               testgc
1255               ja   L1      % jump if above to L1
1256    
1257            L1:
1258               jmp L2
1259    
1260    
1261    After jump chain elimination the 'ja L1' instructions is converted to
1262    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1263    up being implemented in their long form (if L2 is far away) using:
1264    
1265            jbe     L3      % jump if below or equal to L3
1266            jmp     L2
1267         L3:
1268            ...
1269    
1270    
1271    For large compilation units L2  may be far away.
1272    
1273    
1274    ----------------------------------------------------------------------
1275    Name: Matthias Blume
1276    Date: 2002/03/11 13:30:00 EST
1277    Tag: blume-20020311-mltreeeval
1278    Description:
1279    
1280    A functor parameter was missing.
1281    
1282    ----------------------------------------------------------------------
1283    Name: Allen Leung
1284    Date: 2002/03/11 10:30:00 EST
1285    Tag: leunga-20020311-runtime-string0
1286    Description:
1287    
1288       The representation of the empty string now points to a
1289    legal null terminated C string instead of unit.  It is now possible
1290    to convert an ML string into C string with InlineT.CharVector.getData.
1291    This compiles into one single machine instruction.
1292    
1293    ----------------------------------------------------------------------
1294    Name: Allen Leung
1295    Date: 2002/03/10 23:55:00 EST
1296    Tag: leunga-20020310-x86-call
1297    Description:
1298    
1299       Added machine generation for CALL instruction (relative displacement mode)
1300    
1301    ----------------------------------------------------------------------
1302    Name: Matthias Blume
1303    Date: 2002/03/08 16:05:00
1304    Tag: blume-20020308-entrypoints
1305    Description:
1306    
1307    Version number bumped to 110.39.1.  NEW BOOTFILES!
1308    
1309    Entrypoints: non-zero offset into a code object where execution should begin.
1310    
1311    - Added the notion of an entrypoint to CodeObj.
1312    - Added reading/writing of entrypoint info to Binfile.
1313    - Made runtime system bootloader aware of entrypoints.
1314    - Use the address of the label of the first function given to mlriscGen
1315      as the entrypoint.  This address is currently always 0, but it will
1316      not be 0 once we turn on block placement.
1317    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
1318      with entry points) from mlriscGen.
1319    
1320    ----------------------------------------------------------------------
1321    Name: Allen Leung
1322    Date: 2002/03/07 20:45:00 EST
1323    Tag: leunga-20020307-x86-cmov
1324    Description:
1325    
1326       Bug fixes for CMOVcc on x86.
1327    
1328       1. Added machine code generation for CMOVcc
1329       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1330       3. CMOVcc cannot have an immediate operand as argument.
1331    
1332    ----------------------------------------------------------------------
1333    Name: Matthias Blume
1334    Date: 2002/03/07 16:15:00 EST
1335    Tag: blume-20020307-controls
1336    Description:
1337    
1338    This is a very large but mostly boring patch which makes (almost)
1339    every tuneable compiler knob (i.e., pretty much everything under
1340    Control.* plus a few other things) configurable via both the command
1341    line and environment variables in the style CM did its configuration
1342    until now.
1343    
1344    Try starting sml with '-h' (or, if you are brave, '-H')
1345    
1346    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1347    implements the underlying generic mechanism.
1348    
1349    The interface to some of the existing such facilities has changed somewhat.
1350    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1351    (The getFoo interface is still there for backward-compatibility, but its
1352    use is deprecated.)
1353    
1354    The ml-build script passes -Cxxx=yyy command-line arguments through so
1355    that one can now twiddle the compiler settings when using this "batch"
1356    compiler.
1357    
1358    TODO items:
1359    
1360    We should go through and throw out all controls that are no longer
1361    connected to anything.  Moreover, we should go through and provide
1362    meaningful (and correct!) documentation strings for those controls
1363    that still are connected.
1364    
1365    Currently, multiple calls to Controls.new are accepted (only the first
1366    has any effect).  Eventually we should make sure that every control
1367    is being made (via Controls.new) exactly once.  Future access can then
1368    be done using Controls.acc.
1369    
1370    Finally, it would probably be a good idea to use the getter-setter
1371    interface to controls rather than ref cells.  For the time being, both
1372    styles are provided by the Controls module, but getter-setter pairs are
1373    better if thread-safety is of any concern because they can be wrapped.
1374    
1375    *****************************************
1376    
1377    One bug fix: The function blockPlacement in three of the MLRISC
1378    backpatch files used to be hard-wired to one of two possibilities at
1379    link time (according to the value of the placementFlag).  But (I
1380    think) it should rather sense the flag every time.
1381    
1382    *****************************************
1383    
1384    Other assorted changes (by other people who did not supply a HISTORY entry):
1385    
1386    1. the cross-module inliner now works much better (Monnier)
1387    2. representation of weights, frequencies, and probabilities in MLRISC
1388       changed in preparation of using those for weighted block placement
1389       (Reppy, George)
1390    
1391    ----------------------------------------------------------------------
1392    Name: Lal George
1393    Date: 2002/03/07 14:44:24 EST 2002
1394    Tag: george-20020307-weighted-block-placement
1395    
1396    Tested the weighted block placement optimization on all architectures
1397    (except the hppa) using AMPL to generate the block and edge frequencies.
1398    Changes were required in the machine properties to correctly
1399    categorize trap instructions. There is an MLRISC flag
1400    "weighted-block-placement" that can be used to enable weighted block
1401    placement, but this will be ineffective without block/edge
1402    frequencies (coming soon).
1403    
1404    
1405    ----------------------------------------------------------------------
1406    Name: Lal George
1407    Date: 2002/03/05 17:24:48 EST
1408    Tag: george-20020305-linkage-cluster
1409    
1410    In order to support the block placement optimization, a new cluster
1411    is generated as the very first cluster (called the linkage cluster).
1412    It contains a single jump to the 'real' entry point for the compilation
1413    unit. Block placement has no effect on the linkage cluster itself, but
1414    all the other clusters  have full freedom in the manner in which they
1415    reorder blocks or functions.
1416    
1417    On the x86 the typical linkage code that is generated is:
1418       ----------------------
1419            .align 2
1420       L0:
1421            addl    $L1-L0, 72(%esp)
1422            jmp     L1
1423    
1424    
1425            .align  2
1426       L1:
1427       ----------------------
1428    
1429    72(%esp) is the memory location for the stdlink register. This
1430    must contain the address of the CPS function being called. In the
1431    above example, it contains the address of  L0; before
1432    calling L1 (the real entry point for the compilation unit), it
1433    must contain the address for L1, and hence
1434    
1435            addl $L1-L0, 72(%esp)
1436    
1437    I have tested this on all architectures except the hppa.The increase
1438    in code size is of course negligible
1439    
1440    ----------------------------------------------------------------------
1441    Name: Allen Leung
1442    Date: 2002/03/03 13:20:00 EST
1443    Tag: leunga-20020303-mlrisc-tools
1444    
1445      Added #[ ... ] expressions to mlrisc tools
1446    
1447    ----------------------------------------------------------------------
1448    Name: Matthias Blume
1449    Date: 2002/02/27 12:29:00 EST
1450    Tag: blume-20020227-cdebug
1451    Description:
1452    
1453    - made types in structure C and C_Debug to be equal
1454    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1455    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1456    
1457    ----------------------------------------------------------------------
1458    Name: Matthias Blume
1459    Date: 2002/02/26 12:00:00 EST
1460    Tag: blume-20020226-ffi
1461    Description:
1462    
1463    1. Fixed a minor bug in CM's "noweb" tool:
1464       If numbering is turned off, then truly don't number (i.e., do not
1465       supply the -L option to noweb).  The previous behavior was to supply
1466       -L'' -- which caused noweb to use the "default" line numbering scheme.
1467       Thanks to Chris Richards for pointing this out (and supplying the fix).
1468    
1469    2. Once again, I reworked some aspects of the FFI:
1470    
1471       A. The incomplete/complete type business:
1472    
1473       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
1474         gone!
1475       - ML types representing an incomplete type are now *equal* to
1476         ML types representing their corresponding complete types (just like
1477         in C).  This is still safe because ml-nlffigen will not generate
1478         RTTI for incomplete types, nor will it generate functions that
1479         require access to such RTTI.   But when ML code generated from both
1480         incomplete and complete versions of the C type meet, the ML types
1481         are trivially interoperable.
1482    
1483         NOTE:  These changes restore the full generality of the translation
1484         (which was previously lost when I eliminated functorization)!
1485    
1486       B. Enum types:
1487    
1488       - Structure C now has a type constructor "enum" that is similar to
1489         how the "su" constructor works.  However, "enum" is not a phantom
1490         type because each "T enum" has values (and is isomorphic to
1491         MLRep.Signed.int).
1492       - There are generic access operations for enum objects (using
1493         MLRep.Signed.int).
1494       - ml-nlffigen will generate a structure E_foo for each "enum foo".
