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 1044, Mon Jan 28 21:36:08 2002 UTC revision 1393, Fri Sep 26 17:00:32 2003 UTC
# Line 13  Line 13 
13  Description:  Description:
14    
15  ----------------------------------------------------------------------  ----------------------------------------------------------------------
16    Name: Matthias Blume (blume (at) tti - c (dot) org)
17    Date: 2003/09/26 12:00:00 CDT
18    Tag: blume-20030926-ppautoload
19    Description:
20    
21    I modified the read-eval-print loop so that the autoloader gets
22    invoked whenever the prettyprinter tries to look up a symbol that
23    is not currently defined in the toplevel environment but which
24    appears in CM's autoload registry.  As a result, we see far fewer of
25    those ?.Foo.Bar.xxx names in the prettyprinter's output.
26    
27    In addition to this I tried to clean up some pieces of the Basis
28    implementation (e.g., Socket, Word8Array) in order to prevent other
29    instances of these ?.Foo.Bar.xxx names from being printed.
30    
31    The mechanism that picks names for types still needs some work, though.
32    (Right now it seems that if there is a type A.t which is defined to
33    be B.y, but B is unavailable at toplevel, then A.t gets printed as
34    "?.B.t" although the perhaps more sensible solution would be to use
35    "A.t" in this case.  In other words, the prettyprinter should follow
36    a chain of DEFtycs not farther than there are corresponding toplevel
37    names in the current environment.)
38    
39    ----------------------------------------------------------------------
40    Name: Matthias Blume (blume (at) tti - c (dot) org)
41    Date: 2003/09/24 16:31:00 CDT
42    Tag: blume-20030924-installer
43    Description:
44    
45    Another installer tweak:  All the ML code for the installer is now
46    compiled during CMB.make and put into a little library called
47    $smlnj/installer.cm.  The installation then simply invokes
48    
49       sml -m $smlnj/installer.cm
50    
51    and everything happens automagically.
52    
53    Win32:  ML code senses value of environment variable SMLNJ_HOME.
54    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
55            and BINDIR.
56    
57    The new scheme guarantees that the ML code responsible for the installation
58    is in sync with the APIs of the main system.  Also, the installer is
59    somewhat faster because the installer script is precompiled.
60    
61    ----------------------------------------------------------------------
62    Name: Matthias Blume (blume (at) tti - c (dot) org)
63    Date: 2003/09/24 15:35:00 CDT
64    Tag: blume-20030924-synsock
65    Description:
66    
67    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
68    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
69    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
70    implementations of non-blocking operations from CML's Socket
71    structure.
72    
73    ----------------------------------------------------------------------
74    Name: Matthias Blume (blume (at) tti - c (dot) org)
75    Date: 2003/09/24 15:10:05 CDT
76    Tag: blume-20030924-sockets
77    Description:
78    
79    1. Fixed SOCKET API and implementation to match Basis spec.
80       This required changing the internal representation of sockets to one
81       that remembers (for each socket file descriptor) whether it is currently
82       blocking or non-blocking.  This state is maintained lazily (i.e., a system
83       call is made only if the state actually needs to change).
84    
85    2. OS-specific details of sockets were moved into separate files, thus
86       making it possible to unify the bulk of the socket implementations
87       between Unix and Win32.
88    
89    3. CML's socket API changed accordingly.
90       (Note that we need to remove non-blocking functions from this API
91       since they are redundant in the case of CML!)
92    
93    4. CML's socket implementation now makes use of non-blocking functions
94       provided by Basis, thus removing all OS-dependent code from this part
95       of CML.
96    
97    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
98    
99    ----------------------------------------------------------------------
100    Name: Matthias Blume (blume (at) tti - c (dot) org)
101    Date: 2003/09/22 12:10:00 CDT
102    Tag: blume-20030922-110_43_2
103    Description:
104    
105    Made a new interim version and bootfiles for developer's bootstrapping
106    convenience.
107    
108       110.43.2 -- NEW BOOTFILES
109    
110    ----------------------------------------------------------------------
111    Name: Matthias Blume (blume (at) tti - c (dot) org)
112    Date: 2003/09/19 15:55:00 CDT
113    Tag: blume-20030919-cmdir
114    Description:
115    
116    1. new-install.sh -> install.sh
117    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
118    3. tweaked installer so that another name instead of .cm can be chosen
119       at install time (by setting the CM_DIR_ARC environment variable
120       during installation); once installation is complete, the name is
121       fixed
122    
123    ----------------------------------------------------------------------
124    Name: Matthias Blume (blume (at) tti - c (dot) org)
125    Date: 2003/09/18 16:00:00 CDT
126    Tag: blume-20030918-110_43_1
127    Description:
128    
129    Made a new interim version and bootfiles for developer's bootstrapping
130    convenience.
131    
132       110.43.1 -- NEW BOOTFILES
133    
134    ----------------------------------------------------------------------
135    Name: Matthias Blume (blume (at) tti - c (dot) org)
136    Date: 2003/09/18 15:20:00 CDT
137    Tag: blume-20030918-misc
138    Description:
139    
140    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
141       this seems to be controversial at the moment) and used those in
142       Posix.ProcEnv.times.
143    
144    2. Added Time.{from,to}Nanoseconds to Time.
145    
146    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
147       For example, fromLargeInt never needs to look at more than 3 "big
148       digits" to get its 53 bits of precision.
149    
150    ----------------------------------------------------------------------
151    Name: Matthias Blume (blume (at) tti - c (dot) org)
152    Date: 2003/09/17 16:30:00 CDT
153    Tag: blume-20030917-real32-slices
154    Description:
155    
156    Added an entry to the primitive environment
157    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
158    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
159    
160    Removed some of the "magic" constants in real64.sml and replaced them
161    with code that generates these values from their corresponding
162    integer counterparts.
163    
164    Made all(?) the slice-related changes to the Basis and made everything
165    compile again...
166    
167    ----------------------------------------------------------------------
168    Name: Matthias Blume (blume (at) tti - c (dot) org)
169    Date: 2003/09/15 17:45:00 CDT
170    Tag: blume-20030915-rbase
171    Description:
172    
173    Fixed bug in Real.fromLargeInt.
174    
175    ----------------------------------------------------------------------
176    Name: Matthias Blume (blume (at) tti - c (dot) org)
177    Date: 2003/09/13 18:11:00 CDT
178    Tag: blume-20030913-libinstall
179    Description:
180    
181    Minor bugfix in config/libinstall (set anchor with path to
182    standalone tool after installing it, otherwise libraries that
183    need ml-lex or ml-yacc won't compile the first time the installer
184    runs).
185    
186    ----------------------------------------------------------------------
187    Name: Matthias Blume (blume (at) tti - c (dot) org)
188    Date: 2003/09/12 11:45:00 CDT
189    Tag: blume-20030912-various
190    Description:
191    
192     - fixed bug in Real.toLargeInt
193     - fixed bug in Posix.ProcEnv.times
194     - changed inputLine functions to return an option
195     - minor installer improvements / bugfixes
196     - changed default @SMLalloc parameter for x86/celeron to 64k
197    
198    ----------------------------------------------------------------------
199    Name: Matthias Blume (blume (at) tti - c (dot) org)
200    Date: 2003/09/09 22:00:00 CDT
201    Tag: Release_110_43
202    Description:
203    
204        New working release 110.43.  New bootfiles.
205    
206    ----------------------------------------------------------------------
207    Name: Matthias Blume (blume (at) tti - c (dot) org)
208    Date: 2003/09/09 19:20:00 CDT
209    Tag: blume-20030909-installer
210    Description:
211    
212    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
213    Modified config/install.bat to take advantage of it.  Also modified
214    config/install.sh (and called it config/new-install.sh) to take advantage
215    of it on Unix systems.  (The SML code is (supposed to be) platform-
216    independent.)
217    
218    The installer can now install everything under Win32
219    as well as under *nix as long as it compiles.
220    
221    Other changes:
222    
223      - made CML compile again under Win32
224      - made eXene compile under Win32 (by providing a fake structure UnixSock
225        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
226      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
227        same as type int;  under Win32 it isn't
228      - fixed some slice-related problems in the win32-specific parts of CML
229      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
230        (under Win32, the volume name is case-insensitive, and the
231        OS.Path code compares volume names for equality)
232    
233    ----------------------------------------------------------------------
234    Name: Matthias Blume (blume (at) tti - c (dot) org)
235    Date: 2003/09/08 11:55:00 CDT
236    Tag: blume-20030908-fullpath
237    Description:
238    
239    Made Win32 version of OS.FileSys.fullPath return current directory
240    when given an empty string.  This is what the spec says, and incidentally,
241    CM depends on it.  (CM otherwise goes into an infinite loop in certain
242    cases when presented with the name of a non-existing .cm file.)
243    
244    ----------------------------------------------------------------------
245    Name: Matthias Blume (blume (at) tti - c (dot) org)
246    Date: 2003/09/04 16:30:00 CDT
247    Tag: blume-20030905-slices-etc
248    Description:
249    
250      1. Changed interface to vectors and arrays in Basis to match
251         (draft) Basis spec.
252      2. Added signatures and implementations of slices according to
253         Basis spec.
254      3. Edited source code throughout the system to make it compile again
255         under 1. and 2.  (In some cases code had to be added to have it
256         match the new signatures.)
257      4. MLRISC should be backward-compatible: the copies of the originals
258         of files that needed to change under 3. were retained, the .cm files
259         check the compiler version number and use old versions when
260         appropriate.
261      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
262         dirstream -> string option (in accordance with Basis spec).
263      6. When generating code that counts lines, ml-lex used function
264         CharVector.foldli, taking advantage of its old interface.
265         This has been replaced with the corresponding code from
266         CharVectorSlice.  (html-lex must be re-lexed!)
267      7. BitArray in smlnj-lib/Util has been extended/modified to match the
268         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
269      8. Removed temporary additions (fromInternal, toInternal) from the
270         (now obsolete) IntInf in smlnj-lib/Util.
271      9. Cleaned up structure Byte.
272     10. Added localOffset, scan, and fromString to Date (according to spec).
