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 592, Mon Apr 3 07:04:12 2000 UTC revision 1414, Thu Nov 6 17:54:56 2003 UTC
# Line 8  Line 8 
8  The form of an entry should be:  The form of an entry should be:
9    
10  Name:  Name:
11  Date:  Date: yyyy/mm/dd
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14    
15    ----------------------------------------------------------------------
16    Name: Matthias Blume (blume (at) tti - c (dot) org)
17    Date: 2003/11/06 12:00:00 CST
18    Tag: Release_110_44
19    Description:
20    
21    New working version (110.44). New bootfiles.
22    
23    ----------------------------------------------------------------------
24    Name: Matthias Blume (blume (at) tti - c (dot) org)
25    Date: 2003/11/04 11:50:00 CST
26    Tag: blume-20031104-move-libraries
27    Description:
28    
29    Eliminated the "dont_move_libraries" directive in config/targets.
30    (The mechanism was broken and could not be fixed easily.  Moreover,
31    there does not seem to be any reason not to move all libraries into
32    lib during installation.  I originally implemented this directive as a
33    backward-compatibility feature when I first introduced the new CM.
34    Now that things have been stable for a long time and going back to the
35    old CM is not an option, there is no reason to keep it around.)
36    
37    ----------------------------------------------------------------------
38    Name: Matthias Blume (blume (at) tti - c (dot) org)
39    Date: 2003/11/03 16:00:00 CST
40    Tag: blume-20031103-installdir
41    Description:
42    
43    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
44    Richards for pointing out the problem and providing the solution.)
45    
46    ----------------------------------------------------------------------
47    Name: Matthias Blume (blume (at) tti - c (dot) org)
48    Date: 2003/10/01 17:05:00 CDT
49    Tag: blume-20031001-lal-mlrisc
50    Description:
51    
52    MLRISC bug fix from Lal.
53    
54    ----------------------------------------------------------------------
55    Name: Matthias Blume (blume (at) tti - c (dot) org)
56    Date: 2003/09/30 16:10:00 CDT
57    Tag: blume-20030930-primio-bat
58    Description:
59    
60    1. Added openVector, nullRd, and nullWr to PRIM_IO.
61    2. Improved .bat files (for Win32 port) to make things work under Win95.
62       (thanks to Aaron S. Hawley for this one)
63    
64    ----------------------------------------------------------------------
65    Name: Matthias Blume (blume (at) tti - c (dot) org)
66    Date: 2003/09/26 16:05:00 CDT
67    Tag: blume-20030926-wrappriv
68    Description:
69    
70    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
71    
72    ----------------------------------------------------------------------
73    Name: Matthias Blume (blume (at) tti - c (dot) org)
74    Date: 2003/09/26 15:00:00 CDT
75    Tag: blume-20030926-110_43_3
76    Description:
77    
78    - additional cleanup
79    - version number bump, NEW BOOTFILES
80    
81    ----------------------------------------------------------------------
82    Name: Matthias Blume (blume (at) tti - c (dot) org)
83    Date: 2003/09/26 12:00:00 CDT
84    Tag: blume-20030926-ppautoload
85    Description:
86    
87    I modified the read-eval-print loop so that the autoloader gets
88    invoked whenever the prettyprinter tries to look up a symbol that
89    is not currently defined in the toplevel environment but which
90    appears in CM's autoload registry.  As a result, we see far fewer of
91    those ?.Foo.Bar.xxx names in the prettyprinter's output.
92    
93    In addition to this I tried to clean up some pieces of the Basis
94    implementation (e.g., Socket, Word8Array) in order to prevent other
95    instances of these ?.Foo.Bar.xxx names from being printed.
96    
97    The mechanism that picks names for types still needs some work, though.
98    (Right now it seems that if there is a type A.t which is defined to
99    be B.u, but B is unavailable at toplevel, then A.t gets printed as
100    "?.B.u" although the perhaps more sensible solution would be to use
101    "A.t" in this case.  In other words, the prettyprinter should follow
102    a chain of DEFtycs not farther than there are corresponding toplevel
103    names in the current environment.)
104    
105    ----------------------------------------------------------------------
106    Name: Matthias Blume (blume (at) tti - c (dot) org)
107    Date: 2003/09/24 16:31:00 CDT
108    Tag: blume-20030924-installer
109    Description:
110    
111    Another installer tweak:  All the ML code for the installer is now
112    compiled during CMB.make and put into a little library called
113    $smlnj/installer.cm.  The installation then simply invokes
114    
115       sml -m $smlnj/installer.cm
116    
117    and everything happens automagically.
118    
119    Win32:  ML code senses value of environment variable SMLNJ_HOME.
120    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
121            and BINDIR.
122    
123    The new scheme guarantees that the ML code responsible for the installation
124    is in sync with the APIs of the main system.  Also, the installer is
125    somewhat faster because the installer script is precompiled.
126    
127    ----------------------------------------------------------------------
128    Name: Matthias Blume (blume (at) tti - c (dot) org)
129    Date: 2003/09/24 15:35:00 CDT
130    Tag: blume-20030924-synsock
131    Description:
132    
133    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
134    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
135    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
136    implementations of non-blocking operations from CML's Socket
137    structure.
138    
139    ----------------------------------------------------------------------
140    Name: Matthias Blume (blume (at) tti - c (dot) org)
141    Date: 2003/09/24 15:10:05 CDT
142    Tag: blume-20030924-sockets
143    Description:
144    
145    1. Fixed SOCKET API and implementation to match Basis spec.
146       This required changing the internal representation of sockets to one
147       that remembers (for each socket file descriptor) whether it is currently
148       blocking or non-blocking.  This state is maintained lazily (i.e., a system
149       call is made only if the state actually needs to change).
150    
151    2. OS-specific details of sockets were moved into separate files, thus
152       making it possible to unify the bulk of the socket implementations
153       between Unix and Win32.
154    
155    3. CML's socket API changed accordingly.
156       (Note that we need to remove non-blocking functions from this API
157       since they are redundant in the case of CML!)
158    
159    4. CML's socket implementation now makes use of non-blocking functions
160       provided by Basis, thus removing all OS-dependent code from this part
161       of CML.
162    
163    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
164    
165    ----------------------------------------------------------------------
166    Name: Matthias Blume (blume (at) tti - c (dot) org)
167    Date: 2003/09/22 12:10:00 CDT
168    Tag: blume-20030922-110_43_2
169    Description:
170    
171    Made a new interim version and bootfiles for developer's bootstrapping
172    convenience.
173    
174       110.43.2 -- NEW BOOTFILES
175    
176    ----------------------------------------------------------------------
177    Name: Matthias Blume (blume (at) tti - c (dot) org)
178    Date: 2003/09/19 15:55:00 CDT
179    Tag: blume-20030919-cmdir
180    Description:
181    
182    1. new-install.sh -> install.sh
183    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
184    3. tweaked installer so that another name instead of .cm can be chosen
185       at install time (by setting the CM_DIR_ARC environment variable
186       during installation); once installation is complete, the name is
187       fixed
188    
189    ----------------------------------------------------------------------
190    Name: Matthias Blume (blume (at) tti - c (dot) org)
191    Date: 2003/09/18 16:00:00 CDT
192    Tag: blume-20030918-110_43_1
193    Description:
194    
195    Made a new interim version and bootfiles for developer's bootstrapping
196    convenience.
197    
198       110.43.1 -- NEW BOOTFILES
199    
200    ----------------------------------------------------------------------
201    Name: Matthias Blume (blume (at) tti - c (dot) org)
202    Date: 2003/09/18 15:20:00 CDT
203    Tag: blume-20030918-misc
204    Description:
205    
206    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
207       this seems to be controversial at the moment) and used those in
208       Posix.ProcEnv.times.
209    
210    2. Added Time.{from,to}Nanoseconds to Time.
211    
212    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
213       For example, fromLargeInt never needs to look at more than 3 "big
214       digits" to get its 53 bits of precision.
215    
216    ----------------------------------------------------------------------
217    Name: Matthias Blume (blume (at) tti - c (dot) org)
218    Date: 2003/09/17 16:30:00 CDT
219    Tag: blume-20030917-real32-slices
220    Description:
221    
222    Added an entry to the primitive environment
223    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
224    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
225    
226    Removed some of the "magic" constants in real64.sml and replaced them
227    with code that generates these values from their corresponding
228    integer counterparts.
229    
230    Made all(?) the slice-related changes to the Basis and made everything
231    compile again...
232    
233    ----------------------------------------------------------------------
234    Name: Matthias Blume (blume (at) tti - c (dot) org)
235    Date: 2003/09/15 17:45:00 CDT
236    Tag: blume-20030915-rbase
237    Description:
238    
239    Fixed bug in Real.fromLargeInt.
240    
241    ----------------------------------------------------------------------
242    Name: Matthias Blume (blume (at) tti - c (dot) org)
243    Date: 2003/09/13 18:11:00 CDT
244    Tag: blume-20030913-libinstall
245    Description:
246    
247    Minor bugfix in config/libinstall (set anchor with path to
248    standalone tool after installing it, otherwise libraries that
249    need ml-lex or ml-yacc won't compile the first time the installer
250    runs).