1495         * The structure contains the definition of type "mlrep" (the ML-side
1496         representation type of the enum).  Normally, mlrep is the same
1497         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
1498         then mlrep will be defined as a datatype -- thus facilitating
1499         pattern matching on mlrep values.
1500         ("-ec" will be suppressed if there are duplicate values in an
1501          enumeration.)
1502         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
1503         will be generated for each C enum constant xxx.
1504         * Conversion functions m2i and i2m convert between mlrep and
1505         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
1506         * Coversion functions c and ml convert between mlrep and "tag enum".
1507         * Access functions (get/set) fetch and store mlrep values.
1508       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
1509         enumerations are merged into one single enumeration represented by
1510         structure E_'.
1511    
1512    ----------------------------------------------------------------------
1513    Name: Allen Leung
1514    Date: 2002/02/25 04:45:00 EST
1515    Tag: leunga-20020225-cps-spill
1516    
1517    This is a new implementation of the CPS spill phase.
1518    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
1519    In case of problems, replace it with the old file spill.sml
1520    
1521    The current compiler runs into some serious performance problems when
1522    constructing a large record.  This can happen when we try to compile a
1523    structure with many items.  Even a very simple structure like the following
1524    makes the compiler slow down.
1525    
1526        structure Foo = struct
1527           val x_1 = 0w1 : Word32.int
1528           val x_2 = 0w2 : Word32.int
1529           val x_3 = 0w3 : Word32.int
1530           ...
1531           val x_N = 0wN : Word32.int
1532        end
1533    
1534    The following table shows the compile time, from N=1000 to N=4000,
1535    with the old compiler:
1536    
1537    N
1538    1000   CPS 100 spill                           0.04u  0.00s  0.00g
1539           MLRISC ra                               0.06u  0.00s  0.05g
1540              (spills = 0 reloads = 0)
1541           TOTAL                                   0.63u  0.07s  0.21g
1542    
1543    1100   CPS 100 spill                           8.25u  0.32s  0.64g
1544           MLRISC ra                               5.68u  0.59s  3.93g
1545              (spills = 0 reloads = 0)
1546           TOTAL                                   14.71u  0.99s  4.81g
1547    
1548    1500   CPS 100 spill                           58.55u  2.34s  1.74g
1549           MLRISC ra                               5.54u  0.65s  3.91g
1550              (spills = 543 reloads = 1082)
1551           TOTAL                                   65.40u  3.13s  6.00g
1552    
1553    2000   CPS 100 spill                           126.69u  4.84s  3.08g
1554           MLRISC ra                               0.80u  0.10s  0.55g
1555              (spills = 42 reloads = 84)
1556           TOTAL                                   129.42u  5.10s  4.13g
1557    
1558    3000   CPS 100 spill                           675.59u  19.03s  11.64g
1559           MLRISC ra                               2.69u  0.27s  1.38g
1560              (spills = 62 reloads = 124)
1561           TOTAL                                   682.48u  19.61s  13.99g
1562    
1563    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
1564           MLRISC ra                               4.96u  0.27s  2.72g
1565              (spills = 85 reloads = 170)
1566           TOTAL                                   2375.26u  57.21s  48.00g
1567    
1568    As you can see the old cps spill module suffers from some serious
1569    performance problem.  But since I cannot decipher the old code fully,
1570    instead of patching the problems up, I'm reimplementing it
1571    with a different algorithm.  The new code is more modular,
1572    smaller when compiled, and substantially faster
1573    (O(n log n) time and O(n) space).  Timing of the new spill module:
1574    
1575    4000  CPS 100 spill                           0.02u  0.00s  0.00g
1576          MLRISC ra                               0.25u  0.02s  0.15g
1577             (spills=1 reloads=3)
1578          TOTAL                                   7.74u  0.34s  1.62g
1579    
1580    Implementation details:
1581    
1582    As far as I can tell, the purpose of the CPS spill module is to make sure the
1583    number of live variables at any program point (the bandwidth)
1584    does not exceed a certain limit, which is determined by the
1585    size of the spill area.
1586    
1587    When the bandwidth is too large, we decrease the register pressure by
1588    packing live variables into spill records.  How we achieve this is
1589    completely different than what we did in the old code.
1590    
1591    First, there is something about the MLRiscGen code generator
1592    that we should be aware of:
1593    
1594    o MLRiscGen performs code motion!
1595    
1596       In particular, it will move floating point computations and
1597       address computations involving only the heap pointer to
1598       their use sites (if there is only a single use).
1599       What this means is that if we have a CPS record construction
1600       statement
1601    
1602           RECORD(k,vl,w,e)
1603    
1604       we should never count the new record address w as live if w
1605       has only one use (which is often the case).
1606    
1607       We should do something similar to floating point, but the transformation
1608       there is much more complex, so I won't deal with that.
1609    
1610    Secondly, there are now two new cps primops at our disposal:
1611    
1612     1. rawrecord of record_kind option
1613        This pure operator allocates some uninitialized storage from the heap.
1614        There are two forms:
1615    
1616         rawrecord NONE [INT n]  allocates a tagless record of length n
1617         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
1618                                     and initializes the tag.
1619    
1620     2. rawupdate of cty
1621          rawupdate cty (v,i,x)
1622          Assigns to x to the ith component of record v.
1623          The storelist is not updated.
1624    
1625    We use these new primops for both spilling and increment record construction.
1626    
1627     1. Spilling.
1628    
1629        This is implemented with a linear scan algorithm (but generalized
1630        to trees).  The algorithm will create a single spill record at the
1631        beginning of the cps function and use rawupdate to spill to it,
1632        and SELECT or SELp to reload from it.  So both spills and reloads
1633        are fine-grain operations.  In contrast, in the old algorithm
1634        "spills" have to be bundled together in records.
1635    
1636        Ideally, we should sink the spill record construction to where
1637        it is needed.  We can even split the spill record into multiple ones
1638        at the places where they are needed.  But CPS is not a good
1639        representation for global code motion, so I'll keep it simple and
1640        am not attempting this.
1641    
1642     2. Incremental record construction (aka record splitting).
1643    
1644        Long records with many component values which are simulatenously live
1645        (recall that single use record addresses are not considered to
1646         be live) are constructed with rawrecord and rawupdate.
1647        We allocate space on the heap with rawrecord first, then gradually
1648        fill it in with rawupdate.  This is the technique suggested to me
1649        by Matthias.
1650    
1651        Some restrictions on when this is applicable:
1652        1. It is not a VECTOR record.  The code generator currently does not handle
1653           this case. VECTOR record uses double indirection like arrays.
1654        2. All the record component values are defined in the same "basic block"
1655           as the record constructor.  This is to prevent speculative
1656           record construction.
1657    
1658    ----------------------------------------------------------------------
1659    Name: Allen Leung
1660    Date: 2002/02/22 01:02:00 EST
1661    Tag: leunga-20020222-mlrisc-tools
1662    
1663    Minor bug fixes in the parser and rewriter
1664    
1665    ----------------------------------------------------------------------
1666    Name: Allen Leung
1667    Date: 2002/02/21 20:20:00 EST
1668    Tag: leunga-20020221-peephole
1669    
1670    Regenerated the peephole files.  Some contained typos in the specification
1671    and some didn't compile because of pretty printing bugs in the old version
1672    of 'nowhere'.
1673    
1674    ----------------------------------------------------------------------
1675    Name: Allen Leung
1676    Date: 2002/02/19 20:20:00 EST
1677    Tag: leunga-20020219-mlrisc-tools
1678    Description:
1679    
1680       Minor bug fixes to the mlrisc-tools library:
1681    
1682       1.  Fixed up parsing colon suffixed keywords
1683       2.  Added the ability to shut the error messages up
1684       3.  Reimplemented the pretty printer and fixed up/improved
1685           the pretty printing of handle and -> types.
1686       4.  Fixed up generation of literal symbols in the nowhere tool.
1687       5.  Added some SML keywords to to sml.sty
1688    
1689    ----------------------------------------------------------------------
1690    Name: Matthias Blume
1691    Date: 2002/02/19 16:20:00 EST
1692    Tag: blume-20020219-cmffi
1693    Description:
1694    
1695    A wild mix of changes, some minor, some major:
1696    
1697    * All C FFI-related libraries are now anchored under $c:
1698        $/c.cm      --> $c/c.cm
1699        $/c-int.cm  --> $c/internals/c-int.cm
1700        $/memory.cm --> $c/memory/memory.cm
1701    
1702    * "make" tool (in CM) now treats its argument pathname slightly
1703      differently:
1704        1. If the native expansion is an absolute name, then before invoking
1705           the "make" command on it, CM will apply OS.Path.mkRelative
1706           (with relativeTo = OS.FileSys.getDir()) to it.
1707        2. The argument will be passed through to subsequent phases of CM
1708           processing without "going native".  In particular, if the argument
1709           was an anchored path, then "make" will not lose track of that anchor.
1710    
1711    * Compiler backends now "know" their respective C calling conventions
1712      instead of having to be told about it by ml-nlffigen.  This relieves
1713      ml-nlffigen from one of its burdens.