273         Cleaned/corrected implementation of Date.
274         (Still need to check for correctness; implement better canonicalizeDate.)
275     11. Added "scan" to signature IEEE_REAL.
276     12. Some improvements to IntInf [in particular: efficiency-hack for
277         mod and rem when second operand is 2 (for parity checks).]
278     13. Changed representation of type Time.time, using a single IntInf.int
279         value counting microseconds.  This considerably simplified the
280         implementation of structure Time.  We now support negative time
281         values; scan and fromString handle signs.
282     14. Functor PrimIO now takes two additional arguments (VectorSlice and
283         ArraySlice).
284    
285    ----------------------------------------------------------------------
286    Name: Matthias Blume (blume (at) tti - c (dot) org)
287    Date: 2003/08/28 17:00:00 CDT
288    Tag: blume-20030828-intinf
289    Description:
290    
291    This is a major update which comes with a version number bump
292    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
293    and an implementation of IntInf in the Basis.
294    
295    There are a fairly large number of related changes and updates throughout
296    the system:
297    
298    Basis:
299            - Implemented IntInf.
300            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
301            - Added some missing Real64 operations, most notably Real.toLargeInt.
302            - Added FixedInt as a synonym for Int32.
303    
304    compiler:
305            * Added support for a built-in intinf type.
306              - literals
307              - pattern matching
308              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
309              - overloading on literals and operations
310    
311            This required adding a primitive type intinf, some additional
312            primops, and implementations for several non-trivial intinf
313            operations in Core.  (The intinf type is completely abstract
314            to the compiler; all operations get delegated back to the Core.)
315    
316            * Intinf equality is handled by polyequal.  However, the compiler
317            does not print its usual warning in this case (since polyequal
318            is the right thing to do there).
319    
320            * Improved the organization of structure InlineT.
321    
322            * A word about conversion primops:
323            If conversions involving intinf do not cancel out during
324            CPS contract, then the compiler must insert calls to Core functions.
325            Since all core access must be resolved already during the FLINT
326            translate phase, it would be too late a the time of CPS contract
327            to add new Core calls.  For this reason, conversion primops
328            for intinf carry two arguments:  1. the numeric argument that
329            they are supposed to convert, and 2. the Core function that
330            can help with this conversion if necessary.  If CPS contract
331            eliminates a primop, then the associated Core function becomes
332            dead and goes away.  Intinf conversion primops that do not get
333            eliminated by CPS contract get rewritten into calls of their
334            core functions by a separate, new phase.
335    
336    interactive system:
337            - Control.Print.intinfDepth controls max length of intinf constants
338            being printed. (Analogous to Control.Print.stringDepth.)
339            - Cleanup in printutil and pputil: got rid of unused stuff and
340            duplicates; replaced some of the code with code that makes better
341            use of library functionality.
342    
343    CM:
344            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
345            are no longer silent.
346    
347    CKIT:
348            Fixed mismatched uses of Int32 and LargeInt.  I always decided
349            in favor of LargeInt -- which is now the same as IntInf.
350            CKIT-knowledgable people should check whether this is what's
351            intended and otherwise change things back to using Int32 or
352            FixedInt.
353    
354    Throughout the code:
355            Started using IntInf.int literals and built-in operations
356            (e.g., comparison with 0) where this seems appropriate.
357    
358    
359    ----------------------------------------------------------------------
360    Name: Dave MacQueen (dbm@cs.uchicago.edu)
361    Date: 2003/08/13 11:36:00 CDT
362    Tag: dbm-20030813-mcz-merge1
363    Description:
364    
365    Merging changes from the mcz-branch development branch into trunk.
366    These changes involve replacement of the emulated old prettyprinter
367    interface with direct use of the SML/NJ Lib PP library, and fixing
368    of a couple of bugs (895, 1186) relating to error messages.  A new
369    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
370    has been added.
371    
372    ----------------------------------------------------------------------
373    Name: Matthias Blume (blume (at) tti - c (dot) org)
374    Date: 2003/08/11 15:45:00 CDT
375    Tag: blume-20030811-windows
376    Description:
377    
378    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
379    
380        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
381    
382    This patch restores SML/NJ's ability to run under win32.  There are a
383    number of changes, including fixes for several bugs that had gone
384    unnoticed until now:
385    
386      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
387        trying to run the win32 version from within cygwin.)
388    
389      - There are a number of simple .bat scripts that substitute for their
390        corresponding Unix shell-scripts.  (See below.)
391    
392      - The internals of ml-build have been modified slightly.  The main
393        difference is that instead of calling ".link-sml" (or link-sml.bat)
394        using OS.Process.system, the ML process delegates this task back
395        to the script.  Otherwise problems arise in mixed environments such
396        as Cygwin where scripts look and work like Unix scripts, but
397        where OS.Process.system cannot run them.
398    
399      - In CM, the srcpath pickler used native pathname syntax -- which
400        is incorrect in the case of cross-compilation.  The new pickle format
401        is independent of platform-specific naming conventions.
402    
403      - Path configuration files (such as lib/pathconfig) can now choose
404        between native and standard syntax.  Placing a line of the form
405    
406            standard!
407    
408        into the file causes all subsequent paths to be interpreted using
409        CM standard pathname syntax (= Unix conventions); a line
410    
411            native!
412    
413        switches back to native style.  This was needed so that
414        path config files can be written portably, see src/system/pathconfig.
415    
416      - Runtime system:
417    
418          - win32-filesys.c:  get_file_time and set_file_time now
419            access modification time, not creation time.
420    
421          - I/O code made aware of new array representation.
422    
423          - Bug fixes in X86.prim.masm.
424    
425       - src/system/makeml made aware of win32. (For use under cygwin
426            and other Unix-environments for windows.)
427    
428       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
429         which caused BinIO.inputAll to fail consistently.
430    
431    .bat scripts:
432    
433       Windows .bat scripts assume that SMLNJ_HOME is defined.
434    
435       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
436         applications (sml, ml-yacc, ml-lex).
437       - ml-build.bat: analogous to ml-build.
438       - config\install.bat:  Analogous to config/install.sh.  This requires
439         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
440         (nmake etc. must be on the path, and vcvars32 must have been run.)
441         Moreover, sources for ml-lex and ml-yacc need to exist under src,
442         and the bootfile hierarchy must have been unpacked under
443         sml.boot.x86-win32.
444         The script is very primitive and does a poor job at error checking.
445         It only installs the base system, ml-lex, and ml-yacc.  No other
446         libraries are being installed (i.e., you get only those that
447         are part of the compiler.)
448       - link-sml.bat:  analogous to .link-sml, but not currently used
449    
450    Unrelated bug fixes:
451    
452      - ml-nlffigen now exports structures ST_* corresponding to incomplete
453        types.
454      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
455        otherwise.)
456    
457    ----------------------------------------------------------------------
458    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
459    Date: 2003/06/17
460    Tag: macqueen-20030617-bug895
461    Description:
462    
463    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
464    Tag will be used for new development branch (mcz-branch) for
465    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
466    
467    ----------------------------------------------------------------------
468    Name: Matthias Blume (blume (at) tti - c (dot) org)
469    Date: 2003/05/27 16:55:00 CDT
470    Tag: blume-20030527-polyeq
471    Description:
472    
473    Tried to eliminated most cases of polymorphic equality.
474    
475    ----------------------------------------------------------------------
476    Name: Matthias Blume (blume (at) tti - c (dot) org)
477    Date: 2003/05/21 17:45:00 CDT
478    Tag: blume-20030517-complete
479    Description:
480    
481    Two changes:
482    
483      1. Added a flag for controlling whether non-exhaustive bindings will
484         be treated as errors (default is false).
485      2. Cleaned up the *entire* source tree so that CMB.make goes through
486         without a single non-exhaustive match- or bind warning.
487    
488    ----------------------------------------------------------------------
489    Name: Matthias Blume (blume (at) tti - c (dot) org)
490    Date: 2003/05/17 10:20:00 CDT
491    Tag: blume-20030517-absyn
492    Description:
493    
494    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
495    
496       This mainly affects the quality of error messages.  However, some
497       of the code is now more straightforward than before. (Treatment of
498       the above four constructs in translate.sml is much simpler than
499       the "macro-expansion" that was going on before.  Plus, the mach-
500       compiler no longer gets invoked just to be able to compile an
501       if-expression.)
502    
503    2. The ErrorMsg.Error exception is now caught and absorbed by the
504       interactive loop.
505    
506    ----------------------------------------------------------------------
507    Name: Allen Leung
508    Date: 2003/05/16 13:05:00 CDT
509    Tag: leunga-20030516-cygwin-runtime
510    Description:
511    
512        Ported the runtime system to cygwin, which uses the unix
513    x86-unix bin files.  Missing/buggy features:
514    
515       o getnetbyname, getnetbyaddr: these functions seem to be missing in
516         the Cygwin library.
517       o Ctrl-C handling may be flaky.
518       o Windows system calls and Windows I/O are not supported.
519    
520       A new set of binfiles is located at:
521    
522         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
523    
524       This is only needed for bootstrapping the cygwin version of smlnj.
525       Other x86 versions can use the existing binfiles.
526    
527    ----------------------------------------------------------------------
528    Name: Matthias Blume
529    Date: 2003/04/08 15:42:00 CDT
530    Tag: blume-20030408-listpair
531    Description:
532    
533    1. Added a target 'mlrisc' to installer.
534    
535    2. Added missing elements to structure ListPair.
536    
537    ----------------------------------------------------------------------
538    Name: Allen Leung
539    Date: 2003/01/07 10:40:00 EST
540    Tag: leunga-20030107-int-rem
541    Description:
542    
543    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
544    The arguments to the SUBL instruction were swapped.
545    
546    ----------------------------------------------------------------------
547    Name: Matthias Blume
548    Date: 2002/12/12 16:25:00 EST
549    Tag: blume-20021212-risc-ra
550    Description:
551    
552    Fixed a serious bug in the rewrite code for FP spilling/reloading that
553    sent the RA into an infinite loop when floating point registers get
554    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
555    110.37 and 110.38.)