251    
252    ----------------------------------------------------------------------
253    Name: Matthias Blume (blume (at) tti - c (dot) org)
254    Date: 2003/09/12 11:45:00 CDT
255    Tag: blume-20030912-various
256    Description:
257    
258     - fixed bug in Real.toLargeInt
259     - fixed bug in Posix.ProcEnv.times
260     - changed inputLine functions to return an option
261     - minor installer improvements / bugfixes
262     - changed default @SMLalloc parameter for x86/celeron to 64k
263    
264    ----------------------------------------------------------------------
265    Name: Matthias Blume (blume (at) tti - c (dot) org)
266    Date: 2003/09/09 22:00:00 CDT
267    Tag: Release_110_43
268    Description:
269    
270        New working release 110.43.  New bootfiles.
271    
272    ----------------------------------------------------------------------
273    Name: Matthias Blume (blume (at) tti - c (dot) org)
274    Date: 2003/09/09 19:20:00 CDT
275    Tag: blume-20030909-installer
276    Description:
277    
278    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
279    Modified config/install.bat to take advantage of it.  Also modified
280    config/install.sh (and called it config/new-install.sh) to take advantage
281    of it on Unix systems.  (The SML code is (supposed to be) platform-
282    independent.)
283    
284    The installer can now install everything under Win32
285    as well as under *nix as long as it compiles.
286    
287    Other changes:
288    
289      - made CML compile again under Win32
290      - made eXene compile under Win32 (by providing a fake structure UnixSock
291        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
292      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
293        same as type int;  under Win32 it isn't
294      - fixed some slice-related problems in the win32-specific parts of CML
295      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
296        (under Win32, the volume name is case-insensitive, and the
297        OS.Path code compares volume names for equality)
298    
299    ----------------------------------------------------------------------
300    Name: Matthias Blume (blume (at) tti - c (dot) org)
301    Date: 2003/09/08 11:55:00 CDT
302    Tag: blume-20030908-fullpath
303    Description:
304    
305    Made Win32 version of OS.FileSys.fullPath return current directory
306    when given an empty string.  This is what the spec says, and incidentally,
307    CM depends on it.  (CM otherwise goes into an infinite loop in certain
308    cases when presented with the name of a non-existing .cm file.)
309    
310    ----------------------------------------------------------------------
311    Name: Matthias Blume (blume (at) tti - c (dot) org)
312    Date: 2003/09/04 16:30:00 CDT
313    Tag: blume-20030905-slices-etc
314    Description:
315    
316      1. Changed interface to vectors and arrays in Basis to match
317         (draft) Basis spec.
318      2. Added signatures and implementations of slices according to
319         Basis spec.
320      3. Edited source code throughout the system to make it compile again
321         under 1. and 2.  (In some cases code had to be added to have it
322         match the new signatures.)
323      4. MLRISC should be backward-compatible: the copies of the originals
324         of files that needed to change under 3. were retained, the .cm files
325         check the compiler version number and use old versions when
326         appropriate.
327      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
328         dirstream -> string option (in accordance with Basis spec).
329      6. When generating code that counts lines, ml-lex used function
330         CharVector.foldli, taking advantage of its old interface.
331         This has been replaced with the corresponding code from
332         CharVectorSlice.  (html-lex must be re-lexed!)
333      7. BitArray in smlnj-lib/Util has been extended/modified to match the
334         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
335      8. Removed temporary additions (fromInternal, toInternal) from the
336         (now obsolete) IntInf in smlnj-lib/Util.
337      9. Cleaned up structure Byte.
338     10. Added localOffset, scan, and fromString to Date (according to spec).
339         Cleaned/corrected implementation of Date.
340         (Still need to check for correctness; implement better canonicalizeDate.)
341     11. Added "scan" to signature IEEE_REAL.
342     12. Some improvements to IntInf [in particular: efficiency-hack for
343         mod and rem when second operand is 2 (for parity checks).]
344     13. Changed representation of type Time.time, using a single IntInf.int
345         value counting microseconds.  This considerably simplified the
346         implementation of structure Time.  We now support negative time
347         values; scan and fromString handle signs.
348     14. Functor PrimIO now takes two additional arguments (VectorSlice and
349         ArraySlice).
350    
351    ----------------------------------------------------------------------
352    Name: Matthias Blume (blume (at) tti - c (dot) org)
353    Date: 2003/08/28 17:00:00 CDT
354    Tag: blume-20030828-intinf
355    Description:
356    
357    This is a major update which comes with a version number bump
358    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
359    and an implementation of IntInf in the Basis.
360    
361    There are a fairly large number of related changes and updates throughout
362    the system:
363    
364    Basis:
365            - Implemented IntInf.
366            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
367            - Added some missing Real64 operations, most notably Real.toLargeInt.
368            - Added FixedInt as a synonym for Int32.
369    
370    compiler:
371            * Added support for a built-in intinf type.
372              - literals
373              - pattern matching
374              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
375              - overloading on literals and operations
376    
377            This required adding a primitive type intinf, some additional
378            primops, and implementations for several non-trivial intinf
379            operations in Core.  (The intinf type is completely abstract
380            to the compiler; all operations get delegated back to the Core.)
381    
382            * Intinf equality is handled by polyequal.  However, the compiler
383            does not print its usual warning in this case (since polyequal
384            is the right thing to do there).
385    
386            * Improved the organization of structure InlineT.
387    
388            * A word about conversion primops:
389            If conversions involving intinf do not cancel out during
390            CPS contract, then the compiler must insert calls to Core functions.
391            Since all core access must be resolved already during the FLINT
392            translate phase, it would be too late a the time of CPS contract
393            to add new Core calls.  For this reason, conversion primops
394            for intinf carry two arguments:  1. the numeric argument that
395            they are supposed to convert, and 2. the Core function that
396            can help with this conversion if necessary.  If CPS contract
397            eliminates a primop, then the associated Core function becomes
398            dead and goes away.  Intinf conversion primops that do not get
399            eliminated by CPS contract get rewritten into calls of their
400            core functions by a separate, new phase.
401    
402    interactive system:
403            - Control.Print.intinfDepth controls max length of intinf constants
404            being printed. (Analogous to Control.Print.stringDepth.)
405            - Cleanup in printutil and pputil: got rid of unused stuff and
406            duplicates; replaced some of the code with code that makes better
407            use of library functionality.
408    
409    CM:
410            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
411            are no longer silent.
412    
413    CKIT:
414            Fixed mismatched uses of Int32 and LargeInt.  I always decided
415            in favor of LargeInt -- which is now the same as IntInf.
416            CKIT-knowledgable people should check whether this is what's
417            intended and otherwise change things back to using Int32 or
418            FixedInt.
419    
420    Throughout the code:
421            Started using IntInf.int literals and built-in operations
422            (e.g., comparison with 0) where this seems appropriate.
423    
424    
425    ----------------------------------------------------------------------
426    Name: Dave MacQueen (dbm@cs.uchicago.edu)
427    Date: 2003/08/13 11:36:00 CDT
428    Tag: dbm-20030813-mcz-merge1
429    Description:
430    
431    Merging changes from the mcz-branch development branch into trunk.
432    These changes involve replacement of the emulated old prettyprinter
433    interface with direct use of the SML/NJ Lib PP library, and fixing
434    of a couple of bugs (895, 1186) relating to error messages.  A new
435    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
436    has been added.
437    
438    ----------------------------------------------------------------------
439    Name: Matthias Blume (blume (at) tti - c (dot) org)
440    Date: 2003/08/11 15:45:00 CDT
441    Tag: blume-20030811-windows
442    Description:
443    
444    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
445    
446        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
447    
448    This patch restores SML/NJ's ability to run under win32.  There are a
449    number of changes, including fixes for several bugs that had gone
450    unnoticed until now:
451    
452      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
453        trying to run the win32 version from within cygwin.)
454    
455      - There are a number of simple .bat scripts that substitute for their
456        corresponding Unix shell-scripts.  (See below.)
457    
458      - The internals of ml-build have been modified slightly.  The main
459        difference is that instead of calling ".link-sml" (or link-sml.bat)
460        using OS.Process.system, the ML process delegates this task back
461        to the script.  Otherwise problems arise in mixed environments such
462        as Cygwin where scripts look and work like Unix scripts, but
463        where OS.Process.system cannot run them.
464    
465      - In CM, the srcpath pickler used native pathname syntax -- which
466        is incorrect in the case of cross-compilation.  The new pickle format
467        is independent of platform-specific naming conventions.
468    
469      - Path configuration files (such as lib/pathconfig) can now choose
470        between native and standard syntax.  Placing a line of the form
471    
472            standard!
473    
474        into the file causes all subsequent paths to be interpreted using
475        CM standard pathname syntax (= Unix conventions); a line
476    
477            native!
478    
479        switches back to native style.  This was needed so that
480        path config files can be written portably, see src/system/pathconfig.
481    
482      - Runtime system:
483    
484          - win32-filesys.c:  get_file_time and set_file_time now
485            access modification time, not creation time.
486    
487          - I/O code made aware of new array representation.
488    
489          - Bug fixes in X86.prim.masm.
490    
491       - src/system/makeml made aware of win32. (For use under cygwin
492            and other Unix-environments for windows.)
493    
494       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
495         which caused BinIO.inputAll to fail consistently.
496    
497    .bat scripts:
498    
499       Windows .bat scripts assume that SMLNJ_HOME is defined.
500    
501       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
502         applications (sml, ml-yacc, ml-lex).
503       - ml-build.bat: analogous to ml-build.