1714    
1715    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
1716    
1717    * Export C_DEBUG and C_Debug from $c/c.cm.
1718    
1719    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
1720      subtyping relationship between incomplete pointers and their complete
1721      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
1722      with the convention of instantiating 'o with ('t, 'c) obj whenever
1723      the pointer target type is complete.  In the incomplete case, 'o
1724      will be instantiated with some "'c iobj" -- a type obtained by
1725      using one of the functors PointerToIncompleteType or PointerToCompleteType.
1726    
1727      Operations that work on both incomplete and complete pointer types are
1728      typed as taking an 'o ptr while operations that require the target to
1729      be known are typed as taking some ('t, 'c) obj ptr.
1730    
1731      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
1732      where void is an eqtype without any values.  This makes it possible
1733      to work on voidptr values using functions meant to operate on light
1734      incomplete pointers.
1735    
1736    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
1737      been vastly simplified.
1738    
1739    ----------------------------------------------------------------------
1740    Name: Matthias Blume
1741    Date: 2002/02/19 10:48:00 EST
1742    Tag: blume-20020219-pqfix
1743    Description:
1744    
1745    Applied Chris Okasaki's bug fix for priority queues.
1746    
1747    ----------------------------------------------------------------------
1748    Name: Matthias Blume
1749    Date: 2002/02/15 17:05:00
1750    Tag: Release_110_39
1751    Description:
1752    
1753    Last-minute retagging is becoming a tradition... :-(
1754    
1755    This is the working release 110.39.
1756    
1757    ----------------------------------------------------------------------
1758    Name: Matthias Blume
1759    Date: 2002/02/15 16:00:00 EST
1760    Tag: Release_110_39-orig
1761    Description:
1762    
1763    Working release 110.39.  New bootfiles.
1764    
1765    (Update: There was a small bug in the installer so it wouldn't work
1766    with all shells.  So I retagged. -Matthias)
1767    
1768    ----------------------------------------------------------------------
1769    Name: Matthias Blume
1770    Date: 2002/02/15 14:17:00 EST
1771    Tag: blume-20020215-showbindings
1772    Description:
1773    
1774    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
1775    the latter can be useful for exploring what bindings are available at
1776    the interactive prompt.  (The first function returns only the list
1777    of symbols that are really bound, the second prints those but also the
1778    ones that CM's autoloading mechanism knows about.)
1779    
1780    ----------------------------------------------------------------------
1781    Name: Matthias Blume
1782    Date: 2002/02/15 12:08:00 EST
1783    Tag: blume-20020215-iptrs
1784    Description:
1785    
1786    Two improvements to ml-nlffigen:
1787    
1788      1. Write files only if they do not exist or if their current contents
1789         do not coincide with what's being written.  (That is, avoid messing
1790         with the time stamps unless absolutely necessary.)
1791    
1792      2. Implement a "repository" mechanism for generated files related
1793         to "incomplete pointer types".   See the README file for details.
1794    
1795    ----------------------------------------------------------------------
1796    Name: Matthias Blume
1797    Date: 2002/02/14 11:50:00 EST
1798    Tag: blume-20020214-quote
1799    Description:
1800    
1801    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
1802    because of the new and improved tag generation scheme.  (Thanks to Allen
1803    Leung for pointing it out.)
1804    
1805    ----------------------------------------------------------------------
1806    Name: Lal George
1807    Date: 2002/02/14 09:55:27 EST 2002
1808    Tag: george-20020214-isabelle-bug
1809    Description:
1810    
1811    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
1812    of Isabelle on the x86.
1813    
1814    From Allen:
1815    -----------
1816     I've found the problem:
1817    
1818         in ra-core.sml, I use the counter "blocked" to keep track of the
1819         true number of elements in the freeze queue.  When the counter goes
1820         to zero, I skip examining the queue.  But I've messed up the
1821         bookkeeping in combine():
1822    
1823             else ();
1824             case !ucol of
1825               PSEUDO => (if !cntv > 0 then
1826                     (if !cntu > 0 then blocked := !blocked - 1 else ();
1827                                        ^^^^^^^^^^^^^^^^^^^^^^^
1828                      moveu := mergeMoveList(!movev, !moveu)
1829                     )
1830                  else ();
1831    
1832         combine() is called to coalesce two nodes u and v.
1833         I think I was thinking that if the move counts of u and v are both
1834         greater than zero then after they are coalesced then one node is
1835         removed from the freeze queue.  Apparently I was thinking that
1836         both u and v are of low degree, but that's clearly not necessarily true.
1837    
1838    
1839    02/12/2002:
1840        Here's the patch.  HOL now compiles.
1841    
1842        I don't know how this impact on performance (compile
1843        time or runtime).  This bug caused the RA (especially on the x86)
1844        to go thru the potential spill phase when there are still nodes on the
1845        freeze queue.
1846    
1847    
1848    
1849    
1850    ----------------------------------------------------------------------
1851    Name: Matthias Blume
1852    Date: 2002/02/13 22:40:00 EST
1853    Tag: blume-20020213-fptr-rtti
1854    Description:
1855    
1856    Fixed a bug in ml-nlffigen that was introduced with one of the previous
1857    updates.
1858    
1859    ----------------------------------------------------------------------
1860    Name: Matthias Blume
1861    Date: 2002/02/13 16:41:00 EST
1862    Tag: blume-20020213-cmlpq
1863    Description:
1864    
1865    Added new priority queue export symbols (which have just been added to
1866    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
1867    not compile and the installer would choke.)
1868    
1869    ----------------------------------------------------------------------
1870    Name: Matthias Blume
1871    Date: 2002/02/13 16:15:00 EST
1872    Tag: blume-20020213-various
1873    Description:
1874    
1875    1. More tweaks to ml-nlffigen:
1876    
1877       - better internal datastructures (resulting in slight speedup)
1878       - "-match" option requires exact match
1879       - "localized" gensym counters (untagged structs/unions nested within
1880         other structs/unions or within typedefs get a fresh counter; their
1881         tag will be prefixed by a concatenation of their parents' tags)
1882       - bug fixes (related to calculation of transitive closure of types
1883         to be included in the output)
1884    
1885    2. Minor Basis updates:
1886    
1887       - added implementations for List.collate and Option.app
1888    
1889    ----------------------------------------------------------------------
1890    Name: Matthias Blume
1891    Date: 2002/02/11 15:55:00 EST
1892    Tag: blume-20020211-gensym
1893    Description:
1894    
1895    Added a "-gensym" option to command line of ml-nlffigen.  This can be
1896    used to specify a "stem" -- a string that is inserted in all "gensym'd"
1897    names (ML structure names that correspond to unnamed C structs, unions,
1898    and enums), so that separate runs of ml-nlffigen do not clash.
1899    
1900    ----------------------------------------------------------------------
1901    Name: Matthias Blume
1902    Date: 2002/02/11 12:05:00 EST
1903    Tag: blume-20020211-gensml
1904    Description:
1905    
1906    A quick fix for a problem with GenSML (in the pgraph-util library):
1907    Make generation of toplevel "local" optional.  (Strictly speaking,
1908    signature definitions within "local" are not legal SML.)
1909    
1910    Other than that: updates to INSTALL and cm/TODO.
1911    
1912    ----------------------------------------------------------------------
1913    Name: Matthias Blume
1914    Date: 2002/02/08 15:00:00 EST
1915    Tag: blume-20020208-uniquepid
1916    Description:
1917    
1918    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
1919    
1920    1. The installer (config/install.sh) has gotten smarter:
1921    
1922         - Configuration options are a bit easier to specify now
1923           (in config/targets).
1924         - Bug in recognizing .tar.bz2 files fixed.
1925         - Installer automatically resolves dependencies between
1926           configuration options (e.g., if you ask for eXene, you will
1927           also get cml -- regardless whether you asked for it or not).
1928         - Installer can run in "quieter mode" by setting the environment
1929           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
1930           "completely silent", though.
1931         - Build HashCons library as part of smlnj-lib.
1932    
1933    2. A new scheme for assigning persistent identifiers to compilation
1934       units (and, by extension, to types etc.) has been put into place.
1935       This fixes a long-standing bug where types and even dynamic values
1936       can get internally confused, thereby compromising type safety
1937       (abstraction) and dynamic correctness.  See
1938    
1939         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
1940    
1941       for an example of how things could go wrong until now.
1942    
1943       The downside of the new scheme is that pids are not quite as
1944       persistent as they used to be: CM will generate a fresh pid
1945       for every compilation unit that it thinks it sees for the first
1946       time.  That means that if you compile starting from a clean, fresh
1947       source tree at two different times, you end up with different
1948       binaries.
1949    
1950       Cutoff recompilation, however, has not been compromised because
1951       CM keeps pid information in special caches between runs.
1952    
1953    ----------------------------------------------------------------------
1954    Name: Lal George
1955    Date: 2002/02/07 15:34:13 EST 2002
1956    Tag: <none>
1957    Description:
1958    
1959    Compilers that generate assembly code may produce  global labels
1960    whose value is resolved at link time. The various peephole optimization
1961    modules did not take this in account.