556    There was another set of potential problems related to the handling of
557    MLRISC annotations (but those did not yet cause real problems, apparently).
558    
559    ----------------------------------------------------------------------
560    Name: Matthias Blume
561    Date: 2002/12/06 22:40:00 EST
562    Tag: blume-20021206-cm-fileid
563    Description:
564    
565    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
566    This fixes the problem of CM getting confused by files that suddenly
567    change their identity (e.g., by getting unlinked and recreated by some
568    text editor such as vi).  There might be a better/cheaper/cleaner way
569    of doing this, but for now this will have to do.
570    
571    ----------------------------------------------------------------------
572    Name: Matthias Blume
573    Date: 2002/10/28 09:50:00 EST
574    Tag: blume-20021028-typecheck
575    Description:
576    
577    Exported structure Typecheck from $smlnj/viscomp/core.cm.
578    
579    ----------------------------------------------------------------------
580    Name: Matthias Blume
581    Date: 2002/10/17 09:10:00 EDT
582    Tag: Release_110_42
583    Description:
584    
585    In good old tradition, there has been a slight hiccup so that we have
586    to patch 110.42 after the fact.   The old release tag has been replaced
587    (see below).
588    
589    The change solves a problem with two competing approaches the
590    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
591    each other's way.
592    
593    This change only affects the runtime system code and the installer script.
594    (No new bootfiles.)
595    
596    ----------------------------------------------------------------------
597    Name: Matthias Blume
598    Date: 2002/10/16 12:00:00 EDT
599    Tag: Release_110_42_removed
600    Description:
601    
602    New working release.  New bootfiles.
603    
604    ----------------------------------------------------------------------
605    Name: Matthias Blume
606    Date: 2002/10/10 13:10:00 EDT
607    Tag: blume-20021010-ppc-divs
608    Description:
609    
610    The mltree operator DIVS must be implemented with an overflow check on
611    the PPC because the hardware indicates divide-by-zero using "overflow" as
612    well.
613    
614    ----------------------------------------------------------------------
615    Name: Matthias Blume
616    Date: 2002/07/23 11:45:00
617    Tag: blume-20020723-smlnj-home
618    Description:
619    
620    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
621    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
622    CM_PATHCONFIG is also set), the path configuration file is assumed
623    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
624    the entire tree to some other place and everything will "just work".
625    
626    (Companion commands such as ml-build and ml-makedepend also sense this
627    variable.)
628    
629    ----------------------------------------------------------------------
630    Name: Matthias Blume
631    Date: 2002/07/12 21:19:00 EDT
632    Tag: blume-20020712-liveness
633    Description:
634    
635    Exported two useful "step" functions from liveness module (MLRISC).
636    
637    ----------------------------------------------------------------------
638    Name: Matthias Blume
639    Date: 2002/07/05 16:00 EDT
640    Tag: Release_110_41
641    Description:
642    
643    New working release.  New bootfiles.
644    
645    ----------------------------------------------------------------------
646    Name: Matthias Blume
647    Date: 2002/07/05 10:25:00 EDT
648    Tag: blume-20020705-btimp
649    Description:
650    
651    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
652    other clients can set up backtracing support.
653    
654    ----------------------------------------------------------------------
655    Name: Matthias Blume
656    Date: 2002/06/25 17:23:00 EDT
657    Tag: blume-20020625-fpmax
658    Description:
659    
660    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
661    case.  (The sense of the isNaN test was reversed -- which made min and
662    max always return their first argument.)
663    
664    ----------------------------------------------------------------------
665    Name: Matthias Blume
666    Date: 2002/06/11
667    Tag: blume-20020611-unixpath
668    Description:
669    
670    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
671    
672    ----------------------------------------------------------------------
673    Name: Matthias Blume
674    Date: 2002/06/10 16:35:00 EDT
675    Tag: blume-20020610-ieeereal
676    Description:
677    
678    I back-ported my implementation of IEEEReal.fromString from the
679    idlbasis-devel branch so that we can test it.
680    
681    Another small change is that ppDec tries to give more information
682    than just "<sig>" in the case of functors.  However, this code is
683    broken in some mysterious way if the functor's body's signature
684    has not been declared by ascription but gets inferred from the
685    implementation.  This needs fixing...
686    
687    ----------------------------------------------------------------------
688    Name: Matthias Blume
689    Date: 2002/05/31
690    Tag: blume-20020531-btrace-mode
691    Description:
692    
693    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
694    the wayside when I switched over to using Controls everywhere.)
695    
696    ----------------------------------------------------------------------
697    Name: Lal George
698    Date: 2002/05/23 12:21:40 EDT
699    Tag: george-20020523-visual-labels
700    Description:
701    
702    Labels are now displayed in the graphical output to make
703    the fall-through and target blocks obvious.
704    
705    ----------------------------------------------------------------------
706    Name: Matthias Blume
707    Date: 2002/05/22 11:03:00 EDT
708    Tag: blume-20020522-shrink
709    Description:
710    
711    John tweaked yesterday's fix for 1131 to handle an out-of-memory
712    situation that comes up when allocating huge arrays.
713    
714    ----------------------------------------------------------------------
715    Name: Matthias Blume
716    Date: 2002/05/21 16:00:00 EDT
717    Tag: Release_110_40
718    Description:
719    
720    New working release (110.40).  New bootfiles.
721    
722    [Also: John Reppy fixed GC bug 1131.]
723    
724    ----------------------------------------------------------------------
725    Name: Matthias Blume
726    Date: 2002/05/21 12:35:00 EDT
727    Tag: blume-20020521-cmdoc
728    Description:
729    
730    CM documentation update.
731    
732    ----------------------------------------------------------------------
733    Name: Matthias Blume
734    Date: 2002/05/21 10:55:00 EDT
735    Tag: blume-20020521-misc
736    Description:
737    
738    - John tweaked runtime to be silent on heap export (except when
739      GC messages are on).
740    - I added a few more things (cross-compiling versions of CMB) to
741      config/preloads (as suggestions).
742    
743    ----------------------------------------------------------------------
744    Name: Matthias Blume
745    Date: 2002/05/20 22:25:00 EDT
746    Tag: blume-20020520-controls
747    Description:
748    
749    - Added ControlUtil structure to control-lib.cm.
750    - Use it throughout.
751    - Used Controls facility to define MLRISC controls (as opposed to
752      registering MLRISC control ref cells with Controls after the fact)
753    - Fixed messed-up controls priorities.
754    
755    * Removed again all the stuff from config/preloads that one wouldn't
756      be able to preload at the time the initial heap image is built.
757      (Many libraries, e.g., CML, do not exist yet at this time.  The
758       only libraries that can be preloaded via config/preloads are those
759       that come bundled with the bootfiles.)
760    
761    ----------------------------------------------------------------------
762    Name: Matthias Blume
763    Date: 2002/05/20 10:59:00 EDT
764    Tag: blume-20020520-preloads
765    Description:
766    
767    Added a lot of commented-out suggestions for things to be included
768    in config/preloads.
769    
770    ----------------------------------------------------------------------
771    Name: Allen Leung
772    Date: 2002/05/18 14:20:00 EDT
773    Tag: leunga-20020518-mdl
774    Description:
775    
776    o Made the mdl tool stuff compile and run again.
777    o I've disabled all the stuff that depends on RTL specifications; they
778      are all badly broken anyway.
779    
780    ----------------------------------------------------------------------
781    Name: Matthias Blume
782    Date: 2002/05/17 16:49:00 EDT
783    Tag: blume-20020517-controls
784    Description:
785    
786    0. John Reppy made several modifications to the SML/NJ library.
787       In particular, there is a shiny new controls-lib.cm.
788    
789    1. Pushed new controls interface through compiler so that everything
790       compiles again.
791    
792    2. Added FormatComb and FORMAT_COMB to the CML version of the
793       SML/NJ library (so that CML compiles again).
794    
795    3. Modified init scripts because XXX_DEFAULT environment variables
796       are no longer with us.  (Boot-time initialization is now done
797       using the same environment variables that are also used for
798       startup-time initialization of controls.)
799    
800    ----------------------------------------------------------------------
801    Name: Lal George
802    Date: 2002/05/15 09:20:10 EDT
803    Tag: george-20020515-pseudo-op-decls
804    Description:
805    
806            All pseudo-ops emitted before the first segment declaration
807    such as TEXT, DATA, and BSS directives are assumed to be global
808    declarations and are emitted first in the assembly file. This is
809    useful in a number of situations where one has pseudo-ops that are not
810    specific to any segment, and also works around the constraint that one
811    cannot have client pseudo-ops in the TEXT segment.
812    
813            Because no segment is associated with these declarations it is
814    an error to allocate any space or objects before the first segment
815    directive and an exception will be raised. However, we cannot make
816    this check for client pseudo-ops.
817    
818            These top level declarations are a field in the CFG graph_info.
819    In theory you can continue to add to this field after the CFG has been
820    built -- provided you know what you are doing;-)
821    
822    ----------------------------------------------------------------------
823    Name: Matthias Blume
824    Date: 2002/05/13 16:40:00 EDT
825    Tag: blume-20020513-pp-etc
826    Description:
827    
828    A few minor bugfixes:
829    
830      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
831        (Bogus printouts for redefined bindings still occur.  Compiler
832        bug should no longer occur now.  We need to redo the prettyprinter
833        from scratch.)
834    
835      - CM pathname printer now also adds escape sequences for ( and )
836    
837      - commend and docu fixes for ml-nlffi
838    
839    ----------------------------------------------------------------------
840    Name: Matthias Blume
841    Date: 2002/05/10 16:40:00 EDT
842    Tag: blume-20020510-erg-textio
843    Description:
844    
845    Applied the following bugfix provided by Emden Gansner:
846    
847        Output is corrupted when outputSubstr is used rather than output.
848    
849        The problem occurs when a substring
850    
851            ss = (s, dataStart, dataLen)
852    
853        where dataStart > 0, fills a stream buffer with avail bytes left.