504       - config\install.bat:  Analogous to config/install.sh.  This requires
505         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
506         (nmake etc. must be on the path, and vcvars32 must have been run.)
507         Moreover, sources for ml-lex and ml-yacc need to exist under src,
508         and the bootfile hierarchy must have been unpacked under
509         sml.boot.x86-win32.
510         The script is very primitive and does a poor job at error checking.
511         It only installs the base system, ml-lex, and ml-yacc.  No other
512         libraries are being installed (i.e., you get only those that
513         are part of the compiler.)
514       - link-sml.bat:  analogous to .link-sml, but not currently used
515    
516    Unrelated bug fixes:
517    
518      - ml-nlffigen now exports structures ST_* corresponding to incomplete
519        types.
520      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
521        otherwise.)
522    
523    ----------------------------------------------------------------------
524    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
525    Date: 2003/06/17
526    Tag: macqueen-20030617-bug895
527    Description:
528    
529    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
530    Tag will be used for new development branch (mcz-branch) for
531    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
532    
533    ----------------------------------------------------------------------
534    Name: Matthias Blume (blume (at) tti - c (dot) org)
535    Date: 2003/05/27 16:55:00 CDT
536    Tag: blume-20030527-polyeq
537    Description:
538    
539    Tried to eliminated most cases of polymorphic equality.
540    
541    ----------------------------------------------------------------------
542    Name: Matthias Blume (blume (at) tti - c (dot) org)
543    Date: 2003/05/21 17:45:00 CDT
544    Tag: blume-20030517-complete
545    Description:
546    
547    Two changes:
548    
549      1. Added a flag for controlling whether non-exhaustive bindings will
550         be treated as errors (default is false).
551      2. Cleaned up the *entire* source tree so that CMB.make goes through
552         without a single non-exhaustive match- or bind warning.
553    
554    ----------------------------------------------------------------------
555    Name: Matthias Blume (blume (at) tti - c (dot) org)
556    Date: 2003/05/17 10:20:00 CDT
557    Tag: blume-20030517-absyn
558    Description:
559    
560    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
561    
562       This mainly affects the quality of error messages.  However, some
563       of the code is now more straightforward than before. (Treatment of
564       the above four constructs in translate.sml is much simpler than
565       the "macro-expansion" that was going on before.  Plus, the mach-
566       compiler no longer gets invoked just to be able to compile an
567       if-expression.)
568    
569    2. The ErrorMsg.Error exception is now caught and absorbed by the
570       interactive loop.
571    
572    ----------------------------------------------------------------------
573    Name: Allen Leung
574    Date: 2003/05/16 13:05:00 CDT
575    Tag: leunga-20030516-cygwin-runtime
576    Description:
577    
578        Ported the runtime system to cygwin, which uses the unix
579    x86-unix bin files.  Missing/buggy features:
580    
581       o getnetbyname, getnetbyaddr: these functions seem to be missing in
582         the Cygwin library.
583       o Ctrl-C handling may be flaky.
584       o Windows system calls and Windows I/O are not supported.
585    
586       A new set of binfiles is located at:
587    
588         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
589    
590       This is only needed for bootstrapping the cygwin version of smlnj.
591       Other x86 versions can use the existing binfiles.
592    
593    ----------------------------------------------------------------------
594    Name: Matthias Blume
595    Date: 2003/04/08 15:42:00 CDT
596    Tag: blume-20030408-listpair
597    Description:
598    
599    1. Added a target 'mlrisc' to installer.
600    
601    2. Added missing elements to structure ListPair.
602    
603    ----------------------------------------------------------------------
604    Name: Allen Leung
605    Date: 2003/01/07 10:40:00 EST
606    Tag: leunga-20030107-int-rem
607    Description:
608    
609    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
610    The arguments to the SUBL instruction were swapped.
611    
612    ----------------------------------------------------------------------
613    Name: Matthias Blume
614    Date: 2002/12/12 16:25:00 EST
615    Tag: blume-20021212-risc-ra
616    Description:
617    
618    Fixed a serious bug in the rewrite code for FP spilling/reloading that
619    sent the RA into an infinite loop when floating point registers get
620    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
621    110.37 and 110.38.)
622    There was another set of potential problems related to the handling of
623    MLRISC annotations (but those did not yet cause real problems, apparently).
624    
625    ----------------------------------------------------------------------
626    Name: Matthias Blume
627    Date: 2002/12/06 22:40:00 EST
628    Tag: blume-20021206-cm-fileid
629    Description:
630    
631    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
632    This fixes the problem of CM getting confused by files that suddenly
633    change their identity (e.g., by getting unlinked and recreated by some
634    text editor such as vi).  There might be a better/cheaper/cleaner way
635    of doing this, but for now this will have to do.
636    
637    ----------------------------------------------------------------------
638    Name: Matthias Blume
639    Date: 2002/10/28 09:50:00 EST
640    Tag: blume-20021028-typecheck
641    Description:
642    
643    Exported structure Typecheck from $smlnj/viscomp/core.cm.
644    
645    ----------------------------------------------------------------------
646    Name: Matthias Blume
647    Date: 2002/10/17 09:10:00 EDT
648    Tag: Release_110_42
649    Description:
650    
651    In good old tradition, there has been a slight hiccup so that we have
652    to patch 110.42 after the fact.   The old release tag has been replaced
653    (see below).
654    
655    The change solves a problem with two competing approaches the
656    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
657    each other's way.
658    
659    This change only affects the runtime system code and the installer script.
660    (No new bootfiles.)
661    
662    ----------------------------------------------------------------------
663    Name: Matthias Blume
664    Date: 2002/10/16 12:00:00 EDT
665    Tag: Release_110_42_removed
666    Description:
667    
668    New working release.  New bootfiles.
669    
670    ----------------------------------------------------------------------
671    Name: Matthias Blume
672    Date: 2002/10/10 13:10:00 EDT
673    Tag: blume-20021010-ppc-divs
674    Description:
675    
676    The mltree operator DIVS must be implemented with an overflow check on
677    the PPC because the hardware indicates divide-by-zero using "overflow" as
678    well.
679    
680    ----------------------------------------------------------------------
681    Name: Matthias Blume
682    Date: 2002/07/23 11:45:00
683    Tag: blume-20020723-smlnj-home
684    Description:
685    
686    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
687    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
688    CM_PATHCONFIG is also set), the path configuration file is assumed
689    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
690    the entire tree to some other place and everything will "just work".
691    
692    (Companion commands such as ml-build and ml-makedepend also sense this
693    variable.)
694    
695    ----------------------------------------------------------------------
696    Name: Matthias Blume
697    Date: 2002/07/12 21:19:00 EDT
698    Tag: blume-20020712-liveness
699    Description:
700    
701    Exported two useful "step" functions from liveness module (MLRISC).
702    
703    ----------------------------------------------------------------------
704    Name: Matthias Blume
705    Date: 2002/07/05 16:00 EDT
706    Tag: Release_110_41
707    Description:
708    
709    New working release.  New bootfiles.
710    
711    ----------------------------------------------------------------------
712    Name: Matthias Blume
713    Date: 2002/07/05 10:25:00 EDT
714    Tag: blume-20020705-btimp
715    Description:
716    
717    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
718    other clients can set up backtracing support.
719    
720    ----------------------------------------------------------------------
721    Name: Matthias Blume
722    Date: 2002/06/25 17:23:00 EDT
723    Tag: blume-20020625-fpmax
724    Description:
725    
726    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
727    case.  (The sense of the isNaN test was reversed -- which made min and
728    max always return their first argument.)
729    
730    ----------------------------------------------------------------------
731    Name: Matthias Blume
732    Date: 2002/06/11
733    Tag: blume-20020611-unixpath
734    Description:
735    
736    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
737    
738    ----------------------------------------------------------------------
739    Name: Matthias Blume
740    Date: 2002/06/10 16:35:00 EDT
741    Tag: blume-20020610-ieeereal
742    Description:
743    
744    I back-ported my implementation of IEEEReal.fromString from the
745    idlbasis-devel branch so that we can test it.
746    
747    Another small change is that ppDec tries to give more information
748    than just "<sig>" in the case of functors.  However, this code is
749    broken in some mysterious way if the functor's body's signature
750    has not been declared by ascription but gets inferred from the
751    implementation.  This needs fixing...
752    
753    ----------------------------------------------------------------------
754    Name: Matthias Blume
755    Date: 2002/05/31
756    Tag: blume-20020531-btrace-mode
757    Description:
758    
759    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
760    the wayside when I switched over to using Controls everywhere.)
761    
762    ----------------------------------------------------------------------
763    Name: Lal George
764    Date: 2002/05/23 12:21:40 EDT
765    Tag: george-20020523-visual-labels
766    Description:
767    
768    Labels are now displayed in the graphical output to make
769    the fall-through and target blocks obvious.
770    
771    ----------------------------------------------------------------------
772    Name: Matthias Blume
773    Date: 2002/05/22 11:03:00 EDT
774    Tag: blume-20020522-shrink
775    Description:
776    
777    John tweaked yesterday's fix for 1131 to handle an out-of-memory
778    situation that comes up when allocating huge arrays.
779    
780    ----------------------------------------------------------------------
781    Name: Matthias Blume
782    Date: 2002/05/21 16:00:00 EDT
783    Tag: Release_110_40
784    Description:
785    
786    New working release (110.40).  New bootfiles.
787    
788    [Also: John Reppy fixed GC bug 1131.]