1962    
1963    TODO. The Labels.addrOf function should really return an option
1964    type so that clients are forced to deal with this issue, rather
1965    than an exception being raised.
1966    
1967    ----------------------------------------------------------------------
1968    Name: Lal George
1969    Date: 2002/02/06 13:55:02 EST
1970    Tag: george-20020206-ra-breakup
1971    Description:
1972    
1973    1. A bug fix from Allen.
1974    
1975        A typo causes extra fstp %st(0)'s to be generated at compensation
1976        edges, which might cause stack underflow traps at runtime.  This
1977        occurs in fft where there are extraneous fstps right before the 'into'
1978        trap instruction (in this case they are harmless since none of the
1979        integers overflow.)
1980    
1981    2. Pulled out various utility modules that were embedded in the modules
1982       of the register allocator. I need these modules for other purposes, but
1983       they are not complete enough to put into a library (just yet).
1984    ----------------------------------------------------------------------
1985    Name: Matthias Blume
1986    Date: 2002/01/31 16:05:00 EST
1987    Tag: blume-20020131-sparc-ccalls
1988    Description:
1989    
1990    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
1991       of extra stack space by mistake.  Fixed.
1992    
1993    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
1994    
1995    ----------------------------------------------------------------------
1996    Name: Allen Leung
1997    Date: 2002/01/30
1998    Tag: leunga-20020130-nowhere-bug-fix
1999    Description:
2000    
2001       MLRISC bug fixes:
2002       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2003       2. MachineInt.fromString was negating its value.
2004    
2005    ----------------------------------------------------------------------
2006    Name: Matthias Blume
2007    Date: 2002/01/29
2008    Tag: blume-20020129-INSTALL
2009    Description:
2010    
2011    - Added somewhat detailed installation instructions (file INSTALL).
2012    - Fixed curl-detection bug in config/install.sh.
2013    - It is now possible to select the URL getter using the URLGETTER
2014      environment variable:
2015    
2016          not set / "unknown"      --> automatic detection (script tries wget,
2017                                       curl, and lynx)
2018          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2019                                       how to properly invoke them)
2020          other                    --> use $URLGETTER directly, it must take
2021                                       precisely two command-line arguments
2022                                       (source URL and destination file name)
2023    
2024    ----------------------------------------------------------------------
2025    Name: Matthias Blume
2026    Date: 2002/01/28
2027    Tag: blume-20020128-sparc-ccalls
2028    Description:
2029    
2030    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2031    - Make use of the allocParam argument in sparc-c-calls.
2032    
2033    ----------------------------------------------------------------------
2034    Name: Matthias Blume
2035    Date: 2002/01/28
2036    Tag: blume-20020128-allocParam
2037    Description:
2038    
2039    John Reppy:  Changes c-calls API to accept client-callback for
2040    allocating extra stack space.
2041    me: Corresponding changes to mlriscGen (using a dummy argument that
2042        does not change the current behavior).
2043    
2044    ----------------------------------------------------------------------
2045    Name: Matthias Blume
2046    Date: 2002/01/28 12:00:00
2047    Tag: Release_110_38
2048    Description:
2049    
2050    This time for real!!!
2051    
2052    ----------------------------------------------------------------------
2053    Name: Matthias Blume
2054    Date: 2002/01/28 10:56:00 EST
2055    Tag: blume-20020128-retraction
2056    Description:
2057    
2058    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
2059       with blume-Release_110_38-retracted.)
2060    
2061    1. Fixed a problem with incorrect rounding modes in real64.sml.
2062       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2063    
2064    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2065       and enums fixed.  The naming of corresponding ML identifiers should
2066       now be consistent again.
2067    
2068    ----------------------------------------------------------------------
2069    Name: Allen Leung
2070    Date: 2002/01/27
2071    Tag: leunga-20020127-nowhere
2072    Description:
2073    
2074       Added a target called nowhere in the configuration scripts.
2075       Enabling this will build the MLRISC 'nowhere' tool (for translating
2076       programs with where-clauses into legal SML code) during installation.
2077    
2078    ----------------------------------------------------------------------
2079    Name: Matthias Blume
2080    Date: 2002/01/25 21:27:00 EST
2081    Tag: blume-Release_110_38-retracted
2082    Description:
2083    
2084    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
2085    
2086    README will be added later.
2087    
2088    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
2089    (Release_110_38) removed.  Reason: Last-minute bug fixes.
2090    
2091    ----------------------------------------------------------------------
2092    Name: Matthias Blume
2093    Date: 2002/01/25
2094    Tag: blume-20020125-ffi
2095    Description:
2096    
2097    A large number of tweaks and improvements to ml-nlffi-lib and
2098    ml-nlffigen:
2099    
2100       - ML represenation types have been streamlined
2101       - getter and setter functions work with concrete values, not abstract
2102         ones where possible
2103       - ml-nlffigen command line more flexible (see README file there)
2104       - some bugs have been fixed (hopefully)
2105    
2106    ----------------------------------------------------------------------
2107    Name: Lal George
2108    Date: 2002/01/24
2109    Tag: george-20020124-risc-ra-interface
2110    Description:
2111    
2112       There is a dramatic simplification in the interface to the
2113       register allocator for RISC architectures as a result of making
2114       parallel copy instructions explicit.
2115    
2116    ----------------------------------------------------------------------
2117    Name: Matthias Blume
2118    Date: 2002/01/22
2119    Tag: blume-20020122-x86-ccalls
2120    Description:
2121    
2122    Bug fix for c-calls on x86 (having to do with how char- and
2123    short-arguments are being handled).
2124    
2125    ----------------------------------------------------------------------
2126    Name: Matthias Blume
2127    Date: 2002/01/21
2128    Tag: blume-20020121-ff
2129    Description:
2130    
2131    Another day of fiddling with the FFI...
2132    
2133    1. Bug fix/workaround:  CKIT does not complain about negative array
2134       dimensions, so ml-nlffigen has to guard itself against this possibility.
2135       (Otherwise a negative dimension would send it into an infinite loop.)
2136    
2137    2. Some of the abstract types (light objects, light pointers, most "base"
2138       types) in structure C are now eqtypes.
2139    
2140    3. Added constructors and test functions for NULL function pointers.
2141    
2142    ----------------------------------------------------------------------
2143    Name: Matthias Blume
2144    Date: 2002/01/18
2145    Tag: blume-20020118-ready-for-new-release
2146    Description:
2147    
2148    Made config/srcarchiveurl point to a new place.  (Will provide boot
2149    files shortly.)
2150    
2151    Maybe we christen this to be 110.38?
2152    
2153    ----------------------------------------------------------------------
2154    Name: Matthias Blume
2155    Date: 2002/01/18
2156    Tag: blume-20020118-more-ffifiddle
2157    Description:
2158    
2159    Today's FFI fiddling:
2160    
2161      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
2162        C.Set.* that use concrete (MLRep.*) arguments and results instead
2163        of abstract ones.
2164    
2165      - Provided word-style bit operations etc. for "int" representation
2166        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
2167        and Long.
2168    
2169    ----------------------------------------------------------------------
2170    Name: Matthias Blume
2171    Date: 2002/01/18
2172    Tag: blume-20020118-use-x86-fp
2173    Description:
2174    
2175    Now that x86-fast-fp seems to be working, I turned it back on again
2176    by default.  (Seems to work fine now, even with the FFI.)
2177    
2178    Other than that, I added some documentation about the FFI to
2179    src/ml-nlffigen/README and updated the FFI test examples in
2180    src/ml-nlffi-lib/Tests/*.
2181    
2182    ----------------------------------------------------------------------
2183    Name: Allen Leung
2184    Date: 2002/01/17
2185    Tag: leunga-20020117-x86-fast-fp-call
2186    Description:
2187    
2188       1. Fixed a problem with handling return fp values when x86's fast fp
2189          mode is turned on.
2190    
2191       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
2192          of %f32.
2193    
2194       3. Added a constructor INT32lit to the ast of MLRISC tools.
2195    
2196    ----------------------------------------------------------------------
2197    Name: Matthias Blume
2198    Date: 2002/01/16
2199    Tag: blume-20020116-ffifiddle
2200    Description:
2201    
2202    More fiddling with the FFI interface:
2203    
2204     - Make constness 'c instead of rw wherever possible.  This eliminates
2205       the need for certain explicit coercions.  (However, due to ML's
2206       value polymorphism, there will still be many cases where explicit
2207       coercions are necessary.  Phantom types are not the whole answer
2208       to modeling a subtyping relationship in ML.)
2209    
2210     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
2211    
2212     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
2213       clumsy workaround for varargs problem.)
2214    
2215    ----------------------------------------------------------------------
2216    Name: Lal George
2217    Date: 2002/01/15
2218    Tag: <none>
2219    Description:
2220    
2221    1. Since COPY instructions are no longer native to the architecture,
2222       a generic functor can be used to implement the expandCopies function.