854        avail bytes of s, starting at index dataStart, are copied into the
855        buffer, the buffer is flushed, and then the remaining dataLen-avail
856        bytes of ss are copied into the beginning of the buffer. Instead of
857        starting this copy at index dataStart+avail in s, the current code
858        starts the copy at index avail.
859    
860        Fix:
861        In text-io-fn.sml, change line 695 from
862             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
863        to
864             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
865    
866    ----------------------------------------------------------------------
867    Name: Matthias Blume
868    Date: 2002/04/12 13:55:00 EDT
869    Tag: blume-20020412-assyntax
870    Description:
871    
872    1. Grabbed newer assyntax.h from the XFree86 project.
873    2. Fiddled with how to compile X86.prim.asm without warnings.
874    3. (Very) Minor cleanup in CM.
875    
876    ----------------------------------------------------------------------
877    Name: Matthias Blume
878    Date: 2002/04/01 (no joke!) 17:07:00 EST
879    Tag: blume-20020401-x86div
880    Description:
881    
882    Added full support for div/mod/rem/quot on the x86, using the machine
883    instruction's two results (without clumsily recomputing the remainder)
884    directly where appropriate.
885    
886    Some more extensive power-of-two support was added to the x86 instruction
887    selector (avoiding expensive divs, mods, and muls where they can be
888    replaced with cheaper shifts and masks).  However, this sort of thing
889    ought to be done earlier, e.g., within the CPS optimizer so that
890    all architectures benefit from it.
891    
892    The compiler compiles to a fixed point, but changes might be somewhat
893    fragile nevertheless.  Please, report any strange things that you might
894    see wrt. div/mod/quot/rem...
895    
896    ----------------------------------------------------------------------
897    Name: Matthias Blume
898    Date: 2002/03/29 17:22:00
899    Tag: blume-20020329-div
900    Description:
901    
902    Fixed my broken div/mod logic.  Unfortunately, this means that the
903    inline code for div/mod now has one more comparison than before.
904    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
905    The problem was with quotient = 0, because that alone does not tell
906    us which way the rounding went.  One then has to look at whether
907    remainder and divisor have the same sign...  :(
908    
909    Anyway, I replaced the bootfiles with fresh ones...
910    
911    ----------------------------------------------------------------------
912    Name: Matthias Blume
913    Date: 2002/03/29 14:10:00 EST
914    Tag: blume-20020329-inlprims
915    Description:
916    
917    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
918    
919    Primops have changed. This means that the bin/boot-file formats have
920    changed as well.
921    
922    To make sure that there is no confusion, I made a new version.
923    
924    
925    CHANGES:
926    
927    * removed REMT from mltree (remainder should never overflow).
928    
929    * added primops to deal with divisions of all flavors to the frontend
930    
931    * handled these primops all the way through so they map to their respective
932      MLRISC support
933    
934    * used these primops in the implementation of Int, Int32, Word, Word32
935    
936    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
937    
938    * parameterized INLMIN, INLMAX, and INLABS by a numkind
939    
940    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
941      floating point
942    
943    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
944      and Real (but Real.abs maps to a separate floating-point-only primop)
945    
946    
947    TODO items:
948    
949    * Hacked Alpha32 instruction selection, disabling the selection of REMx
950      instructions because the machine instruction encoder cannot handle
951      them.  (Hppa, PPC, and Sparc instruction selection did not handle
952      REM in the first place, and REM is supported by the x86 machine coder.)
953    
954    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
955      selection phase.  (The two can be streamlined because the hardware
956      delivers both quotient and remainder at the same time anyway.)
957    
958    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
959      (Currently the behavior is inconsistent both across architectures and
960      wrt. the draft Basis spec.)
961    
962    * Word8 should eventually be handled natively, too.
963    
964    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
965      as if there currently is no execution path that could trigger it in
966      SML/NJ.  (The assumptions underlying functions arith and promotable do not
967      hold for things like multiplication and division.)
968    
969    ----------------------------------------------------------------------
970    Name: Matthias Blume
971    Date: 2002/03/27 16:27:00 EST
972    Tag: blume-20020327-mlrisc-divisions
973    Description:
974    
975    Added support for all four division operations (ML's div, mod, quot,
976    and rem) to MLRISC.  In the course of doing so, I also rationalized
977    the naming (no more annoying switch-around of DIV and QUOT), by
978    parameterizing the operation by div_rounding_mode (which can be either
979    DIV_TO_ZERO or DIV_TO_NEGINF).
980    
981    The generic MLTreeGen functor takes care of compiling all four
982    operations down to only round-to-zero div.
983    
984    Missing pieces:
985    
986      * Doing something smarter than relying on MLTreeGen on architectures
987        like, e.g., the x86 where hardware division delivers both quotient and
988        remainder at the same time.  With this, the implementation of the
989        round-to-neginf operations could be further streamlined.
990    
991      * Remove inlining support for div/mod/rem from the frontend and replace it
992        with primops that get carried through to the backend.  Do this for all
993        int and word types.
994    
995    ----------------------------------------------------------------------
996    Name: Matthias Blume
997    Date: 2002/03/25 17:25:00 EST
998    Tag: blume-20020325-divmod
999    Description:
1000    
1001    I improved (hopefully without breaking them) the implementation of Int.div,
1002    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1003    advantage of the following observations:
1004    
1005      Let  q = x quot y      r = x rem y
1006           d = x div  y      m = x mod y
1007    
1008    where "quot" is the round-to-zero version of integer division that
1009    hardware usually provides.  Then we have:
1010    
1011         r = x - q * y        where neither the * nor the - will overflow
1012         d = if q >= 0 orelse x = q * y then q else q - 1
1013                              where neither the * nor the - will overflow
1014         m = if q >= 0 orelse r = 0 then r else r + y
1015                              where the + will not overflow
1016    
1017    This results in substantial simplification of the generated code.
1018    The following table shows the number of CFG nodes and edges generated
1019    for
1020            fun f (x, y) = x OPER y
1021            (* with OPER \in div, mod, quot, rem *)
1022    
1023    
1024        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1025        --------------------------------------------------------
1026         div |         24 |         39 |         12 |         16
1027         mod |         41 |         71 |         12 |         16
1028        quot |          8 |         10 |          8 |         10
1029         rem |         10 |         14 |          8 |         10
1030    
1031    
1032    ----------------------------------------------------------------------
1033    Name: Matthias Blume
1034    Date: 2002/03/25 22:06:00 EST
1035    Tag: blume-20020325-cprotobug
1036    Description:
1037    
1038    Fixed a bug in cproto (c prototype decoder).
1039    
1040    ----------------------------------------------------------------------
1041    Name: Matthias Blume
1042    Date: 2002/03/25 16:00:00 EST
1043    Tag: blume-20020325-raw-primops
1044    Description:
1045    
1046    I did some cleanup to Allen's new primop code and
1047    replaced yesterday's bootfiles with new ones.
1048    (But they are stored in the same place.)
1049    
1050    ----------------------------------------------------------------------
1051    Name: Matthias Blume
1052    Date: 2002/03/24 22:40:00 EST
1053    Tag: blume-20020324-bootfiles
1054    Description:
1055    
1056    Made the bootfiles that Allen asked for.
1057    
1058    ----------------------------------------------------------------------
1059    Name: Allen Leung
1060    Date: 2002/03/23 15:50:00 EST
1061    Tag: leunga-20020323-flint-cps-rcc-primops
1062    Description:
1063    
1064      1. Changes to FLINT primops:
1065    
1066        (* make a call to a C-function;
1067         * The primop carries C function prototype information and specifies
1068         * which of its (ML-) arguments are floating point. C prototype
1069         * information is for use by the backend, ML information is for
1070         * use by the CPS converter. *)
1071      | RAW_CCALL of { c_proto: CTypes.c_proto,
1072                       ml_args: ccall_type list,
1073                       ml_res_opt: ccall_type option,
1074                       reentrant : bool
1075                     } option
1076       (* Allocate uninitialized storage on the heap.
1077        * The record is meant to hold short-lived C objects, i.e., they
1078        * are not ML pointers.  With the tag, the representation is
1079        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1080        *)
1081      | RAW_RECORD of {tag:bool,sz:int}
1082      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1083    
1084      2.  These CPS primops are now overloaded:
1085    
1086           rawload of {kind:numkind}
1087           rawstore of {kind:numkind}
1088    
1089          The one argument form is:
1090    
1091             rawload {kind} address
1092    
1093          The two argument form is:
1094    
1095             rawload {kind} [ml object, byte-offset]
1096    
1097      3. RAW_CCALL/RCC now takes two extra arguments:
1098    
1099         a. The first is whether the C call is reentrant, i.e., whether
1100            ML state should be saved and restored.
1101         b. The second argument is a string argument specifying the name of
1102            library and the C function.
1103    
1104         These things are currently not handled in the code generator, yet.
1105    
1106      4. In CProto,
1107    
1108         An encoding type of "bool" means "ml object" and is mapped into
1109         C prototype of PTR.  Note that "bool" is different than "string",
1110         even though "string" is also mapped into PTR, because "bool"
1111         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1112    
1113      5. Pickler/unpicker
1114    
1115         Changed to handle RAW_RECORD and newest RAW_CCALL
1116    
1117      6. MLRiscGen,
1118    
1119         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1120         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1121            in the file CodeGen/cpscompile/cps-c-calls.sml
1122    
1123      7. Added the conditional move operator
1124    
1125             condmove of branch
1126    
1127         to cps.  Generation of this is still buggy so it is currently
1128         disabled.
1129    
1130    ----------------------------------------------------------------------
1131    Name: Lal George
1132    Date: 2002/03/22 14:18:25 EST
1133    Tag: george-20020322-cps-branch-prob
1134    Description:
1135    
1136    Implemented the Ball-Larus branch prediction-heuristics, and
1137    incorporated graphical viewers for control flow graphs.
1138    
1139    Ball-Larus Heuristics:
1140    ---------------------
1141    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1142    
1143    By design it uses the Dempster-Shafer theory for combining
1144    probabilities.  For example, in the function:
1145    
1146        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1147    
1148    the ball-larus heuristics predicts that the n=0 is unlikely
1149    (OH-heuristic), and the 'then' branch is unlikely because of the
1150    RH-heuristic -- giving the 'then' branch an even lower combined
1151    probability using the Dempster-Shafer theory.