789    
790    ----------------------------------------------------------------------
791    Name: Matthias Blume
792    Date: 2002/05/21 12:35:00 EDT
793    Tag: blume-20020521-cmdoc
794    Description:
795    
796    CM documentation update.
797    
798    ----------------------------------------------------------------------
799    Name: Matthias Blume
800    Date: 2002/05/21 10:55:00 EDT
801    Tag: blume-20020521-misc
802    Description:
803    
804    - John tweaked runtime to be silent on heap export (except when
805      GC messages are on).
806    - I added a few more things (cross-compiling versions of CMB) to
807      config/preloads (as suggestions).
808    
809    ----------------------------------------------------------------------
810    Name: Matthias Blume
811    Date: 2002/05/20 22:25:00 EDT
812    Tag: blume-20020520-controls
813    Description:
814    
815    - Added ControlUtil structure to control-lib.cm.
816    - Use it throughout.
817    - Used Controls facility to define MLRISC controls (as opposed to
818      registering MLRISC control ref cells with Controls after the fact)
819    - Fixed messed-up controls priorities.
820    
821    * Removed again all the stuff from config/preloads that one wouldn't
822      be able to preload at the time the initial heap image is built.
823      (Many libraries, e.g., CML, do not exist yet at this time.  The
824       only libraries that can be preloaded via config/preloads are those
825       that come bundled with the bootfiles.)
826    
827    ----------------------------------------------------------------------
828    Name: Matthias Blume
829    Date: 2002/05/20 10:59:00 EDT
830    Tag: blume-20020520-preloads
831    Description:
832    
833    Added a lot of commented-out suggestions for things to be included
834    in config/preloads.
835    
836    ----------------------------------------------------------------------
837    Name: Allen Leung
838    Date: 2002/05/18 14:20:00 EDT
839    Tag: leunga-20020518-mdl
840    Description:
841    
842    o Made the mdl tool stuff compile and run again.
843    o I've disabled all the stuff that depends on RTL specifications; they
844      are all badly broken anyway.
845    
846    ----------------------------------------------------------------------
847    Name: Matthias Blume
848    Date: 2002/05/17 16:49:00 EDT
849    Tag: blume-20020517-controls
850    Description:
851    
852    0. John Reppy made several modifications to the SML/NJ library.
853       In particular, there is a shiny new controls-lib.cm.
854    
855    1. Pushed new controls interface through compiler so that everything
856       compiles again.
857    
858    2. Added FormatComb and FORMAT_COMB to the CML version of the
859       SML/NJ library (so that CML compiles again).
860    
861    3. Modified init scripts because XXX_DEFAULT environment variables
862       are no longer with us.  (Boot-time initialization is now done
863       using the same environment variables that are also used for
864       startup-time initialization of controls.)
865    
866    ----------------------------------------------------------------------
867    Name: Lal George
868    Date: 2002/05/15 09:20:10 EDT
869    Tag: george-20020515-pseudo-op-decls
870    Description:
871    
872            All pseudo-ops emitted before the first segment declaration
873    such as TEXT, DATA, and BSS directives are assumed to be global
874    declarations and are emitted first in the assembly file. This is
875    useful in a number of situations where one has pseudo-ops that are not
876    specific to any segment, and also works around the constraint that one
877    cannot have client pseudo-ops in the TEXT segment.
878    
879            Because no segment is associated with these declarations it is
880    an error to allocate any space or objects before the first segment
881    directive and an exception will be raised. However, we cannot make
882    this check for client pseudo-ops.
883    
884            These top level declarations are a field in the CFG graph_info.
885    In theory you can continue to add to this field after the CFG has been
886    built -- provided you know what you are doing;-)
887    
888    ----------------------------------------------------------------------
889    Name: Matthias Blume
890    Date: 2002/05/13 16:40:00 EDT
891    Tag: blume-20020513-pp-etc
892    Description:
893    
894    A few minor bugfixes:
895    
896      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
897        (Bogus printouts for redefined bindings still occur.  Compiler
898        bug should no longer occur now.  We need to redo the prettyprinter
899        from scratch.)
900    
901      - CM pathname printer now also adds escape sequences for ( and )
902    
903      - commend and docu fixes for ml-nlffi
904    
905    ----------------------------------------------------------------------
906    Name: Matthias Blume
907    Date: 2002/05/10 16:40:00 EDT
908    Tag: blume-20020510-erg-textio
909    Description:
910    
911    Applied the following bugfix provided by Emden Gansner:
912    
913        Output is corrupted when outputSubstr is used rather than output.
914    
915        The problem occurs when a substring
916    
917            ss = (s, dataStart, dataLen)
918    
919        where dataStart > 0, fills a stream buffer with avail bytes left.
920        avail bytes of s, starting at index dataStart, are copied into the
921        buffer, the buffer is flushed, and then the remaining dataLen-avail
922        bytes of ss are copied into the beginning of the buffer. Instead of
923        starting this copy at index dataStart+avail in s, the current code
924        starts the copy at index avail.
925    
926        Fix:
927        In text-io-fn.sml, change line 695 from
928             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
929        to
930             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
931    
932    ----------------------------------------------------------------------
933    Name: Matthias Blume
934    Date: 2002/04/12 13:55:00 EDT
935    Tag: blume-20020412-assyntax
936    Description:
937    
938    1. Grabbed newer assyntax.h from the XFree86 project.
939    2. Fiddled with how to compile X86.prim.asm without warnings.
940    3. (Very) Minor cleanup in CM.
941    
942    ----------------------------------------------------------------------
943    Name: Matthias Blume
944    Date: 2002/04/01 (no joke!) 17:07:00 EST
945    Tag: blume-20020401-x86div
946    Description:
947    
948    Added full support for div/mod/rem/quot on the x86, using the machine
949    instruction's two results (without clumsily recomputing the remainder)
950    directly where appropriate.
951    
952    Some more extensive power-of-two support was added to the x86 instruction
953    selector (avoiding expensive divs, mods, and muls where they can be
954    replaced with cheaper shifts and masks).  However, this sort of thing
955    ought to be done earlier, e.g., within the CPS optimizer so that
956    all architectures benefit from it.
957    
958    The compiler compiles to a fixed point, but changes might be somewhat
959    fragile nevertheless.  Please, report any strange things that you might
960    see wrt. div/mod/quot/rem...
961    
962    ----------------------------------------------------------------------
963    Name: Matthias Blume
964    Date: 2002/03/29 17:22:00
965    Tag: blume-20020329-div
966    Description:
967    
968    Fixed my broken div/mod logic.  Unfortunately, this means that the
969    inline code for div/mod now has one more comparison than before.
970    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
971    The problem was with quotient = 0, because that alone does not tell
972    us which way the rounding went.  One then has to look at whether
973    remainder and divisor have the same sign...  :(
974    
975    Anyway, I replaced the bootfiles with fresh ones...
976    
977    ----------------------------------------------------------------------
978    Name: Matthias Blume
979    Date: 2002/03/29 14:10:00 EST
980    Tag: blume-20020329-inlprims
981    Description:
982    
983    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
984    
985    Primops have changed. This means that the bin/boot-file formats have
986    changed as well.
987    
988    To make sure that there is no confusion, I made a new version.
989    
990    
991    CHANGES:
992    
993    * removed REMT from mltree (remainder should never overflow).
994    
995    * added primops to deal with divisions of all flavors to the frontend
996    
997    * handled these primops all the way through so they map to their respective
998      MLRISC support
999    
1000    * used these primops in the implementation of Int, Int32, Word, Word32
1001    
1002    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1003    
1004    * parameterized INLMIN, INLMAX, and INLABS by a numkind
1005    
1006    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1007      floating point
1008    
1009    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1010      and Real (but Real.abs maps to a separate floating-point-only primop)
1011    
1012    
1013    TODO items:
1014    
1015    * Hacked Alpha32 instruction selection, disabling the selection of REMx
1016      instructions because the machine instruction encoder cannot handle
1017      them.  (Hppa, PPC, and Sparc instruction selection did not handle
1018      REM in the first place, and REM is supported by the x86 machine coder.)
1019    
1020    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1021      selection phase.  (The two can be streamlined because the hardware
1022      delivers both quotient and remainder at the same time anyway.)
1023    
1024    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1025      (Currently the behavior is inconsistent both across architectures and
1026      wrt. the draft Basis spec.)
1027    
1028    * Word8 should eventually be handled natively, too.
1029    
1030    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1031      as if there currently is no execution path that could trigger it in
1032      SML/NJ.  (The assumptions underlying functions arith and promotable do not
1033      hold for things like multiplication and division.)
1034    
1035    ----------------------------------------------------------------------
1036    Name: Matthias Blume
1037    Date: 2002/03/27 16:27:00 EST
1038    Tag: blume-20020327-mlrisc-divisions
1039    Description:
1040    
1041    Added support for all four division operations (ML's div, mod, quot,
1042    and rem) to MLRISC.  In the course of doing so, I also rationalized
1043    the naming (no more annoying switch-around of DIV and QUOT), by
1044    parameterizing the operation by div_rounding_mode (which can be either
1045    DIV_TO_ZERO or DIV_TO_NEGINF).
1046    
1047    The generic MLTreeGen functor takes care of compiling all four
1048    operations down to only round-to-zero div.