2223    
2224    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
2225       TEXT segment.
2226    
2227    ----------------------------------------------------------------------
2228    Name: Matthias Blume
2229    Date: 2002/01/15
2230    Tag: blume-20020115-ffiupdates
2231    Description:
2232    
2233    1. Fix for bug resulting in single-precision float values being returned
2234       incorrectly from FFI calls.
2235    
2236    2. Small modifications to C FFI API:
2237    
2238        - memory-allocation routines return straight objects (no options)
2239          and raise an exception in out-of-memory situations
2240        - unsafe extensions to cast between function pointers and pointers
2241          from/to ints
2242        - added structure C_Debug as an alternative to structure C where
2243          pointer-dereferencing (|*| and |*!) always check for null-pointers
2244        - added open_lib' to DynLinkage;  open_lib' works like open_lib
2245          but also takes a (possibly empty) list of existing library handles
2246          that the current library depends on
2247    
2248    ----------------------------------------------------------------------
2249    Name: Matthias Blume
2250    Date: 2002/01/10
2251    Tag: blume-20020110-newffigen
2252    Description:
2253    
2254    1. Updates to portable graph code.
2255    
2256    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
2257       more scalable now so that even huge interfaces such as the one
2258       for GTK compile in finite time and space. :-)
2259       See src/ml-nlffigen/README for details on what's new.
2260    
2261    ----------------------------------------------------------------------
2262    Name: Lal George
2263    Date: 2001/01/09 14:31:35 EST 2002
2264    Tag: george-20011206-rm-native-copy
2265    Description:
2266    
2267            Removed the native COPY and FCOPY instructions
2268            from all the architectures and replaced it with the
2269            explicit COPY instruction from the previous commit.
2270    
2271            It is now possible to simplify many of the optimizations
2272            modules that manipulate copies. This has not been
2273            done in this change.
2274    
2275    ----------------------------------------------------------------------
2276    Name: Lal George
2277    Date: 2001/12/06 16:50:13 EST 2001
2278    Tag: george-20011206-mlrisc-instruction
2279    Description:
2280    
2281    Changed the representation of instructions from being fully abstract
2282    to being partially concrete. That is to say:
2283    
2284      from
2285            type instruction
2286    
2287      to
2288            type instr                              (* machine instruction *)
2289    
2290            datatype instruction =
2291                LIVE of {regs: C.cellset, spilled: C.cellset}
2292              | KILL of {regs: C.cellset, spilled: C.cellset}
2293              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
2294              | ANNOTATION of {i: instruction, a: Annotations.annotation}
2295              | INSTR of instr
2296    
2297    This makes the handling of certain special instructions that appear on
2298    all architectures easier and uniform.
2299    
2300    LIVE and KILL say that a list of registers are live or killed at the
2301    program point where they appear. No spill code is generated when an
2302    element of the 'regs' field is spilled, but the register is moved to
2303    the 'spilled' (which is present, more for debugging than anything else).
2304    
2305    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
2306    We used to generate:
2307    
2308            DEFFREG f1
2309            f1 := f2 + f3
2310            trapb
2311    
2312    but now generate:
2313    
2314            f1 := f2 + f3
2315            trapb
2316            LIVE {regs=[f1,f2,f3], spilled=[]}
2317    
2318    Furthermore, the DEFFREG (hack) required that all floating point instruction
2319    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
2320    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
2321    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
2322    this architecture).
2323    
2324    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
2325    all the architectures. This will result in further simplification of the
2326    register allocator that must be aware of them for coalescing purposes, and
2327    will also simplify certain aspects of the machine description that provides
2328    callbacks related to parallel copies.
2329    
2330    ANNOTATION should be obvious, and now INSTR represents the honest to God
2331    machine instruction set!
2332    
2333    The <arch>/instructions/<arch>Instr.sml files define certain utility
2334    functions for making porting easier -- essentially converting upper case
2335    to lower case. All machine instructions (of type instr) are in upper case,
2336    and the lower case form generates an MLRISC instruction. For example on
2337    the alpha we have:
2338    
2339      datatype instr =
2340         LDA of {r:cell, b:cell, d:operand}
2341       | ...
2342    
2343      val lda : {r:cell, b:cell, d:operand} -> instruction
2344        ...
2345    
2346    where lda is just (INSTR o LDA), etc.
2347    
2348    ----------------------------------------------------------------------
2349    Name: Matthias Blume
2350    Date: 2001/11/22 21:40:00 EST
2351    Tag: Release_110_37
2352    Description:
2353    
2354    Release 110.37.  This time for real.
2355    
2356    ----------------------------------------------------------------------
2357    Name: Matthias Blume
2358    Date: 2001/11/21 16:35:00 EST
2359    Tag: blume-20011121-foot-in-mouth
2360    Description:
2361    
2362    Removed the "Release_110_37" tag because of a serious bug.
2363    This will be re-tagged once the bug is fixed.
2364    
2365    ----------------------------------------------------------------------
2366    Name: Matthias Blume
2367    Date: 2001/11/21 16:14:00 EST
2368    Tag: blume-20011121-forgottenfile
2369    Description:
2370    
2371    Forgot to add a file.  (Just a .tex-file -- part of
2372    the CM manual source.)
2373    
2374    ----------------------------------------------------------------------
2375    Name: Matthias Blume
2376    Date: 2001/11/21 16:10:00 EST
2377    Tag: blume-20011121-invalid_110_37
2378    Description:
2379    
2380    Note: I removed the original tag "Release_110_37" from this commit
2381          because we found a serious bug in all non-x86 backends.
2382          - Matthias
2383    
2384    1. Modifications to the SML/NJ code generator and to the runtime system
2385       so that code object name strings are directly inserted into code
2386       objects at code generation time.  The only business the runtime system
2387       has with this is now to read the name strings on occasions.
2388       (The encoding of the name string has also changed somewhat.)
2389    
2390    2. CM now implements a simple "set calculus" for specifying export lists.
2391       In particular, it is now possible to refer to the export lists of
2392       other libraries/groups/sources and form unions as well as differences.
2393       See the latest CM manual for details.
2394    
2395    3. An separate notion of "proxy" libraries has again be eliminated from
2396       CM's model.  (Proxy libraries are now simply a special case of using
2397       the export list calculus.)
2398    
2399    4. Some of the existing libraries now take advantage of the new set
2400       calculus.
2401       (Notice that not all libraries have been converted because some
2402       of the existing .cm-files are supposed to be backward compatible
2403       with 110.0.x.)
2404    
2405    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
2406       "General.exnMessage"!  The former relies on a certain hook to be
2407       initialized, and that often does not happen in the stand-alone case.)
2408    
2409    ----------------------------------------------------------------------
2410    Name: Lal George
2411    Date: 2001/11/21  13:56:18 EST
2412    Tag: george-2001121-pseudo-ops
2413    Description:
2414    
2415      Implemented a complete redesign of MLRISC pseudo-ops. Now there
2416      ought to never be any question of incompatabilities with
2417      pseudo-op syntax expected by host assemblers.
2418    
2419      For now, only modules supporting GAS syntax are implemented
2420      but more should follow, such as MASM, and vendor assembler
2421      syntax, e.g. IBM as, Sun as, etc.
2422    
2423    ----------------------------------------------------------------------
2424    Name: Matthias Blume
2425    Date: 2001/11/14 11:52:00 EST
2426    Tag: blume-20011114-srcname
2427    Description:
2428    
2429    1. Routed the name of the current source file to mlriscgen where it
2430       should be directly emitted into the code object.  (This last part
2431       is yet to be done.)
2432    
2433    2. Some cleanup of the pgraph code to make it match the proposal that
2434       I put out the other day.  (The proposal notwithstanding, things are
2435       still in flux here.)
2436    
2437    ----------------------------------------------------------------------
2438    Name: Lal George
2439    Date: 2001/11/14 09:44:04 EST
2440    Tag:
2441    Description:
2442    
2443      Fix for a backpatching bug reported by Allen.
2444    
2445      Because the boundary between short and long span-dependent
2446      instructions is +/- 128, there are an astounding number of
2447      span-dependent instructions whose size is over estimated.
2448    
2449      Allen came up with the idea of letting the size of span
2450      dependent instructions be non-monotonic, for a maxIter
2451      number of times, after which the size must be monotonically
2452      increasing.
2453    
2454      This table shows the number of span-dependent instructions
2455      whose size was over-estimated as a function of maxIter, for the
2456      file Parse/parse/ml.grm.sml:
2457    
2458         maxIter            # of instructions:
2459            10                      687
2460            20                      438
2461            30                      198
2462            40                        0
2463    
2464      In compiling the compiler, there is no significant difference in
2465      compilation speed between maxIter=10 and maxIter=40. Actually,
2466      my measurements showed that maxIter=40 was a tad faster than
2467      maxIter=10! Also 96% of the  files in the compiler reach a fix
2468      point within 13 iterations, so fixing maxIter at 40, while high,
2469      is okay.