1152    
1153    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1154    probability of the 'then' branch because of the loop in the else
1155    branch.
1156    
1157    
1158    Graphical Viewing:
1159    ------------------
1160    I merely plugged in Allen's graphical viewers into the compiler. The
1161    additional code is not much. At the top level, saying:
1162    
1163            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1164    
1165    will display the graphical view of the control flow graph just before
1166    back-patching.  daVinci must be in your path for this to work. If
1167    daVinci is not available, then the default viewer can be changed
1168    using:
1169    
1170            Control.MLRISC.getString "viewer"
1171    
1172    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1173    course, these viewers must be in your path.
1174    
1175    The above will display the compilation unit at the level of clusters,
1176    many of which are small, boring, and un-interesting. Also setting:
1177    
1178            Control.MLRISC.getInt "cfg-graphical-view_size"
1179    
1180    will display clusters that are larger than the value set by the above.
1181    
1182    
1183    ----------------------------------------------------------------------
1184    Name: Matthias Blume
1185    Date: 2002/03/21 22:20:00 EST
1186    Tag: blume-20020321-kmp-bugfix
1187    Description:
1188    
1189    Changed the interface to the KMP routine in PreString and fixed
1190    a minor bug in one place where it was used.
1191    
1192    ----------------------------------------------------------------------
1193    Name: Allen Leung
1194    Date: 2002/03/21 20:30:00 EST
1195    Tag: leunga-20020321-cfg
1196    Description:
1197    
1198      Fixed a potential problem in cfg edge splitting.
1199    
1200    ----------------------------------------------------------------------
1201    Name: Allen Leung
1202    Date: 2002/03/21 17:15:00 EST
1203    Tag: leunga-20020321-x86-fp-cfg
1204    Description:
1205    
1206      1. Recoded the buggy parts of x86-fp.
1207    
1208         a. All the block reordering code has been removed.
1209            We now depend on the block placement phases to do this work.
1210    
1211         b. Critical edge splitting code has been simplified and moved into the
1212            CFG modules, as where they belong.
1213    
1214         Both of these were quite buggy and complex.  The code is now much, much
1215         simpler.
1216    
1217      2. X86 backend.
1218    
1219         a. Added instructions for 64-bit support.  Instruction selection for
1220            64-bit has not been committed, however, since that
1221            requires changes to MLTREE which haven't been approved by
1222            Lal and John.
1223    
1224         b. Added support for FUCOMI and FUCOMIP when generating code for
1225            PentiumPro and above.  We only generate these instructions in
1226            the fast-fp mode.
1227    
1228         c. Added cases for JP and JNP in X86FreqProps.
1229    
1230      3. CFG
1231    
1232         CFG now has a bunch of methods for edge splitting and merging.
1233    
1234      4. Machine description.
1235    
1236         John's simplification of MLTREE_BASIS.fcond broke a few machine
1237         description things:
1238    
1239         rtl-build.{sig,sml} and hppa.mdl fixed.
1240    
1241         NOTE: the machine description stuff in the repository is still broken.
1242               Again, I can't put my fixes in because that involves
1243               changes to MLTREE.
1244    
1245    ----------------------------------------------------------------------
1246    Name: Matthias Blume
1247    Date: 2002/03/20 15:55:00 EST
1248    Tag: blume-20020320-kmp
1249    Description:
1250    
1251    Implemented Knuth-Morris-Pratt string matching in PreString and used
1252    it for String.isSubstring, Substring.isSubstring, and
1253    Substring.position.
1254    
1255    (Might need some stress-testing.  Simple examples worked fine.)
1256    
1257    ----------------------------------------------------------------------
1258    Name: Matthias Blume
1259    Date: 2002/03/19 16:37:00 EST
1260    Tag: blume-20020319-witnesses
1261    Description:
1262    
1263    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1264    
1265    This implements a generic mechanism for changing constness qualifiers
1266    anywhere within big C types without resorting to outright "casts".
1267    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1268    modify the constness at the outermost level.)
1269    The implementation of "convert" is based on the idea of "witness"
1270    values -- values that are not used by the operation but whose types
1271    "testify" to their applicability.  On the implementation side, "convert"
1272    is simply a projection (returning its second curried argument).  With
1273    cross-module inlining, it should not result in any machine code being
1274    generated.
1275    
1276    ----------------------------------------------------------------------
1277    Name: Matthias Blume
1278    Date: 2002/03/15 16:40:00 EST
1279    Tag: blume-20020315-basis
1280    Description:
1281    
1282    Provided (preliminary?) implementations for
1283    
1284      {String,Substring}.{concatWith,isSuffix,isSubstring}
1285    
1286    and
1287    
1288      Substring.full
1289    
1290    Those are in the Basis spec but they were missing in SML/NJ.
1291    
1292    ----------------------------------------------------------------------
1293    Name: Matthias Blume
1294    Date: 2002/03/14 21:30:00 EST
1295    Tag: blume-20020314-controls
1296    Description:
1297    
1298    Controls:
1299    ---------
1300    
1301    1. Factored out the recently-added Controls : CONTROLS stuff and put
1302       it into its own library $/controls-lib.cm.  The source tree for
1303       this is under src/smlnj-lib/Controls.
1304    
1305    2. Changed the names of types and functions in this interface, so they
1306       make a bit more "sense":
1307    
1308          module -> registry
1309          'a registry -> 'a group
1310    
1311    3. The interface now deals in ref cells only.  The getter/setter interface
1312       is (mostly) gone.
1313    
1314    4. Added a function that lets one register an already-existing ref cell.
1315    
1316    5. Made the corresponding modifications to the rest of the code so that
1317       everything compiles again.
1318    
1319    6. Changed the implementation of Controls.MLRISC back to something closer
1320       to the original.  In particular, this module (and therefore MLRISC)
1321       does not depend on Controls.  There now is some link-time code in
1322       int-sys.sml that registers the MLRISC controls with the Controls
1323       module.
1324    
1325    CM:
1326    ---
1327    
1328      * One can now specify the lambda-split aggressiveness in init.cmi.
1329    
1330    ----------------------------------------------------------------------
1331    Name: Allen Leung
1332    Date: 2002/03/13 17:30:00 EST
1333    Tag: leunga-20020313-x86-fp-unary
1334    Description:
1335    
1336    Bug fix for:
1337    
1338    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1339    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1340    > - fun f(x,(y,z)) = Real.~ y;
1341    > [autoloading]
1342    > [autoloading done]
1343    >       fchsl   (%eax), 184(%esp)
1344    > Error: MLRisc bug: X86MCEmitter.emitInstr
1345    >
1346    > uncaught exception Error
1347    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1348    
1349    The problem was that the code generator did not generate any fp registers
1350    in this case, and the ra didn't know that it needed to run the X86FP phase to
1351    translate the pseudo fp instruction.   This only happened with unary fp
1352    operators in certain situations.
1353    
1354    ----------------------------------------------------------------------
1355    Name: Matthias Blume
1356    Date: 2002/03/13 14:00:00 EST
1357    Tag: blume-20020313-overload-etc
1358    Description:
1359    
1360    1. Added _overload as a synonym for overload for backward compatibility.
1361       (Control.overloadKW must be true for either version to be accepted.)
1362    
1363    2. Fixed bug in install script that caused more things to be installed
1364       than what was requested in config/targets.
1365    
1366    3. Made CM aware of the (_)overload construct so that autoloading
1367       works.
1368    
1369    ----------------------------------------------------------------------
1370    Name: Matthias Blume
1371    Date: 2002/03/12 22:03:00 EST
1372    Tag: blume-20020312-url
1373    Description:
1374    
1375    Forgot to update BOOT and srcarchiveurl.
1376    
1377    ----------------------------------------------------------------------
1378    Name: Matthias Blume
1379    Date: 2002/03/12 17:30:00 EST
1380    Tag: blume-20020312-version110392
1381    Description:
1382    
1383    Yet another version number bump (because of small changes to the
1384    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1385    
1386    Changes:
1387    
1388      The new pid generation scheme described a few weeks ago was overly
1389      complicated.  I implemented a new mechanism that is simpler and
1390      provides a bit more "stability":  Once CM has seen a compilation
1391      unit, it keeps its identity constant (as long as you do not delete
1392      those crucial CM/GUID/* files).  This means that when you change
1393      an interface, compile, then go back to the old interface, and
1394      compile again, you arrive at the original pid.
1395    
1396      There now also is a mechanism that instructs CM to use the plain
1397      environment hash as a module's pid (effectively making its GUID
1398      the empty string).  For this, "noguid" must be specified as an
1399      option to the .sml file in question within its .cm file.
1400      This is most useful for code that is being generated by tools such
1401      as ml-nlffigen (because during development programmers tend to
1402      erase the tool's entire output directory tree including CM's cached
1403      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1404      revert to the old, broken behavior of SML/NJ, but in specific cases
1405      where there is no danger of interface confusion, its use is ok
1406      (I think).
1407    
1408      ml-nlffigen by default generates "noguid" annotations.  They can be
1409      turned off by specifying -guid in its command line.
1410    
1411    ----------------------------------------------------------------------
1412    Name: Lal George
1413    Date: 2002/03/12 12 14:42:36 EST
1414    Tag: george-20020312-frequency-computation
1415    Description:
1416    
1417    Integrated jump chaining and static block frequency into the
1418    compiler. More details and numbers later.
1419    
1420    ----------------------------------------------------------------------
1421    Name: Lal George
1422    Date: 2002/03/11 11 22:38:53 EST
1423    Tag: george-20020311-jump-chain-elim
1424    Description:
1425    
1426    Tested the jump chain elimination on all architectures (except the
1427    hppa).  This is on by default right now and is profitable for the
1428    alpha and x86, however, it may not be profitable for the sparc and ppc
1429    when compiling the compiler.