1049    
1050    Missing pieces:
1051    
1052      * Doing something smarter than relying on MLTreeGen on architectures
1053        like, e.g., the x86 where hardware division delivers both quotient and
1054        remainder at the same time.  With this, the implementation of the
1055        round-to-neginf operations could be further streamlined.
1056    
1057      * Remove inlining support for div/mod/rem from the frontend and replace it
1058        with primops that get carried through to the backend.  Do this for all
1059        int and word types.
1060    
1061    ----------------------------------------------------------------------
1062    Name: Matthias Blume
1063    Date: 2002/03/25 17:25:00 EST
1064    Tag: blume-20020325-divmod
1065    Description:
1066    
1067    I improved (hopefully without breaking them) the implementation of Int.div,
1068    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1069    advantage of the following observations:
1070    
1071      Let  q = x quot y      r = x rem y
1072           d = x div  y      m = x mod y
1073    
1074    where "quot" is the round-to-zero version of integer division that
1075    hardware usually provides.  Then we have:
1076    
1077         r = x - q * y        where neither the * nor the - will overflow
1078         d = if q >= 0 orelse x = q * y then q else q - 1
1079                              where neither the * nor the - will overflow
1080         m = if q >= 0 orelse r = 0 then r else r + y
1081                              where the + will not overflow
1082    
1083    This results in substantial simplification of the generated code.
1084    The following table shows the number of CFG nodes and edges generated
1085    for
1086            fun f (x, y) = x OPER y
1087            (* with OPER \in div, mod, quot, rem *)
1088    
1089    
1090        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1091        --------------------------------------------------------
1092         div |         24 |         39 |         12 |         16
1093         mod |         41 |         71 |         12 |         16
1094        quot |          8 |         10 |          8 |         10
1095         rem |         10 |         14 |          8 |         10
1096    
1097    
1098    ----------------------------------------------------------------------
1099    Name: Matthias Blume
1100    Date: 2002/03/25 22:06:00 EST
1101    Tag: blume-20020325-cprotobug
1102    Description:
1103    
1104    Fixed a bug in cproto (c prototype decoder).
1105    
1106    ----------------------------------------------------------------------
1107    Name: Matthias Blume
1108    Date: 2002/03/25 16:00:00 EST
1109    Tag: blume-20020325-raw-primops
1110    Description:
1111    
1112    I did some cleanup to Allen's new primop code and
1113    replaced yesterday's bootfiles with new ones.
1114    (But they are stored in the same place.)
1115    
1116    ----------------------------------------------------------------------
1117    Name: Matthias Blume
1118    Date: 2002/03/24 22:40:00 EST
1119    Tag: blume-20020324-bootfiles
1120    Description:
1121    
1122    Made the bootfiles that Allen asked for.
1123    
1124    ----------------------------------------------------------------------
1125    Name: Allen Leung
1126    Date: 2002/03/23 15:50:00 EST
1127    Tag: leunga-20020323-flint-cps-rcc-primops
1128    Description:
1129    
1130      1. Changes to FLINT primops:
1131    
1132        (* make a call to a C-function;
1133         * The primop carries C function prototype information and specifies
1134         * which of its (ML-) arguments are floating point. C prototype
1135         * information is for use by the backend, ML information is for
1136         * use by the CPS converter. *)
1137      | RAW_CCALL of { c_proto: CTypes.c_proto,
1138                       ml_args: ccall_type list,
1139                       ml_res_opt: ccall_type option,
1140                       reentrant : bool
1141                     } option
1142       (* Allocate uninitialized storage on the heap.
1143        * The record is meant to hold short-lived C objects, i.e., they
1144        * are not ML pointers.  With the tag, the representation is
1145        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1146        *)
1147      | RAW_RECORD of {tag:bool,sz:int}
1148      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1149    
1150      2.  These CPS primops are now overloaded:
1151    
1152           rawload of {kind:numkind}
1153           rawstore of {kind:numkind}
1154    
1155          The one argument form is:
1156    
1157             rawload {kind} address
1158    
1159          The two argument form is:
1160    
1161             rawload {kind} [ml object, byte-offset]
1162    
1163      3. RAW_CCALL/RCC now takes two extra arguments:
1164    
1165         a. The first is whether the C call is reentrant, i.e., whether
1166            ML state should be saved and restored.
1167         b. The second argument is a string argument specifying the name of
1168            library and the C function.
1169    
1170         These things are currently not handled in the code generator, yet.
1171    
1172      4. In CProto,
1173    
1174         An encoding type of "bool" means "ml object" and is mapped into
1175         C prototype of PTR.  Note that "bool" is different than "string",
1176         even though "string" is also mapped into PTR, because "bool"
1177         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1178    
1179      5. Pickler/unpicker
1180    
1181         Changed to handle RAW_RECORD and newest RAW_CCALL
1182    
1183      6. MLRiscGen,
1184    
1185         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1186         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1187            in the file CodeGen/cpscompile/cps-c-calls.sml
1188    
1189      7. Added the conditional move operator
1190    
1191             condmove of branch
1192    
1193         to cps.  Generation of this is still buggy so it is currently
1194         disabled.
1195    
1196    ----------------------------------------------------------------------
1197    Name: Lal George
1198    Date: 2002/03/22 14:18:25 EST
1199    Tag: george-20020322-cps-branch-prob
1200    Description:
1201    
1202    Implemented the Ball-Larus branch prediction-heuristics, and
1203    incorporated graphical viewers for control flow graphs.
1204    
1205    Ball-Larus Heuristics:
1206    ---------------------
1207    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1208    
1209    By design it uses the Dempster-Shafer theory for combining
1210    probabilities.  For example, in the function:
1211    
1212        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1213    
1214    the ball-larus heuristics predicts that the n=0 is unlikely
1215    (OH-heuristic), and the 'then' branch is unlikely because of the
1216    RH-heuristic -- giving the 'then' branch an even lower combined
1217    probability using the Dempster-Shafer theory.
1218    
1219    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1220    probability of the 'then' branch because of the loop in the else
1221    branch.
1222    
1223    
1224    Graphical Viewing:
1225    ------------------
1226    I merely plugged in Allen's graphical viewers into the compiler. The
1227    additional code is not much. At the top level, saying:
1228    
1229            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1230    
1231    will display the graphical view of the control flow graph just before
1232    back-patching.  daVinci must be in your path for this to work. If
1233    daVinci is not available, then the default viewer can be changed
1234    using:
1235    
1236            Control.MLRISC.getString "viewer"
1237    
1238    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1239    course, these viewers must be in your path.
1240    
1241    The above will display the compilation unit at the level of clusters,
1242    many of which are small, boring, and un-interesting. Also setting:
1243    
1244            Control.MLRISC.getInt "cfg-graphical-view_size"
1245    
1246    will display clusters that are larger than the value set by the above.
1247    
1248    
1249    ----------------------------------------------------------------------
1250    Name: Matthias Blume
1251    Date: 2002/03/21 22:20:00 EST
1252    Tag: blume-20020321-kmp-bugfix
1253    Description:
1254    
1255    Changed the interface to the KMP routine in PreString and fixed
1256    a minor bug in one place where it was used.
1257    
1258    ----------------------------------------------------------------------
1259    Name: Allen Leung
1260    Date: 2002/03/21 20:30:00 EST
1261    Tag: leunga-20020321-cfg
1262    Description:
1263    
1264      Fixed a potential problem in cfg edge splitting.
1265    
1266    ----------------------------------------------------------------------
1267    Name: Allen Leung
1268    Date: 2002/03/21 17:15:00 EST
1269    Tag: leunga-20020321-x86-fp-cfg
1270    Description:
1271    
1272      1. Recoded the buggy parts of x86-fp.
1273    
1274         a. All the block reordering code has been removed.
1275            We now depend on the block placement phases to do this work.
1276    
1277         b. Critical edge splitting code has been simplified and moved into the
1278            CFG modules, as where they belong.
1279    
1280         Both of these were quite buggy and complex.  The code is now much, much
1281         simpler.
1282    
1283      2. X86 backend.
1284    
1285         a. Added instructions for 64-bit support.  Instruction selection for
1286            64-bit has not been committed, however, since that
1287            requires changes to MLTREE which haven't been approved by
1288            Lal and John.
1289    
1290         b. Added support for FUCOMI and FUCOMIP when generating code for
1291            PentiumPro and above.  We only generate these instructions in
1292            the fast-fp mode.
1293    
1294         c. Added cases for JP and JNP in X86FreqProps.
1295    
1296      3. CFG
1297    
1298         CFG now has a bunch of methods for edge splitting and merging.
1299    
1300      4. Machine description.
1301    
1302         John's simplification of MLTREE_BASIS.fcond broke a few machine
1303         description things:
1304    
1305         rtl-build.{sig,sml} and hppa.mdl fixed.
1306    
1307         NOTE: the machine description stuff in the repository is still broken.
1308               Again, I can't put my fixes in because that involves
1309               changes to MLTREE.
1310    
1311    ----------------------------------------------------------------------
1312    Name: Matthias Blume
1313    Date: 2002/03/20 15:55:00 EST
1314    Tag: blume-20020320-kmp
1315    Description:
1316    
1317    Implemented Knuth-Morris-Pratt string matching in PreString and used
1318    it for String.isSubstring, Substring.isSubstring, and
1319    Substring.position.
1320    
1321    (Might need some stress-testing.  Simple examples worked fine.)