2470    
2471    ----------------------------------------------------------------------
2472    Name: Matthias Blume
2473    Date: 2001/10/31 15:25:00 EST
2474    Tag: blume-20011031-pgraph
2475    Description:
2476    
2477    CKIT:
2478    * Changed the "Function" constructor of type Ast.ctype to carry optional
2479      argument identifiers.
2480    * Changed the return type of TypeUtil.getFunction accordingly.
2481    * Type equality ignores the argument names.
2482    * TypeUtil.composite tries to preserve argument names but gives up quickly
2483      if there is a mismatch.
2484    
2485    installation script:
2486    * attempts to use "curl" if available (unless "wget" is available as well)
2487    
2488    CM:
2489    * has an experimental implementation of "portable graphs" which I will
2490      soon propose as an implementation-independent library format
2491    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
2492    
2493    NLFFI-LIB:
2494    * some cleanup (all cosmetic)
2495    
2496    NLFFIGEN:
2497    * temporarily disabled the mechanism that suppresses ML output for
2498      C definitions whose identifiers start with an underscore character
2499    * generate val bindings for enum constants
2500    * user can request that only one style (light or heavy) is being used;
2501      default is to use both (command-line arguments: -heavy and -light)
2502    * fixed bug in handling of function types involving incomplete pointers
2503    * generate ML entry points that take record arguments (i.e., using
2504      named arguments) for C functions that have a prototype with named
2505      arguments
2506      (see changes to CKIT)
2507    
2508    ----------------------------------------------------------------------
2509    Name: Allen Leung
2510    Date: 2001/10/27 20:34:00 EDT
2511    Tag: leunga-20011027-x86-fast-fp-call
2512    Description:
2513    
2514       Fixed the bug described in blume-20010920-slowfp.
2515    
2516       The fix involves
2517          1. generating FCOPYs in FSTP in ia32-svid
2518          2. marking a CALL with the appropriate annotation
2519    
2520    ----------------------------------------------------------------------
2521    Name: Matthias Blume
2522    Date: 2001/10/16 11:32:00 EDT
2523    Tag: blume-20011016-netbsd
2524    Description:
2525    
2526    Underscore patch from Chris Richards (fixing problem with compiling
2527    runtime system under recent NetBSD).
2528    
2529    ----------------------------------------------------------------------
2530    Name: Allen Leung
2531    Date: 2001/10/12 17:18:32 EDT 2001
2532    Tag: leung-20011012-x86-printflowgraph
2533    Description:
2534    
2535    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
2536    
2537    ----------------------------------------------------------------------
2538    Name: Lal George
2539    Date: 2001/10/11 23:51:34 EDT
2540    Tag: george-20011011-too-many-instrs
2541    Description:
2542    
2543    The representation of a program point never expected to see more
2544    than 65536 instructions in a basic block!
2545    
2546    ----------------------------------------------------------------------
2547    Name: Lal George
2548    Date: 2001/10/09 09:41:37 EDT
2549    Tag: george-20011008-mlrisc-labels
2550    Description:
2551    
2552    Changed the machine description files to support printing of
2553    local and global labels in assembly code, based on host assembler
2554    conventions.
2555    
2556    ----------------------------------------------------------------------
2557    Name: Matthias Blume
2558    Date: 2001/09/25 15:25:00 EDT
2559    Tag: blume-20010925-exninfo
2560    Description:
2561    
2562    I provided a non-hook implementation of exnName (at the toplevel) and
2563    made the "dummy" implementation of exnMessage (at the toplevel) more
2564    useful: if nothing gets "hooked in", then at least you are going to
2565    see the exception name and a message indicating why you don't see more.
2566    
2567    [For the time being, programs that need exnMessage and want to use
2568    ml-build should either use General.exnMessage (strongly recommended) or
2569    refer to structure General at some other point so that CM sees a
2570    static dependency.]
2571    
2572    [Similar remarks go for "print" and "use":  If you want to use their
2573    functionality in stand-alone programs generated by ml-build, then use
2574    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
2575    
2576    ----------------------------------------------------------------------
2577    Name: Matthias Blume
2578    Date: 2001/09/20 17:28:00 EDT
2579    Tag: blume-20010920-slowfp
2580    Description:
2581    
2582    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
2583    
2584    ----------------------------------------------------------------------
2585    Name: Matthias Blume
2586    Date: 2001/09/20 17:20:00 EDT
2587    Tag: blume-20010920-canonicalpaths
2588    Description:
2589    
2590    0. Updated the BOOT file (something that I forgot to do earlier).
2591    
2592    1. Small internal change to CM so that it avoids "/../" in filenames
2593       as much as possible (but only where it is safe).
2594    
2595    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
2596       that arguments that contain delimiters are passed through correctly.
2597       This change also means that all "special" arguments of the form
2598       @SMLxxx... must come first.
2599    
2600    3. Changed install script to put relative anchor names for tool commands
2601       into pathconfig.
2602    
2603    ----------------------------------------------------------------------
2604    Name: Matthias Blume
2605    Date: 2001/09/18 15:35:00 EDT
2606    Tag: blume-20010918-readme11036
2607    Description:
2608    
2609    Added README files.
2610    
2611    ----------------------------------------------------------------------
2612    Name: Matthias Blume
2613    Date: 2001/09/18 11:45:00 EDT
2614    Tag: Release_110_36 (retag)
2615    Description:
2616    
2617    Fixed mistake in config/preloads. Retagged as 110.36.
2618    
2619    ----------------------------------------------------------------------
2620    Name: Matthias Blume
2621    Date: 2001/09/18 09:40:00 EDT
2622    Tag: Release_110_36_orig (tag changed)
2623    Description:
2624    
2625    New version (110.36).  New bootfiles.
2626    
2627    ----------------------------------------------------------------------
2628    Name: Matthias Blume
2629    Date: 2001/09/14 16:15:00 EDT
2630    Tag: blume-20010914-x86fastfp
2631    Description:
2632    
2633    John committed some changes that Allen made, in particular a (hopefully)
2634    correctly working version of the x86-fp module.
2635    
2636    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
2637    flag to "true".  Everything seems to compile to a fixpoint ok, and
2638    "mandelbrot" speeds up by about 15%.
2639    
2640    ----------------------------------------------------------------------
2641    Name: Matthias Blume
2642    Date: 2001/09/13 11:20:00 EDT
2643    Tag: blume-20010913-minimal
2644    Description:
2645    
2646    1. Stefan Monnier's patch to fix a miscompilation problem that
2647       was brought to light by John Reppy's work on Moby.
2648    
2649    2. Implemented a minimal "structure Compiler" that contains just
2650       "version" and "architecture".  The minimal version will be
2651       available when the full version is not.  This is for backward-
2652       compatibility with code that wants to test Compiler.version.
2653    
2654    ----------------------------------------------------------------------
2655    Name: Matthias Blume
2656    Date: 2001/08/28 14:03:00 EDT
2657    Tag: blume-20010828-ml-lex
2658    Description:
2659    
2660    Fix for bug 1581, received from Neophytos Michael.
2661    
2662    ----------------------------------------------------------------------
2663    Name: Matthias Blume
2664    Date: 2001/08/27 11:20:00 EDT
2665    Tag: blume-20010827-readme11035
2666    Description:
2667    
2668    Fleshed out the README file for 110.35.
2669    
2670    ----------------------------------------------------------------------
2671    Name: Matthias Blume
2672    Date: 2001/08/24 17:10:00 EDT
2673    Tag: Release_110_35
2674    Description:
2675    
2676    New version number (110.35).  New bootfiles.
2677    
2678    ----------------------------------------------------------------------
2679    Name: Lal George
2680    Date: 2001/08/24 13:47:18 EDT 2001
2681    Tag: george-20010824-MLRISC-graphs
2682    Description:
2683    
2684     removed clusters from MLRISC completely and replaced with graphs.
2685    
2686    ----------------------------------------------------------------------
2687    Name: Matthias Blume
2688    Date: 2001/08/23 17:50:00 EDT
2689    Tag: blume-20010823-toplevel
2690    Description:
2691    
2692    - some reorganization of the code that implements various kinds of
2693      environments in the compiler (static, dynamic, symbolic, combined)
2694    - re-implemented the EnvRef module so that evalStream works properly
2695      (if the stream contains references to "use", "CM.make", etc.)
2696    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
2697    
2698    ----------------------------------------------------------------------
2699    Name: Matthias Blume
2700    Date: 2001/08/20 15:50 EDT
2701    Tag: blume20010820-slipup
2702    Description:
2703    
2704    I forgot to commit a few files.  Here they are...
2705    
2706    ----------------------------------------------------------------------
2707    Name: Matthias Blume
2708    Date: 2001/08/20 15:35:00 EDT
2709    Tag: blume-20010820-debugprof
2710    Description:
2711    
2712    !!!! NEW BOOTFILES !!!!