1430    
1431    The gc test will typically jump to a label at the end of the cluster,
1432    where there is another jump to an external cluster containing the actual
1433    code to invoke gc. This is to allow factoring of common gc invocation
1434    sequences. That is to say, we generate:
1435    
1436            f:
1437               testgc
1438               ja   L1      % jump if above to L1
1439    
1440            L1:
1441               jmp L2
1442    
1443    
1444    After jump chain elimination the 'ja L1' instructions is converted to
1445    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1446    up being implemented in their long form (if L2 is far away) using:
1447    
1448            jbe     L3      % jump if below or equal to L3
1449            jmp     L2
1450         L3:
1451            ...
1452    
1453    
1454    For large compilation units L2  may be far away.
1455    
1456    
1457    ----------------------------------------------------------------------
1458    Name: Matthias Blume
1459    Date: 2002/03/11 13:30:00 EST
1460    Tag: blume-20020311-mltreeeval
1461    Description:
1462    
1463    A functor parameter was missing.
1464    
1465    ----------------------------------------------------------------------
1466    Name: Allen Leung
1467    Date: 2002/03/11 10:30:00 EST
1468    Tag: leunga-20020311-runtime-string0
1469    Description:
1470    
1471       The representation of the empty string now points to a
1472    legal null terminated C string instead of unit.  It is now possible
1473    to convert an ML string into C string with InlineT.CharVector.getData.
1474    This compiles into one single machine instruction.
1475    
1476    ----------------------------------------------------------------------
1477    Name: Allen Leung
1478    Date: 2002/03/10 23:55:00 EST
1479    Tag: leunga-20020310-x86-call
1480    Description:
1481    
1482       Added machine generation for CALL instruction (relative displacement mode)
1483    
1484    ----------------------------------------------------------------------
1485    Name: Matthias Blume
1486    Date: 2002/03/08 16:05:00
1487    Tag: blume-20020308-entrypoints
1488    Description:
1489    
1490    Version number bumped to 110.39.1.  NEW BOOTFILES!
1491    
1492    Entrypoints: non-zero offset into a code object where execution should begin.
1493    
1494    - Added the notion of an entrypoint to CodeObj.
1495    - Added reading/writing of entrypoint info to Binfile.
1496    - Made runtime system bootloader aware of entrypoints.
1497    - Use the address of the label of the first function given to mlriscGen
1498      as the entrypoint.  This address is currently always 0, but it will
1499      not be 0 once we turn on block placement.
1500    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
1501      with entry points) from mlriscGen.
1502    
1503    ----------------------------------------------------------------------
1504    Name: Allen Leung
1505    Date: 2002/03/07 20:45:00 EST
1506    Tag: leunga-20020307-x86-cmov
1507    Description:
1508    
1509       Bug fixes for CMOVcc on x86.
1510    
1511       1. Added machine code generation for CMOVcc
1512       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1513       3. CMOVcc cannot have an immediate operand as argument.
1514    
1515    ----------------------------------------------------------------------
1516    Name: Matthias Blume
1517    Date: 2002/03/07 16:15:00 EST
1518    Tag: blume-20020307-controls
1519    Description:
1520    
1521    This is a very large but mostly boring patch which makes (almost)
1522    every tuneable compiler knob (i.e., pretty much everything under
1523    Control.* plus a few other things) configurable via both the command
1524    line and environment variables in the style CM did its configuration
1525    until now.
1526    
1527    Try starting sml with '-h' (or, if you are brave, '-H')
1528    
1529    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1530    implements the underlying generic mechanism.
1531    
1532    The interface to some of the existing such facilities has changed somewhat.
1533    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1534    (The getFoo interface is still there for backward-compatibility, but its
1535    use is deprecated.)
1536    
1537    The ml-build script passes -Cxxx=yyy command-line arguments through so
1538    that one can now twiddle the compiler settings when using this "batch"
1539    compiler.
1540    
1541    TODO items:
1542    
1543    We should go through and throw out all controls that are no longer
1544    connected to anything.  Moreover, we should go through and provide
1545    meaningful (and correct!) documentation strings for those controls
1546    that still are connected.
1547    
1548    Currently, multiple calls to Controls.new are accepted (only the first
1549    has any effect).  Eventually we should make sure that every control
1550    is being made (via Controls.new) exactly once.  Future access can then
1551    be done using Controls.acc.
1552    
1553    Finally, it would probably be a good idea to use the getter-setter
1554    interface to controls rather than ref cells.  For the time being, both
1555    styles are provided by the Controls module, but getter-setter pairs are
1556    better if thread-safety is of any concern because they can be wrapped.
1557    
1558    *****************************************
1559    
1560    One bug fix: The function blockPlacement in three of the MLRISC
1561    backpatch files used to be hard-wired to one of two possibilities at
1562    link time (according to the value of the placementFlag).  But (I
1563    think) it should rather sense the flag every time.
1564    
1565    *****************************************
1566    
1567    Other assorted changes (by other people who did not supply a HISTORY entry):
1568    
1569    1. the cross-module inliner now works much better (Monnier)
1570    2. representation of weights, frequencies, and probabilities in MLRISC
1571       changed in preparation of using those for weighted block placement
1572       (Reppy, George)
1573    
1574    ----------------------------------------------------------------------
1575    Name: Lal George
1576    Date: 2002/03/07 14:44:24 EST 2002
1577    Tag: george-20020307-weighted-block-placement
1578    
1579    Tested the weighted block placement optimization on all architectures
1580    (except the hppa) using AMPL to generate the block and edge frequencies.
1581    Changes were required in the machine properties to correctly
1582    categorize trap instructions. There is an MLRISC flag
1583    "weighted-block-placement" that can be used to enable weighted block
1584    placement, but this will be ineffective without block/edge
1585    frequencies (coming soon).
1586    
1587    
1588    ----------------------------------------------------------------------
1589    Name: Lal George
1590    Date: 2002/03/05 17:24:48 EST
1591    Tag: george-20020305-linkage-cluster
1592    
1593    In order to support the block placement optimization, a new cluster
1594    is generated as the very first cluster (called the linkage cluster).
1595    It contains a single jump to the 'real' entry point for the compilation
1596    unit. Block placement has no effect on the linkage cluster itself, but
1597    all the other clusters  have full freedom in the manner in which they
1598    reorder blocks or functions.
1599    
1600    On the x86 the typical linkage code that is generated is:
1601       ----------------------
1602            .align 2
1603       L0:
1604            addl    $L1-L0, 72(%esp)
1605            jmp     L1
1606    
1607    
1608            .align  2
1609       L1:
1610       ----------------------
1611    
1612    72(%esp) is the memory location for the stdlink register. This
1613    must contain the address of the CPS function being called. In the
1614    above example, it contains the address of  L0; before
1615    calling L1 (the real entry point for the compilation unit), it
1616    must contain the address for L1, and hence
1617    
1618            addl $L1-L0, 72(%esp)
1619    
1620    I have tested this on all architectures except the hppa.The increase
1621    in code size is of course negligible
1622    
1623    ----------------------------------------------------------------------
1624    Name: Allen Leung
1625    Date: 2002/03/03 13:20:00 EST
1626    Tag: leunga-20020303-mlrisc-tools
1627    
1628      Added #[ ... ] expressions to mlrisc tools
1629    
1630    ----------------------------------------------------------------------
1631    Name: Matthias Blume
1632    Date: 2002/02/27 12:29:00 EST
1633    Tag: blume-20020227-cdebug
1634    Description:
1635    
1636    - made types in structure C and C_Debug to be equal
1637    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1638    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1639    
1640    ----------------------------------------------------------------------
1641    Name: Matthias Blume
1642    Date: 2002/02/26 12:00:00 EST
1643    Tag: blume-20020226-ffi
1644    Description:
1645    
1646    1. Fixed a minor bug in CM's "noweb" tool:
1647       If numbering is turned off, then truly don't number (i.e., do not
1648       supply the -L option to noweb).  The previous behavior was to supply
1649       -L'' -- which caused noweb to use the "default" line numbering scheme.
1650       Thanks to Chris Richards for pointing this out (and supplying the fix).
1651    
1652    2. Once again, I reworked some aspects of the FFI:
1653    
1654       A. The incomplete/complete type business:
1655    
1656       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
1657         gone!
1658       - ML types representing an incomplete type are now *equal* to
1659         ML types representing their corresponding complete types (just like
1660         in C).  This is still safe because ml-nlffigen will not generate
1661         RTTI for incomplete types, nor will it generate functions that
1662         require access to such RTTI.   But when ML code generated from both
1663         incomplete and complete versions of the C type meet, the ML types
1664         are trivially interoperable.
1665    
1666         NOTE:  These changes restore the full generality of the translation
1667         (which was previously lost when I eliminated functorization)!
1668    
1669       B. Enum types:
1670    
1671       - Structure C now has a type constructor "enum" that is similar to
1672         how the "su" constructor works.  However, "enum" is not a phantom
1673         type because each "T enum" has values (and is isomorphic to
1674         MLRep.Signed.int).
1675       - There are generic access operations for enum objects (using
1676         MLRep.Signed.int).
1677       - ml-nlffigen will generate a structure E_foo for each "enum foo".
1678         * The structure contains the definition of type "mlrep" (the ML-side
1679         representation type of the enum).  Normally, mlrep is the same
1680         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
1681         then mlrep will be defined as a datatype -- thus facilitating
1682         pattern matching on mlrep values.
1683         ("-ec" will be suppressed if there are duplicate values in an
1684          enumeration.)
1685         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
1686         will be generated for each C enum constant xxx.
1687         * Conversion functions m2i and i2m convert between mlrep and
1688         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
1689         * Coversion functions c and ml convert between mlrep and "tag enum".
1690         * Access functions (get/set) fetch and store mlrep values.
1691       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
1692         enumerations are merged into one single enumeration represented by
1693         structure E_'.
1694    
1695    ----------------------------------------------------------------------
1696    Name: Allen Leung
1697    Date: 2002/02/25 04:45:00 EST
1698    Tag: leunga-20020225-cps-spill
1699    
1700    This is a new implementation of the CPS spill phase.
1701    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
1702    In case of problems, replace it with the old file spill.sml
1703    
1704    The current compiler runs into some serious performance problems when
1705    constructing a large record.  This can happen when we try to compile a
1706    structure with many items.  Even a very simple structure like the following
1707    makes the compiler slow down.