1322    
1323    ----------------------------------------------------------------------
1324    Name: Matthias Blume
1325    Date: 2002/03/19 16:37:00 EST
1326    Tag: blume-20020319-witnesses
1327    Description:
1328    
1329    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1330    
1331    This implements a generic mechanism for changing constness qualifiers
1332    anywhere within big C types without resorting to outright "casts".
1333    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1334    modify the constness at the outermost level.)
1335    The implementation of "convert" is based on the idea of "witness"
1336    values -- values that are not used by the operation but whose types
1337    "testify" to their applicability.  On the implementation side, "convert"
1338    is simply a projection (returning its second curried argument).  With
1339    cross-module inlining, it should not result in any machine code being
1340    generated.
1341    
1342    ----------------------------------------------------------------------
1343    Name: Matthias Blume
1344    Date: 2002/03/15 16:40:00 EST
1345    Tag: blume-20020315-basis
1346    Description:
1347    
1348    Provided (preliminary?) implementations for
1349    
1350      {String,Substring}.{concatWith,isSuffix,isSubstring}
1351    
1352    and
1353    
1354      Substring.full
1355    
1356    Those are in the Basis spec but they were missing in SML/NJ.
1357    
1358    ----------------------------------------------------------------------
1359    Name: Matthias Blume
1360    Date: 2002/03/14 21:30:00 EST
1361    Tag: blume-20020314-controls
1362    Description:
1363    
1364    Controls:
1365    ---------
1366    
1367    1. Factored out the recently-added Controls : CONTROLS stuff and put
1368       it into its own library $/controls-lib.cm.  The source tree for
1369       this is under src/smlnj-lib/Controls.
1370    
1371    2. Changed the names of types and functions in this interface, so they
1372       make a bit more "sense":
1373    
1374          module -> registry
1375          'a registry -> 'a group
1376    
1377    3. The interface now deals in ref cells only.  The getter/setter interface
1378       is (mostly) gone.
1379    
1380    4. Added a function that lets one register an already-existing ref cell.
1381    
1382    5. Made the corresponding modifications to the rest of the code so that
1383       everything compiles again.
1384    
1385    6. Changed the implementation of Controls.MLRISC back to something closer
1386       to the original.  In particular, this module (and therefore MLRISC)
1387       does not depend on Controls.  There now is some link-time code in
1388       int-sys.sml that registers the MLRISC controls with the Controls
1389       module.
1390    
1391    CM:
1392    ---
1393    
1394      * One can now specify the lambda-split aggressiveness in init.cmi.
1395    
1396    ----------------------------------------------------------------------
1397    Name: Allen Leung
1398    Date: 2002/03/13 17:30:00 EST
1399    Tag: leunga-20020313-x86-fp-unary
1400    Description:
1401    
1402    Bug fix for:
1403    
1404    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1405    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1406    > - fun f(x,(y,z)) = Real.~ y;
1407    > [autoloading]
1408    > [autoloading done]
1409    >       fchsl   (%eax), 184(%esp)
1410    > Error: MLRisc bug: X86MCEmitter.emitInstr
1411    >
1412    > uncaught exception Error
1413    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1414    
1415    The problem was that the code generator did not generate any fp registers
1416    in this case, and the ra didn't know that it needed to run the X86FP phase to
1417    translate the pseudo fp instruction.   This only happened with unary fp
1418    operators in certain situations.
1419    
1420    ----------------------------------------------------------------------
1421    Name: Matthias Blume
1422    Date: 2002/03/13 14:00:00 EST
1423    Tag: blume-20020313-overload-etc
1424    Description:
1425    
1426    1. Added _overload as a synonym for overload for backward compatibility.
1427       (Control.overloadKW must be true for either version to be accepted.)
1428    
1429    2. Fixed bug in install script that caused more things to be installed
1430       than what was requested in config/targets.
1431    
1432    3. Made CM aware of the (_)overload construct so that autoloading
1433       works.
1434    
1435    ----------------------------------------------------------------------
1436    Name: Matthias Blume
1437    Date: 2002/03/12 22:03:00 EST
1438    Tag: blume-20020312-url
1439    Description:
1440    
1441    Forgot to update BOOT and srcarchiveurl.
1442    
1443    ----------------------------------------------------------------------
1444    Name: Matthias Blume
1445    Date: 2002/03/12 17:30:00 EST
1446    Tag: blume-20020312-version110392
1447    Description:
1448    
1449    Yet another version number bump (because of small changes to the
1450    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1451    
1452    Changes:
1453    
1454      The new pid generation scheme described a few weeks ago was overly
1455      complicated.  I implemented a new mechanism that is simpler and
1456      provides a bit more "stability":  Once CM has seen a compilation
1457      unit, it keeps its identity constant (as long as you do not delete
1458      those crucial CM/GUID/* files).  This means that when you change
1459      an interface, compile, then go back to the old interface, and
1460      compile again, you arrive at the original pid.
1461    
1462      There now also is a mechanism that instructs CM to use the plain
1463      environment hash as a module's pid (effectively making its GUID
1464      the empty string).  For this, "noguid" must be specified as an
1465      option to the .sml file in question within its .cm file.
1466      This is most useful for code that is being generated by tools such
1467      as ml-nlffigen (because during development programmers tend to
1468      erase the tool's entire output directory tree including CM's cached
1469      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1470      revert to the old, broken behavior of SML/NJ, but in specific cases
1471      where there is no danger of interface confusion, its use is ok
1472      (I think).
1473    
1474      ml-nlffigen by default generates "noguid" annotations.  They can be
1475      turned off by specifying -guid in its command line.
1476    
1477    ----------------------------------------------------------------------
1478    Name: Lal George
1479    Date: 2002/03/12 12 14:42:36 EST
1480    Tag: george-20020312-frequency-computation
1481    Description:
1482    
1483    Integrated jump chaining and static block frequency into the
1484    compiler. More details and numbers later.
1485    
1486    ----------------------------------------------------------------------
1487    Name: Lal George
1488    Date: 2002/03/11 11 22:38:53 EST
1489    Tag: george-20020311-jump-chain-elim
1490    Description:
1491    
1492    Tested the jump chain elimination on all architectures (except the
1493    hppa).  This is on by default right now and is profitable for the
1494    alpha and x86, however, it may not be profitable for the sparc and ppc
1495    when compiling the compiler.
1496    
1497    The gc test will typically jump to a label at the end of the cluster,
1498    where there is another jump to an external cluster containing the actual
1499    code to invoke gc. This is to allow factoring of common gc invocation
1500    sequences. That is to say, we generate:
1501    
1502            f:
1503               testgc
1504               ja   L1      % jump if above to L1
1505    
1506            L1:
1507               jmp L2
1508    
1509    
1510    After jump chain elimination the 'ja L1' instructions is converted to
1511    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1512    up being implemented in their long form (if L2 is far away) using:
1513    
1514            jbe     L3      % jump if below or equal to L3
1515            jmp     L2
1516         L3:
1517            ...
1518    
1519    
1520    For large compilation units L2  may be far away.
1521    
1522    
1523    ----------------------------------------------------------------------
1524    Name: Matthias Blume
1525    Date: 2002/03/11 13:30:00 EST
1526    Tag: blume-20020311-mltreeeval
1527    Description:
1528    
1529    A functor parameter was missing.
1530    
1531    ----------------------------------------------------------------------
1532    Name: Allen Leung
1533    Date: 2002/03/11 10:30:00 EST
1534    Tag: leunga-20020311-runtime-string0
1535    Description:
1536    
1537       The representation of the empty string now points to a
1538    legal null terminated C string instead of unit.  It is now possible
1539    to convert an ML string into C string with InlineT.CharVector.getData.
1540    This compiles into one single machine instruction.
1541    
1542    ----------------------------------------------------------------------
1543    Name: Allen Leung
1544    Date: 2002/03/10 23:55:00 EST
1545    Tag: leunga-20020310-x86-call
1546    Description:
1547    
1548       Added machine generation for CALL instruction (relative displacement mode)
1549    
1550    ----------------------------------------------------------------------
1551    Name: Matthias Blume
1552    Date: 2002/03/08 16:05:00
1553    Tag: blume-20020308-entrypoints
1554    Description:
1555    
1556    Version number bumped to 110.39.1.  NEW BOOTFILES!
1557    
1558    Entrypoints: non-zero offset into a code object where execution should begin.
1559    
1560    - Added the notion of an entrypoint to CodeObj.
1561    - Added reading/writing of entrypoint info to Binfile.
1562    - Made runtime system bootloader aware of entrypoints.
1563    - Use the address of the label of the first function given to mlriscGen
1564      as the entrypoint.  This address is currently always 0, but it will
1565      not be 0 once we turn on block placement.
1566    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
1567      with entry points) from mlriscGen.
1568    
1569    ----------------------------------------------------------------------
1570    Name: Allen Leung
1571    Date: 2002/03/07 20:45:00 EST
1572    Tag: leunga-20020307-x86-cmov
1573    Description:
1574    
1575       Bug fixes for CMOVcc on x86.
1576    
1577       1. Added machine code generation for CMOVcc
1578       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1579       3. CMOVcc cannot have an immediate operand as argument.