2713    
2714    This is another round of reorganizing the compiler sources.  This
2715    time the main goal was to factor out all the "instrumentation"
2716    passes (for profiling and backtracing) into their own library.
2717    The difficulty was to do it in such a way that it does not depend
2718    on elaborate.cm but only on elabdata.cm.
2719    
2720    Therefore there have been further changes to both elaborate.cm and
2721    elabdata.cm -- more "generic" things have been moved from the former
2722    to the latter.  As a result, I was forced to split the assignment
2723    of numbers indicating "primtyc"s into two portions: SML-generic and
2724    SML/NJ-specific.  Since it would have been awkward to maintain,
2725    I bit the bullet and actually _changed_ the mapping between these
2726    numbers and primtycs.  The bottom line of this is that you need
2727    a new set of bin- and bootfiles.
2728    
2729    I have built new bootfiles for all architectures, so doing a fresh
2730    checkout and config/install.sh should be all you need.
2731    
2732    The newly created library's name is
2733    
2734        $smlnj/viscomp/debugprof.cm
2735    
2736    and its sources live under
2737    
2738        src/compiler/DebugProf
2739    
2740    ----------------------------------------------------------------------
2741    Name: Matthias Blume
2742    Date: 2001/08/15 17:15:00 EDT
2743    Tag: blume-20010815-compreorg
2744    Description:
2745    
2746    This is a first cut at reorganizing the CM libraries that make up the
2747    core of the compiler.  The idea is to separate out pieces that could
2748    be used independently by tools, e.g., the parser, the typechecker, etc.
2749    
2750    The current status is a step in this direction, but it is not quite
2751    satisfactory yet.  Expect more changes in the future.
2752    
2753    Here is the current (new) organization...
2754    
2755        What used to be $smlnj/viscomp/core.cm is now divided into
2756        six CM libraries:
2757    
2758             $smlnj/viscomp/basics.cm
2759                           /parser.cm
2760                           /elabdata.cm
2761                           /elaborate.cm
2762                           /execute.cm
2763                           /core.cm
2764    
2765        The CM files for these libraries live under src/system/smlnj/viscomp.
2766        All these libraries are proxy libraries that contain precisely
2767        one CM library component.  Here are the locations of the components
2768        (all within the src/compiler tree):
2769    
2770             Basics/basics.cm
2771             Parse/parser.cm
2772             ElabData/elabdata.cm
2773             Elaborator/elaborate.cm
2774             Execution/execute.cm
2775             core.cm
2776    
2777         [This organization is the same that has been used already
2778         for a while for the architecture-specific parts of the visible
2779         compiler and for the old version of core.cm.]
2780    
2781         As you will notice, many source files have been moved from their
2782         respective original locations to a new home in one of the above
2783         subtrees.
2784    
2785         The division of labor between the new libraries is the following:
2786    
2787             basics.cm:
2788                - Simple, basic definitions that pertain to many (or all) of
2789                  the other libraries.
2790             parser.cm:
2791                - The SML parser, producing output of type Ast.dec.
2792                - The type family for Ast is also defined and exported here.
2793             elabdata.cm:
2794                - The datatypes that describe input and output of the elaborator.
2795                  This includes types, absyn, and static environments.
2796             elaborator.cm:
2797                - The SML/NJ type checker and elaborator.
2798                  This maps an Ast.dec (with a given static environment) to
2799                  an Absyn.dec (with a new static environment).
2800                - This libraries implements certain modules that used to be
2801                  structures as functors (to remove dependencies on FLINT).
2802             execute.cm:
2803                - Everything having to do with executing binary code objects.
2804                - Dynamic environments.
2805             core.cm:
2806                - SML/NJ-specific instantiations of the elaborator and MLRISC.
2807                - Top-level modules.
2808                - FLINT (this should eventually become its own library)
2809    
2810    Notes:
2811    
2812    I am not 100% happy with the way I separated the elaborator (and its
2813    data structures) from FLINT.  Two instances of the same problem:
2814    
2815        1. Data structures contain certain fields that carry FLINT-specific
2816           information.  I hacked around this using exn and the property list
2817           module from smlnj-lib.  But the fact that there are middle-end
2818           specific fields around at all is a bit annoying.
2819    
2820        2. The elaborator calculates certain FLINT-related information.  I tried
2821           to make this as abstract as I could using functorization, but, again,
2822           the fact that the elaborator has to perform calculations on behalf
2823           of the middle-end at all is not nice.
2824    
2825        3. Having to used exn and property lists is unfortunate because it
2826           weakens type checking.  The other alternative (parameterizing
2827           nearly *everything*) is not appealing, though.
2828    
2829    I removed the "rebinding =" warning hack because due to the new organization
2830    it was awkward to maintain it.  As a result, the compiler now issues some of
2831    these warnings when compiling init.cmi during bootstrap compilation. On
2832    the plus side, you also get a warning when you do, for example:
2833       val op = = Int32.+
2834    which was not the case up to now.
2835    
2836    I placed "assign" and "deref" into the _Core structure so that the
2837    code that deals with the "lazy" keyword can find them there.  This
2838    removes the need for having access to the primitive environment
2839    during elaboration.
2840    
2841    ----------------------------------------------------------------------
2842    Name: Matthias Blume
2843    Date: 2001/08/13
2844    Tag: blume-20010813-closures
2845    Description:
2846    
2847    This fix was sent to us by Zhong Shao.  It is supposed to improve the
2848    performance of certain loops by avoiding needless closure allocation.
2849    
2850    ----------------------------------------------------------------------
2851    Name: Lal George
2852    Date: 2001/07/31 10:03:23 EDT 2001
2853    Tag: george-20010731-x86-fmalloc
2854    Description: Fixed bug in x86 calls
2855    
2856        There was a bug where call instructions would mysteriously
2857        vanish. The call instruction had to be one that returned
2858        a floating point value.
2859    
2860    ----------------------------------------------------------------------
2861    Name: Lal George
2862    Date: 2001/07/19 16:36:29 EDT 2001
2863    Tag: george-20010719-simple-cells
2864    Description:
2865    
2866    I have dramatically simplified the interface for CELLS in MLRISC.
2867    
2868    In summary, the cells interface is broken up into three parts:
2869    
2870      1. CellsBasis : CELLS_BASIS
2871    
2872            CellsBasis is a top level structure and common for all
2873            architectures.  it contains the definitions of basic datatypes
2874            and utility  functions over these types.
2875    
2876      2. functor Cells() : CELLS
2877    
2878            Cells generates an interface for CELLS that incorporates the
2879            specific resources on the target architecture, such as the
2880            presence of special register classes, their number and size,
2881            and various useful substructures.
2882    
2883      3. <ARCH>CELLS
2884    
2885            e.g. SparcCells: SPARCCELLS
2886    
2887            <ARCH>CELLS usually contains additional bindings for special
2888            registers  on the architecture, such as:
2889    
2890                    val r0 : cell           (* register zero *)
2891                    val y : cell            (* Y register *)
2892                    val psr : cell          (* processor status register *)
2893                    ...
2894    
2895            The structure returned by applying the Cells functor is opened
2896            in this interface.
2897    
2898    The main implication of all this is that the datatypes for cells is
2899    split between CellsBasis and CELLS -- a fairly simple change for user
2900    code.
2901    
2902    In the old scheme the CELLS interface had a definitional binding of
2903    the form:
2904    
2905            signature CELLS = sig
2906    
2907               structure CellsBasis = CellsBasis
2908    
2909               ...
2910    
2911            end
2912    
2913    With all the sharing constraints that goes on in MLRISC, this old
2914    design  quickly leads to errors such as:
2915    
2916            "structure definition spec inside of sharing ... "
2917    
2918    
2919    and appears to require an unacceptable amount of sharing and where
2920    constraint hackery.
2921    
2922    I think this error message (the interaction of definitional specs and
2923    sharing) requires more explanation on our web page.
2924    
2925    ----------------------------------------------------------------------
2926    Name: Matthias Blume
2927    Date: 2001/07/19 15:00:00 EDT
2928    Tag: blume-20010719-libreorg
2929    Description:
2930    
2931    This update puts together a fairly extensive but straightforward change
2932    to the way the libraries that implement the interactive system are
2933    organized:
2934    
2935       The biggest change is the elimination of structure Compiler.  As a
2936       replacement for this structure, there is now a CM library
2937       (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
2938       that exports all the substructures of the original structure Compiler
2939       directly.  So instead of saying Compiler.Foo.bar one now simply
2940       says Foo.bar.  (The CM libraries actually export a collection of
2941       structures that is richer than the collection of substructures of
2942       structure Compiler.)
2943    
2944       To make the transition smooth, there is a separate library called
2945       $smlnj/compiler/compiler.cm which puts together and exports the
2946       original structure Compiler (or at least something very close to it).
2947    
2948       There are five members of the original structure Compiler
2949       that are not exported directly but which instead became members
2950       of a new structure Backend (described by signature BACKEND).  These are:
2951       structure Profile (: PROFILE), structure Compile (: COMPILE), structure
2952       Interact (: INTERACT), structure Machine (: MACHINE), and val
2953       architecture (: string).