1708    
1709        structure Foo = struct
1710           val x_1 = 0w1 : Word32.int
1711           val x_2 = 0w2 : Word32.int
1712           val x_3 = 0w3 : Word32.int
1713           ...
1714           val x_N = 0wN : Word32.int
1715        end
1716    
1717    The following table shows the compile time, from N=1000 to N=4000,
1718    with the old compiler:
1719    
1720    N
1721    1000   CPS 100 spill                           0.04u  0.00s  0.00g
1722           MLRISC ra                               0.06u  0.00s  0.05g
1723              (spills = 0 reloads = 0)
1724           TOTAL                                   0.63u  0.07s  0.21g
1725    
1726    1100   CPS 100 spill                           8.25u  0.32s  0.64g
1727           MLRISC ra                               5.68u  0.59s  3.93g
1728              (spills = 0 reloads = 0)
1729           TOTAL                                   14.71u  0.99s  4.81g
1730    
1731    1500   CPS 100 spill                           58.55u  2.34s  1.74g
1732           MLRISC ra                               5.54u  0.65s  3.91g
1733              (spills = 543 reloads = 1082)
1734           TOTAL                                   65.40u  3.13s  6.00g
1735    
1736    2000   CPS 100 spill                           126.69u  4.84s  3.08g
1737           MLRISC ra                               0.80u  0.10s  0.55g
1738              (spills = 42 reloads = 84)
1739           TOTAL                                   129.42u  5.10s  4.13g
1740    
1741    3000   CPS 100 spill                           675.59u  19.03s  11.64g
1742           MLRISC ra                               2.69u  0.27s  1.38g
1743              (spills = 62 reloads = 124)
1744           TOTAL                                   682.48u  19.61s  13.99g
1745    
1746    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
1747           MLRISC ra                               4.96u  0.27s  2.72g
1748              (spills = 85 reloads = 170)
1749           TOTAL                                   2375.26u  57.21s  48.00g
1750    
1751    As you can see the old cps spill module suffers from some serious
1752    performance problem.  But since I cannot decipher the old code fully,
1753    instead of patching the problems up, I'm reimplementing it
1754    with a different algorithm.  The new code is more modular,
1755    smaller when compiled, and substantially faster
1756    (O(n log n) time and O(n) space).  Timing of the new spill module:
1757    
1758    4000  CPS 100 spill                           0.02u  0.00s  0.00g
1759          MLRISC ra                               0.25u  0.02s  0.15g
1760             (spills=1 reloads=3)
1761          TOTAL                                   7.74u  0.34s  1.62g
1762    
1763    Implementation details:
1764    
1765    As far as I can tell, the purpose of the CPS spill module is to make sure the
1766    number of live variables at any program point (the bandwidth)
1767    does not exceed a certain limit, which is determined by the
1768    size of the spill area.
1769    
1770    When the bandwidth is too large, we decrease the register pressure by
1771    packing live variables into spill records.  How we achieve this is
1772    completely different than what we did in the old code.
1773    
1774    First, there is something about the MLRiscGen code generator
1775    that we should be aware of:
1776    
1777    o MLRiscGen performs code motion!
1778    
1779       In particular, it will move floating point computations and
1780       address computations involving only the heap pointer to
1781       their use sites (if there is only a single use).
1782       What this means is that if we have a CPS record construction
1783       statement
1784    
1785           RECORD(k,vl,w,e)
1786    
1787       we should never count the new record address w as live if w
1788       has only one use (which is often the case).
1789    
1790       We should do something similar to floating point, but the transformation
1791       there is much more complex, so I won't deal with that.
1792    
1793    Secondly, there are now two new cps primops at our disposal:
1794    
1795     1. rawrecord of record_kind option
1796        This pure operator allocates some uninitialized storage from the heap.
1797        There are two forms:
1798    
1799         rawrecord NONE [INT n]  allocates a tagless record of length n
1800         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
1801                                     and initializes the tag.
1802    
1803     2. rawupdate of cty
1804          rawupdate cty (v,i,x)
1805          Assigns to x to the ith component of record v.
1806          The storelist is not updated.
1807    
1808    We use these new primops for both spilling and increment record construction.
1809    
1810     1. Spilling.
1811    
1812        This is implemented with a linear scan algorithm (but generalized
1813        to trees).  The algorithm will create a single spill record at the
1814        beginning of the cps function and use rawupdate to spill to it,
1815        and SELECT or SELp to reload from it.  So both spills and reloads
1816        are fine-grain operations.  In contrast, in the old algorithm
1817        "spills" have to be bundled together in records.
1818    
1819        Ideally, we should sink the spill record construction to where
1820        it is needed.  We can even split the spill record into multiple ones
1821        at the places where they are needed.  But CPS is not a good
1822        representation for global code motion, so I'll keep it simple and
1823        am not attempting this.
1824    
1825     2. Incremental record construction (aka record splitting).
1826    
1827        Long records with many component values which are simulatenously live
1828        (recall that single use record addresses are not considered to
1829         be live) are constructed with rawrecord and rawupdate.
1830        We allocate space on the heap with rawrecord first, then gradually
1831        fill it in with rawupdate.  This is the technique suggested to me
1832        by Matthias.
1833    
1834        Some restrictions on when this is applicable:
1835        1. It is not a VECTOR record.  The code generator currently does not handle
1836           this case. VECTOR record uses double indirection like arrays.
1837        2. All the record component values are defined in the same "basic block"
1838           as the record constructor.  This is to prevent speculative
1839           record construction.
1840    
1841    ----------------------------------------------------------------------
1842    Name: Allen Leung
1843    Date: 2002/02/22 01:02:00 EST
1844    Tag: leunga-20020222-mlrisc-tools
1845    
1846    Minor bug fixes in the parser and rewriter
1847    
1848    ----------------------------------------------------------------------
1849    Name: Allen Leung
1850    Date: 2002/02/21 20:20:00 EST
1851    Tag: leunga-20020221-peephole
1852    
1853    Regenerated the peephole files.  Some contained typos in the specification
1854    and some didn't compile because of pretty printing bugs in the old version
1855    of 'nowhere'.
1856    
1857    ----------------------------------------------------------------------
1858    Name: Allen Leung
1859    Date: 2002/02/19 20:20:00 EST
1860    Tag: leunga-20020219-mlrisc-tools
1861    Description:
1862    
1863       Minor bug fixes to the mlrisc-tools library:
1864    
1865       1.  Fixed up parsing colon suffixed keywords
1866       2.  Added the ability to shut the error messages up
1867       3.  Reimplemented the pretty printer and fixed up/improved
1868           the pretty printing of handle and -> types.
1869       4.  Fixed up generation of literal symbols in the nowhere tool.
1870       5.  Added some SML keywords to to sml.sty
1871    
1872    ----------------------------------------------------------------------
1873    Name: Matthias Blume
1874    Date: 2002/02/19 16:20:00 EST
1875    Tag: blume-20020219-cmffi
1876    Description:
1877    
1878    A wild mix of changes, some minor, some major:
1879    
1880    * All C FFI-related libraries are now anchored under $c:
1881        $/c.cm      --> $c/c.cm
1882        $/c-int.cm  --> $c/internals/c-int.cm
1883        $/memory.cm --> $c/memory/memory.cm
1884    
1885    * "make" tool (in CM) now treats its argument pathname slightly
1886      differently:
1887        1. If the native expansion is an absolute name, then before invoking
1888           the "make" command on it, CM will apply OS.Path.mkRelative
1889           (with relativeTo = OS.FileSys.getDir()) to it.
1890        2. The argument will be passed through to subsequent phases of CM
1891           processing without "going native".  In particular, if the argument
1892           was an anchored path, then "make" will not lose track of that anchor.
1893    
1894    * Compiler backends now "know" their respective C calling conventions
1895      instead of having to be told about it by ml-nlffigen.  This relieves
1896      ml-nlffigen from one of its burdens.
1897    
1898    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
1899    
1900    * Export C_DEBUG and C_Debug from $c/c.cm.
1901    
1902    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
1903      subtyping relationship between incomplete pointers and their complete
1904      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
1905      with the convention of instantiating 'o with ('t, 'c) obj whenever
1906      the pointer target type is complete.  In the incomplete case, 'o
1907      will be instantiated with some "'c iobj" -- a type obtained by
1908      using one of the functors PointerToIncompleteType or PointerToCompleteType.
1909    
1910      Operations that work on both incomplete and complete pointer types are
1911      typed as taking an 'o ptr while operations that require the target to
1912      be known are typed as taking some ('t, 'c) obj ptr.
1913    
1914      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
1915      where void is an eqtype without any values.  This makes it possible
1916      to work on voidptr values using functions meant to operate on light
1917      incomplete pointers.
1918    
1919    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
1920      been vastly simplified.
1921    
1922    ----------------------------------------------------------------------
1923    Name: Matthias Blume
1924    Date: 2002/02/19 10:48:00 EST
1925    Tag: blume-20020219-pqfix
1926    Description:
1927    
1928    Applied Chris Okasaki's bug fix for priority queues.
1929    
1930    ----------------------------------------------------------------------
1931    Name: Matthias Blume
1932    Date: 2002/02/15 17:05:00
1933    Tag: Release_110_39
1934    Description:
1935    
1936    Last-minute retagging is becoming a tradition... :-(
1937    
1938    This is the working release 110.39.
1939    
1940    ----------------------------------------------------------------------
1941    Name: Matthias Blume
1942    Date: 2002/02/15 16:00:00 EST
1943    Tag: Release_110_39-orig
1944    Description:
1945    
1946    Working release 110.39.  New bootfiles.
1947    
1948    (Update: There was a small bug in the installer so it wouldn't work
1949    with all shells.  So I retagged. -Matthias)
1950    
1951    ----------------------------------------------------------------------
1952    Name: Matthias Blume
1953    Date: 2002/02/15 14:17:00 EST
1954    Tag: blume-20020215-showbindings
1955    Description:
1956    
1957    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
1958    the latter can be useful for exploring what bindings are available at
1959    the interactive prompt.  (The first function returns only the list
1960    of symbols that are really bound, the second prints those but also the
1961    ones that CM's autoloading mechanism knows about.)