1580    
1581    ----------------------------------------------------------------------
1582    Name: Matthias Blume
1583    Date: 2002/03/07 16:15:00 EST
1584    Tag: blume-20020307-controls
1585    Description:
1586    
1587    This is a very large but mostly boring patch which makes (almost)
1588    every tuneable compiler knob (i.e., pretty much everything under
1589    Control.* plus a few other things) configurable via both the command
1590    line and environment variables in the style CM did its configuration
1591    until now.
1592    
1593    Try starting sml with '-h' (or, if you are brave, '-H')
1594    
1595    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1596    implements the underlying generic mechanism.
1597    
1598    The interface to some of the existing such facilities has changed somewhat.
1599    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1600    (The getFoo interface is still there for backward-compatibility, but its
1601    use is deprecated.)
1602    
1603    The ml-build script passes -Cxxx=yyy command-line arguments through so
1604    that one can now twiddle the compiler settings when using this "batch"
1605    compiler.
1606    
1607    TODO items:
1608    
1609    We should go through and throw out all controls that are no longer
1610    connected to anything.  Moreover, we should go through and provide
1611    meaningful (and correct!) documentation strings for those controls
1612    that still are connected.
1613    
1614    Currently, multiple calls to Controls.new are accepted (only the first
1615    has any effect).  Eventually we should make sure that every control
1616    is being made (via Controls.new) exactly once.  Future access can then
1617    be done using Controls.acc.
1618    
1619    Finally, it would probably be a good idea to use the getter-setter
1620    interface to controls rather than ref cells.  For the time being, both
1621    styles are provided by the Controls module, but getter-setter pairs are
1622    better if thread-safety is of any concern because they can be wrapped.
1623    
1624    *****************************************
1625    
1626    One bug fix: The function blockPlacement in three of the MLRISC
1627    backpatch files used to be hard-wired to one of two possibilities at
1628    link time (according to the value of the placementFlag).  But (I
1629    think) it should rather sense the flag every time.
1630    
1631    *****************************************
1632    
1633    Other assorted changes (by other people who did not supply a HISTORY entry):
1634    
1635    1. the cross-module inliner now works much better (Monnier)
1636    2. representation of weights, frequencies, and probabilities in MLRISC
1637       changed in preparation of using those for weighted block placement
1638       (Reppy, George)
1639    
1640    ----------------------------------------------------------------------
1641    Name: Lal George
1642    Date: 2002/03/07 14:44:24 EST 2002
1643    Tag: george-20020307-weighted-block-placement
1644    
1645    Tested the weighted block placement optimization on all architectures
1646    (except the hppa) using AMPL to generate the block and edge frequencies.
1647    Changes were required in the machine properties to correctly
1648    categorize trap instructions. There is an MLRISC flag
1649    "weighted-block-placement" that can be used to enable weighted block
1650    placement, but this will be ineffective without block/edge
1651    frequencies (coming soon).
1652    
1653    
1654    ----------------------------------------------------------------------
1655    Name: Lal George
1656    Date: 2002/03/05 17:24:48 EST
1657    Tag: george-20020305-linkage-cluster
1658    
1659    In order to support the block placement optimization, a new cluster
1660    is generated as the very first cluster (called the linkage cluster).
1661    It contains a single jump to the 'real' entry point for the compilation
1662    unit. Block placement has no effect on the linkage cluster itself, but
1663    all the other clusters  have full freedom in the manner in which they
1664    reorder blocks or functions.
1665    
1666    On the x86 the typical linkage code that is generated is:
1667       ----------------------
1668            .align 2
1669       L0:
1670            addl    $L1-L0, 72(%esp)
1671            jmp     L1
1672    
1673    
1674            .align  2
1675       L1:
1676       ----------------------
1677    
1678    72(%esp) is the memory location for the stdlink register. This
1679    must contain the address of the CPS function being called. In the
1680    above example, it contains the address of  L0; before
1681    calling L1 (the real entry point for the compilation unit), it
1682    must contain the address for L1, and hence
1683    
1684            addl $L1-L0, 72(%esp)
1685    
1686    I have tested this on all architectures except the hppa.The increase
1687    in code size is of course negligible
1688    
1689    ----------------------------------------------------------------------
1690    Name: Allen Leung
1691    Date: 2002/03/03 13:20:00 EST
1692    Tag: leunga-20020303-mlrisc-tools
1693    
1694      Added #[ ... ] expressions to mlrisc tools
1695    
1696    ----------------------------------------------------------------------
1697    Name: Matthias Blume
1698    Date: 2002/02/27 12:29:00 EST
1699    Tag: blume-20020227-cdebug
1700    Description:
1701    
1702    - made types in structure C and C_Debug to be equal
1703    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1704    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1705    
1706    ----------------------------------------------------------------------
1707    Name: Matthias Blume
1708    Date: 2002/02/26 12:00:00 EST
1709    Tag: blume-20020226-ffi
1710    Description:
1711    
1712    1. Fixed a minor bug in CM's "noweb" tool:
1713       If numbering is turned off, then truly don't number (i.e., do not
1714       supply the -L option to noweb).  The previous behavior was to supply
1715       -L'' -- which caused noweb to use the "default" line numbering scheme.
1716       Thanks to Chris Richards for pointing this out (and supplying the fix).
1717    
1718    2. Once again, I reworked some aspects of the FFI:
1719    
1720       A. The incomplete/complete type business:
1721    
1722       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
1723         gone!
1724       - ML types representing an incomplete type are now *equal* to
1725         ML types representing their corresponding complete types (just like
1726         in C).  This is still safe because ml-nlffigen will not generate
1727         RTTI for incomplete types, nor will it generate functions that
1728         require access to such RTTI.   But when ML code generated from both
1729         incomplete and complete versions of the C type meet, the ML types
1730         are trivially interoperable.
1731    
1732         NOTE:  These changes restore the full generality of the translation
1733         (which was previously lost when I eliminated functorization)!
1734    
1735       B. Enum types:
1736    
1737       - Structure C now has a type constructor "enum" that is similar to
1738         how the "su" constructor works.  However, "enum" is not a phantom
1739         type because each "T enum" has values (and is isomorphic to
1740         MLRep.Signed.int).
1741       - There are generic access operations for enum objects (using
1742         MLRep.Signed.int).
1743       - ml-nlffigen will generate a structure E_foo for each "enum foo".
1744         * The structure contains the definition of type "mlrep" (the ML-side
1745         representation type of the enum).  Normally, mlrep is the same
1746         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
1747         then mlrep will be defined as a datatype -- thus facilitating
1748         pattern matching on mlrep values.
1749         ("-ec" will be suppressed if there are duplicate values in an
1750          enumeration.)
1751         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
1752         will be generated for each C enum constant xxx.
1753         * Conversion functions m2i and i2m convert between mlrep and
1754         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
1755         * Coversion functions c and ml convert between mlrep and "tag enum".
1756         * Access functions (get/set) fetch and store mlrep values.
1757       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
1758         enumerations are merged into one single enumeration represented by
1759         structure E_'.
1760    
1761    ----------------------------------------------------------------------
1762    Name: Allen Leung
1763    Date: 2002/02/25 04:45:00 EST
1764    Tag: leunga-20020225-cps-spill
1765    
1766    This is a new implementation of the CPS spill phase.
1767    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
1768    In case of problems, replace it with the old file spill.sml
1769    
1770    The current compiler runs into some serious performance problems when
1771    constructing a large record.  This can happen when we try to compile a
1772    structure with many items.  Even a very simple structure like the following
1773    makes the compiler slow down.
1774    
1775        structure Foo = struct
1776           val x_1 = 0w1 : Word32.int
1777           val x_2 = 0w2 : Word32.int
1778           val x_3 = 0w3 : Word32.int
1779           ...
1780           val x_N = 0wN : Word32.int
1781        end
1782    
1783    The following table shows the compile time, from N=1000 to N=4000,
1784    with the old compiler:
1785    
1786    N
1787    1000   CPS 100 spill                           0.04u  0.00s  0.00g
1788           MLRISC ra                               0.06u  0.00s  0.05g
1789              (spills = 0 reloads = 0)
1790           TOTAL                                   0.63u  0.07s  0.21g
1791    
1792    1100   CPS 100 spill                           8.25u  0.32s  0.64g
1793           MLRISC ra                               5.68u  0.59s  3.93g
1794              (spills = 0 reloads = 0)
1795           TOTAL                                   14.71u  0.99s  4.81g
1796    
1797    1500   CPS 100 spill                           58.55u  2.34s  1.74g
1798           MLRISC ra                               5.54u  0.65s  3.91g
1799              (spills = 543 reloads = 1082)
1800           TOTAL                                   65.40u  3.13s  6.00g
1801    
1802    2000   CPS 100 spill                           126.69u  4.84s  3.08g
1803           MLRISC ra                               0.80u  0.10s  0.55g
1804              (spills = 42 reloads = 84)
1805           TOTAL                                   129.42u  5.10s  4.13g
1806    
1807    3000   CPS 100 spill                           675.59u  19.03s  11.64g
1808           MLRISC ra                               2.69u  0.27s  1.38g
1809              (spills = 62 reloads = 124)
1810           TOTAL                                   682.48u  19.61s  13.99g
1811    
1812    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
1813           MLRISC ra                               4.96u  0.27s  2.72g
1814              (spills = 85 reloads = 170)
1815           TOTAL                                   2375.26u  57.21s  48.00g
1816    
1817    As you can see the old cps spill module suffers from some serious
1818    performance problem.  But since I cannot decipher the old code fully,
1819    instead of patching the problems up, I'm reimplementing it
1820    with a different algorithm.  The new code is more modular,
1821    smaller when compiled, and substantially faster
1822    (O(n log n) time and O(n) space).  Timing of the new spill module:
1823    
1824    4000  CPS 100 spill                           0.02u  0.00s  0.00g
1825          MLRISC ra                               0.25u  0.02s  0.15g
1826             (spills=1 reloads=3)
1827          TOTAL                                   7.74u  0.34s  1.62g
1828    
1829    Implementation details:
1830    
1831    As far as I can tell, the purpose of the CPS spill module is to make sure the
1832    number of live variables at any program point (the bandwidth)
1833    does not exceed a certain limit, which is determined by the
1834    size of the spill area.