2954    
2955       Structure Compiler.Version has become structure CompilerVersion.
2956    
2957       Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
2958       by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
2959       or x86, respectively.
2960       Each of these exports the same frontend structures that
2961       $smlnj/compiler.cm exports.  But they do not have a structure Backend
2962       and instead export some structure <Arch>Backend where <Arch> is Alpha32,
2963       Hppa, PPC, Sparc, or X86, respectively.
2964    
2965       Library $smlnj/compiler/all.cm exports the union of the exports of
2966       $smlnj/compiler/<arch>.cm
2967    
2968       There are no structures <Arch>Compiler anymore, use
2969       $smlnj/compiler/<arch>.cm instead.
2970    
2971       Library host-compiler-0.cm is gone.  Instead, the internal library
2972       that instantiates CM is now called cm0.cm.  Selection of the host
2973       compiler (backend) is no longer done here but. (Responsibility for it
2974       now lies with $smlnj/compiler/current.cm.  This seems to be more
2975       logical.)
2976    
2977       Many individual files have been moved or renamed.  Some files have
2978       been split into multiple files, and some "dead" files have been deleted.
2979    
2980    Aside from these changes to library organization, there are also changes
2981    to the way the code itself is organized:
2982    
2983       Structure Binfile has been re-implemented in such a way that it no
2984       longer needs any knowledge of the compiler.  It exclusively deals
2985       with the details of binfile layout.  It no longer invokes the
2986       compiler (for the purpose of creating new prospective binfile
2987       content), and it no longer has any knowledge of how to interpret
2988       pickles.
2989    
2990       Structure Compile (: COMPILE) has been stripped down to the bare
2991       essentials of compilation.  It no longer deals with linking/execution.
2992       The interface has been cleaned up considerably.
2993    
2994       Utility routines for dealing with linking and execution have been
2995       moved into their own substructures.
2996    
2997       (The ultimate goal of these changes is to provide a light-weight
2998       binfile loader/linker (at least for, e.g., stable libraries) that
2999       does not require CM or the compiler to be present.)
3000    
3001    CM documentation has been updated to reflect the changes to library
3002    organization.
3003    
3004    ----------------------------------------------------------------------
3005    Name: Matthias Blume
3006    Date: 2001/07/10 17:30:00 EDT
3007    Tag: Release_110_34
3008    Description:
3009    
3010    Minor tweak to 110.34 (re-tagged):
3011    
3012      - README.html file added to CVS repository
3013      - runtime compiles properly under FreeBSD 3.X and 4.X
3014    
3015    ----------------------------------------------------------------------
3016    Name: Matthias Blume
3017    Date: 2001/07/10 17:30:00 EDT
3018    Tag: Release_110_34
3019    Description:
3020    
3021    New version number (110.34). New bootfiles.
3022    
3023    ----------------------------------------------------------------------
3024    Name: Matthias Blume
3025    Date: 2001/07/09 16:00:00 EDT
3026    Tag: blume-20010709-more-varargs
3027    Description:
3028    
3029    I changed the handling of varargs in ml-nlffigen again:
3030    The ellipsis ... will now simply be ignored (with an accompanying warning).
3031    
3032    The immediate effect is that you can actually call a varargs function
3033    from ML -- but you can't actually supply any arguments beyond the ones
3034    specified explicitly.  (For example, you can call printf with its format
3035    string, but you cannot pass additional arguments.)
3036    
3037    This behavior is only marginally more useful than the one before, but
3038    it has the advantage that a function or, more importantly, a function
3039    type never gets dropped on the floor, thus avoiding follow-up problems with
3040    other types that refer to the offending one.
3041    
3042    ----------------------------------------------------------------------
3043    Name: Matthias Blume
3044    Date: 2001/07/09 11:25:00 EDT
3045    Tag: blume-20010709-varargs
3046    Description:
3047    
3048    1. ckit-lib.cm now exports structure Error
3049    2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
3050       with a warning accompanied by a source location.  Moreover, it
3051       merely skips the offending function or type and proceeds with the
3052       rest of its work.u  As a result, one can safely feed C code containing
3053       "..." to ml-nlffigen.
3054    3. There are some internal improvements to CM, providing slightly
3055       more general string substitutions in the tools subsystem.
3056    
3057    ----------------------------------------------------------------------
3058    Name: Matthias Blume
3059    Date: 2001/06/27 15:10:00 EDT
3060    Tag: blume-20010627-concur
3061    Description:
3062    
3063    Fixed a small bug in CM's handling of parallel compilation.
3064    (You could observe the bug by Control-C-interrupting an ordinary
3065    CMB.make or CM.stabilize and then attaching some compile servers.
3066    The result was that all of a sudden the previously interrupted
3067    compilation would continue on its own.  This was because of
3068    an over-optimization: CM did not bother to clean out certain queues
3069    when no servers were attached "anyway", resulting in the contents
3070    of these queues to grab control when new servers did get attached.)
3071    
3072    There is also another minor update to the CM manual.
3073    
3074    ----------------------------------------------------------------------
3075    Name: Matthias Blume
3076    Date: 2001/06/26 16:15:00 EDT
3077    Tag: blume-20010626-cmdoc
3078    Description:
3079    
3080    Minor typo fixed in CM manual (syntax diagram for libraries).
3081    
3082    ----------------------------------------------------------------------
3083    Name: Matthias Blume
3084    Date: 2001/06/25 22:55:00 EDT
3085    Tag: blume-20010625-x86pc
3086    Description:
3087    
3088    Fixed a nasty bug in the X86 assembly code that caused signal
3089    handlers to fail (crash) randomly.
3090    
3091    ----------------------------------------------------------------------
3092    Name: Matthias Blume
3093    Date: 2001/06/25 12:05:00 EDT
3094    Tag: blume-20010625-nlffigen
3095    Description:
3096    
3097    This update fixes a number of minor bugs in ml-nlffigen as reported by
3098    Nick Carter <nbc@andrew.cmu.edu>.
3099    
3100      1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
3101      2. Default names for generated files are now derived from the name of
3102         the C file *without its directory*.  In particular, this causes generated
3103         files to be placed locally even if the C file is in some system directory.
3104      3. Default names for generated signatures and structures are also derived
3105         from the C file name without its directory.  This avoids silly things
3106         like "structure GL/GL".
3107         (Other silly names are still possible because ml-nlffigen does not do
3108          a thorough check of whether generated names are legal ML identifiers.
3109          When in doubt, use command line arguments to force particular names.)
3110    
3111    ----------------------------------------------------------------------
3112    Name: Matthias Blume
3113    Date: 2001/06/21 12:25:00 EDT
3114    Tag: blume-20010621-eXene
3115    Description:
3116    
3117    eXene now compiles and (sort of) works again.
3118    
3119    The library name (for version > 110.33) is $/eXene.cm.
3120    
3121    I also added an new example in src/eXene/examples/nbody.  See the
3122    README file there for details.
3123    
3124    ----------------------------------------------------------------------
3125    Name: Matthias Blume
3126    Date: 2001/06/20 16:40:00 EDT
3127    Tag: blume-20010620-cml
3128    Description:
3129    
3130    CML now compiles and works again.
3131    
3132    Libraries (for version > 110.33):
3133    
3134      $cml/cml.cm            Main CML library.
3135      $cml/basis.cm          CML's version of $/basis.cm.
3136      $cml/cml-internal.cm   Internal helper library.
3137      $cml/core-cml.cm       Internal helper library.
3138      $cml-lib/trace-cml.cm  Tracing facility.
3139      $cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
3140    
3141    The installer (config/install.sh) has been taught how to properly
3142    install this stuff.
3143    
3144    ----------------------------------------------------------------------
3145    Name: Matthias Blume
3146    Date: 2001/06/19 17:55:00 EDT
3147    Tag: blume-20010619-instantiate
3148    Description:
3149    
3150    This un-breaks the fix for bug 1432.
3151    (The bug was originally fixed in 110.9 but I broke it again some
3152    time after that.)
3153    
3154    ----------------------------------------------------------------------
3155  Name: Matthias Blume  Name: Matthias Blume
3156  Date: 2001/06/19 17:25:00 EDT  Date: 2001/06/19 17:25:00 EDT
3157  Tag: blume-20010619-signals  Tag: blume-20010619-signals
# Line 2439  Line 5578 
5578       elaborator). There were a lot of changes during my "linkpath" trials       elaborator). There were a lot of changes during my "linkpath" trials
5579       that could have been reverted to their original state but weren't.       that could have been reverted to their original state but weren't.
5580       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
5581       than what was strictly necessary...  (I _did_ resist the tempation       than what was strictly necessary...  (I _did_ resist the temptation
5582       of doing any "global reformatting" to avoid an untimely death at       of doing any "global reformatting" to avoid an untimely death at
5583       Dave's hands. :)       Dave's hands. :)
5584    

Legend:
Removed from v.842  
changed lines
  Added in v.1356

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