1962    
1963    ----------------------------------------------------------------------
1964    Name: Matthias Blume
1965    Date: 2002/02/15 12:08:00 EST
1966    Tag: blume-20020215-iptrs
1967    Description:
1968    
1969    Two improvements to ml-nlffigen:
1970    
1971      1. Write files only if they do not exist or if their current contents
1972         do not coincide with what's being written.  (That is, avoid messing
1973         with the time stamps unless absolutely necessary.)
1974    
1975      2. Implement a "repository" mechanism for generated files related
1976         to "incomplete pointer types".   See the README file for details.
1977    
1978    ----------------------------------------------------------------------
1979    Name: Matthias Blume
1980    Date: 2002/02/14 11:50:00 EST
1981    Tag: blume-20020214-quote
1982    Description:
1983    
1984    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
1985    because of the new and improved tag generation scheme.  (Thanks to Allen
1986    Leung for pointing it out.)
1987    
1988    ----------------------------------------------------------------------
1989    Name: Lal George
1990    Date: 2002/02/14 09:55:27 EST 2002
1991    Tag: george-20020214-isabelle-bug
1992    Description:
1993    
1994    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
1995    of Isabelle on the x86.
1996    
1997    From Allen:
1998    -----------
1999     I've found the problem:
2000    
2001         in ra-core.sml, I use the counter "blocked" to keep track of the
2002         true number of elements in the freeze queue.  When the counter goes
2003         to zero, I skip examining the queue.  But I've messed up the
2004         bookkeeping in combine():
2005    
2006             else ();
2007             case !ucol of
2008               PSEUDO => (if !cntv > 0 then
2009                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2010                                        ^^^^^^^^^^^^^^^^^^^^^^^
2011                      moveu := mergeMoveList(!movev, !moveu)
2012                     )
2013                  else ();
2014    
2015         combine() is called to coalesce two nodes u and v.
2016         I think I was thinking that if the move counts of u and v are both
2017         greater than zero then after they are coalesced then one node is
2018         removed from the freeze queue.  Apparently I was thinking that
2019         both u and v are of low degree, but that's clearly not necessarily true.
2020    
2021    
2022    02/12/2002:
2023        Here's the patch.  HOL now compiles.
2024    
2025        I don't know how this impact on performance (compile
2026        time or runtime).  This bug caused the RA (especially on the x86)
2027        to go thru the potential spill phase when there are still nodes on the
2028        freeze queue.
2029    
2030    
2031    
2032    
2033    ----------------------------------------------------------------------
2034    Name: Matthias Blume
2035    Date: 2002/02/13 22:40:00 EST
2036    Tag: blume-20020213-fptr-rtti
2037    Description:
2038    
2039    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2040    updates.
2041    
2042    ----------------------------------------------------------------------
2043    Name: Matthias Blume
2044    Date: 2002/02/13 16:41:00 EST
2045    Tag: blume-20020213-cmlpq
2046    Description:
2047    
2048    Added new priority queue export symbols (which have just been added to
2049    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2050    not compile and the installer would choke.)
2051    
2052    ----------------------------------------------------------------------
2053    Name: Matthias Blume
2054    Date: 2002/02/13 16:15:00 EST
2055    Tag: blume-20020213-various
2056    Description:
2057    
2058    1. More tweaks to ml-nlffigen:
2059    
2060       - better internal datastructures (resulting in slight speedup)
2061       - "-match" option requires exact match
2062       - "localized" gensym counters (untagged structs/unions nested within
2063         other structs/unions or within typedefs get a fresh counter; their
2064         tag will be prefixed by a concatenation of their parents' tags)
2065       - bug fixes (related to calculation of transitive closure of types
2066         to be included in the output)
2067    
2068    2. Minor Basis updates:
2069    
2070       - added implementations for List.collate and Option.app
2071    
2072    ----------------------------------------------------------------------
2073    Name: Matthias Blume
2074    Date: 2002/02/11 15:55:00 EST
2075    Tag: blume-20020211-gensym
2076    Description:
2077    
2078    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2079    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2080    names (ML structure names that correspond to unnamed C structs, unions,
2081    and enums), so that separate runs of ml-nlffigen do not clash.
2082    
2083    ----------------------------------------------------------------------
2084    Name: Matthias Blume
2085    Date: 2002/02/11 12:05:00 EST
2086    Tag: blume-20020211-gensml
2087    Description:
2088    
2089    A quick fix for a problem with GenSML (in the pgraph-util library):
2090    Make generation of toplevel "local" optional.  (Strictly speaking,
2091    signature definitions within "local" are not legal SML.)
2092    
2093    Other than that: updates to INSTALL and cm/TODO.
2094    
2095    ----------------------------------------------------------------------
2096    Name: Matthias Blume
2097    Date: 2002/02/08 15:00:00 EST
2098    Tag: blume-20020208-uniquepid
2099    Description:
2100    
2101    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2102    
2103    1. The installer (config/install.sh) has gotten smarter:
2104    
2105         - Configuration options are a bit easier to specify now
2106           (in config/targets).
2107         - Bug in recognizing .tar.bz2 files fixed.
2108         - Installer automatically resolves dependencies between
2109           configuration options (e.g., if you ask for eXene, you will
2110           also get cml -- regardless whether you asked for it or not).
2111         - Installer can run in "quieter mode" by setting the environment
2112           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2113           "completely silent", though.
2114         - Build HashCons library as part of smlnj-lib.
2115    
2116    2. A new scheme for assigning persistent identifiers to compilation
2117       units (and, by extension, to types etc.) has been put into place.
2118       This fixes a long-standing bug where types and even dynamic values
2119       can get internally confused, thereby compromising type safety
2120       (abstraction) and dynamic correctness.  See
2121    
2122         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2123    
2124       for an example of how things could go wrong until now.
2125    
2126       The downside of the new scheme is that pids are not quite as
2127       persistent as they used to be: CM will generate a fresh pid
2128       for every compilation unit that it thinks it sees for the first
2129       time.  That means that if you compile starting from a clean, fresh
2130       source tree at two different times, you end up with different
2131       binaries.
2132    
2133       Cutoff recompilation, however, has not been compromised because
2134       CM keeps pid information in special caches between runs.
2135    
2136    ----------------------------------------------------------------------
2137    Name: Lal George
2138    Date: 2002/02/07 15:34:13 EST 2002
2139    Tag: <none>
2140    Description:
2141    
2142    Compilers that generate assembly code may produce  global labels
2143    whose value is resolved at link time. The various peephole optimization
2144    modules did not take this in account.
2145    
2146    TODO. The Labels.addrOf function should really return an option
2147    type so that clients are forced to deal with this issue, rather
2148    than an exception being raised.
2149    
2150    ----------------------------------------------------------------------
2151    Name: Lal George
2152    Date: 2002/02/06 13:55:02 EST
2153    Tag: george-20020206-ra-breakup
2154    Description:
2155    
2156    1. A bug fix from Allen.
2157    
2158        A typo causes extra fstp %st(0)'s to be generated at compensation
2159        edges, which might cause stack underflow traps at runtime.  This
2160        occurs in fft where there are extraneous fstps right before the 'into'
2161        trap instruction (in this case they are harmless since none of the
2162        integers overflow.)
2163    
2164    2. Pulled out various utility modules that were embedded in the modules
2165       of the register allocator. I need these modules for other purposes, but
2166       they are not complete enough to put into a library (just yet).
2167    ----------------------------------------------------------------------
2168    Name: Matthias Blume
2169    Date: 2002/01/31 16:05:00 EST
2170    Tag: blume-20020131-sparc-ccalls
2171    Description:
2172    
2173    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2174       of extra stack space by mistake.  Fixed.
2175    
2176    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2177    
2178    ----------------------------------------------------------------------
2179    Name: Allen Leung
2180    Date: 2002/01/30
2181    Tag: leunga-20020130-nowhere-bug-fix
2182    Description:
2183    
2184       MLRISC bug fixes:
2185       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2186       2. MachineInt.fromString was negating its value.
2187    
2188    ----------------------------------------------------------------------
2189    Name: Matthias Blume
2190    Date: 2002/01/29
2191    Tag: blume-20020129-INSTALL
2192    Description:
2193    
2194    - Added somewhat detailed installation instructions (file INSTALL).
2195    - Fixed curl-detection bug in config/install.sh.
2196    - It is now possible to select the URL getter using the URLGETTER
2197      environment variable:
2198    
2199          not set / "unknown"      --> automatic detection (script tries wget,
2200                                       curl, and lynx)
2201          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2202                                       how to properly invoke them)
2203          other                    --> use $URLGETTER directly, it must take
2204                                       precisely two command-line arguments
2205                                       (source URL and destination file name)
2206    
2207    ----------------------------------------------------------------------
2208    Name: Matthias Blume
2209    Date: 2002/01/28
2210    Tag: blume-20020128-sparc-ccalls
2211    Description:
2212    
2213    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2214    - Make use of the allocParam argument in sparc-c-calls.
2215    
2216    ----------------------------------------------------------------------
2217  Name: Matthias Blume  Name: Matthias Blume
2218  Date: 2002/01/28  Date: 2002/01/28
2219  Tag: blume-20020128-allocParam  Tag: blume-20020128-allocParam
# Line 584  Line 2785 
2785    
2786  ----------------------------------------------------------------------  ----------------------------------------------------------------------
2787  Name: Matthias Blume  Name: Matthias Blume
 >>>>>>> 1.169  
2788  Date: 2001/09/18 15:35:00 EDT  Date: 2001/09/18 15:35:00 EDT
2789  Tag: blume-20010918-readme11036  Tag: blume-20010918-readme11036
2790  Description:  Description:

Legend:
Removed from v.1044  
changed lines
  Added in v.1393

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