1835    
1836    When the bandwidth is too large, we decrease the register pressure by
1837    packing live variables into spill records.  How we achieve this is
1838    completely different than what we did in the old code.
1839    
1840    First, there is something about the MLRiscGen code generator
1841    that we should be aware of:
1842    
1843    o MLRiscGen performs code motion!
1844    
1845       In particular, it will move floating point computations and
1846       address computations involving only the heap pointer to
1847       their use sites (if there is only a single use).
1848       What this means is that if we have a CPS record construction
1849       statement
1850    
1851           RECORD(k,vl,w,e)
1852    
1853       we should never count the new record address w as live if w
1854       has only one use (which is often the case).
1855    
1856       We should do something similar to floating point, but the transformation
1857       there is much more complex, so I won't deal with that.
1858    
1859    Secondly, there are now two new cps primops at our disposal:
1860    
1861     1. rawrecord of record_kind option
1862        This pure operator allocates some uninitialized storage from the heap.
1863        There are two forms:
1864    
1865         rawrecord NONE [INT n]  allocates a tagless record of length n
1866         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
1867                                     and initializes the tag.
1868    
1869     2. rawupdate of cty
1870          rawupdate cty (v,i,x)
1871          Assigns to x to the ith component of record v.
1872          The storelist is not updated.
1873    
1874    We use these new primops for both spilling and increment record construction.
1875    
1876     1. Spilling.
1877    
1878        This is implemented with a linear scan algorithm (but generalized
1879        to trees).  The algorithm will create a single spill record at the
1880        beginning of the cps function and use rawupdate to spill to it,
1881        and SELECT or SELp to reload from it.  So both spills and reloads
1882        are fine-grain operations.  In contrast, in the old algorithm
1883        "spills" have to be bundled together in records.
1884    
1885        Ideally, we should sink the spill record construction to where
1886        it is needed.  We can even split the spill record into multiple ones
1887        at the places where they are needed.  But CPS is not a good
1888        representation for global code motion, so I'll keep it simple and
1889        am not attempting this.
1890    
1891     2. Incremental record construction (aka record splitting).
1892    
1893        Long records with many component values which are simulatenously live
1894        (recall that single use record addresses are not considered to
1895         be live) are constructed with rawrecord and rawupdate.
1896        We allocate space on the heap with rawrecord first, then gradually
1897        fill it in with rawupdate.  This is the technique suggested to me
1898        by Matthias.
1899    
1900        Some restrictions on when this is applicable:
1901        1. It is not a VECTOR record.  The code generator currently does not handle
1902           this case. VECTOR record uses double indirection like arrays.
1903        2. All the record component values are defined in the same "basic block"
1904           as the record constructor.  This is to prevent speculative
1905           record construction.
1906    
1907    ----------------------------------------------------------------------
1908    Name: Allen Leung
1909    Date: 2002/02/22 01:02:00 EST
1910    Tag: leunga-20020222-mlrisc-tools
1911    
1912    Minor bug fixes in the parser and rewriter
1913    
1914    ----------------------------------------------------------------------
1915    Name: Allen Leung
1916    Date: 2002/02/21 20:20:00 EST
1917    Tag: leunga-20020221-peephole
1918    
1919    Regenerated the peephole files.  Some contained typos in the specification
1920    and some didn't compile because of pretty printing bugs in the old version
1921    of 'nowhere'.
1922    
1923    ----------------------------------------------------------------------
1924    Name: Allen Leung
1925    Date: 2002/02/19 20:20:00 EST
1926    Tag: leunga-20020219-mlrisc-tools
1927    Description:
1928    
1929       Minor bug fixes to the mlrisc-tools library:
1930    
1931       1.  Fixed up parsing colon suffixed keywords
1932       2.  Added the ability to shut the error messages up
1933       3.  Reimplemented the pretty printer and fixed up/improved
1934           the pretty printing of handle and -> types.
1935       4.  Fixed up generation of literal symbols in the nowhere tool.
1936       5.  Added some SML keywords to to sml.sty
1937    
1938    ----------------------------------------------------------------------
1939    Name: Matthias Blume
1940    Date: 2002/02/19 16:20:00 EST
1941    Tag: blume-20020219-cmffi
1942    Description:
1943    
1944    A wild mix of changes, some minor, some major:
1945    
1946    * All C FFI-related libraries are now anchored under $c:
1947        $/c.cm      --> $c/c.cm
1948        $/c-int.cm  --> $c/internals/c-int.cm
1949        $/memory.cm --> $c/memory/memory.cm
1950    
1951    * "make" tool (in CM) now treats its argument pathname slightly
1952      differently:
1953        1. If the native expansion is an absolute name, then before invoking
1954           the "make" command on it, CM will apply OS.Path.mkRelative
1955           (with relativeTo = OS.FileSys.getDir()) to it.
1956        2. The argument will be passed through to subsequent phases of CM
1957           processing without "going native".  In particular, if the argument
1958           was an anchored path, then "make" will not lose track of that anchor.
1959    
1960    * Compiler backends now "know" their respective C calling conventions
1961      instead of having to be told about it by ml-nlffigen.  This relieves
1962      ml-nlffigen from one of its burdens.
1963    
1964    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
1965    
1966    * Export C_DEBUG and C_Debug from $c/c.cm.
1967    
1968    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
1969      subtyping relationship between incomplete pointers and their complete
1970      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
1971      with the convention of instantiating 'o with ('t, 'c) obj whenever
1972      the pointer target type is complete.  In the incomplete case, 'o
1973      will be instantiated with some "'c iobj" -- a type obtained by
1974      using one of the functors PointerToIncompleteType or PointerToCompleteType.
1975    
1976      Operations that work on both incomplete and complete pointer types are
1977      typed as taking an 'o ptr while operations that require the target to
1978      be known are typed as taking some ('t, 'c) obj ptr.
1979    
1980      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
1981      where void is an eqtype without any values.  This makes it possible
1982      to work on voidptr values using functions meant to operate on light
1983      incomplete pointers.
1984    
1985    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
1986      been vastly simplified.
1987    
1988    ----------------------------------------------------------------------
1989    Name: Matthias Blume
1990    Date: 2002/02/19 10:48:00 EST
1991    Tag: blume-20020219-pqfix
1992    Description:
1993    
1994    Applied Chris Okasaki's bug fix for priority queues.
1995    
1996    ----------------------------------------------------------------------
1997    Name: Matthias Blume
1998    Date: 2002/02/15 17:05:00
1999    Tag: Release_110_39
2000    Description:
2001    
2002    Last-minute retagging is becoming a tradition... :-(
2003    
2004    This is the working release 110.39.
2005    
2006    ----------------------------------------------------------------------
2007    Name: Matthias Blume
2008    Date: 2002/02/15 16:00:00 EST
2009    Tag: Release_110_39-orig
2010    Description:
2011    
2012    Working release 110.39.  New bootfiles.
2013    
2014    (Update: There was a small bug in the installer so it wouldn't work
2015    with all shells.  So I retagged. -Matthias)
2016    
2017    ----------------------------------------------------------------------
2018    Name: Matthias Blume
2019    Date: 2002/02/15 14:17:00 EST
2020    Tag: blume-20020215-showbindings
2021    Description:
2022    
2023    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
2024    the latter can be useful for exploring what bindings are available at
2025    the interactive prompt.  (The first function returns only the list
2026    of symbols that are really bound, the second prints those but also the
2027    ones that CM's autoloading mechanism knows about.)
2028    
2029    ----------------------------------------------------------------------
2030    Name: Matthias Blume
2031    Date: 2002/02/15 12:08:00 EST
2032    Tag: blume-20020215-iptrs
2033    Description:
2034    
2035    Two improvements to ml-nlffigen:
2036    
2037      1. Write files only if they do not exist or if their current contents
2038         do not coincide with what's being written.  (That is, avoid messing
2039         with the time stamps unless absolutely necessary.)
2040    
2041      2. Implement a "repository" mechanism for generated files related
2042         to "incomplete pointer types".   See the README file for details.
2043    
2044    ----------------------------------------------------------------------
2045    Name: Matthias Blume
2046    Date: 2002/02/14 11:50:00 EST
2047    Tag: blume-20020214-quote
2048    Description:
2049    
2050    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
2051    because of the new and improved tag generation scheme.  (Thanks to Allen
2052    Leung for pointing it out.)
2053    
2054    ----------------------------------------------------------------------
2055    Name: Lal George
2056    Date: 2002/02/14 09:55:27 EST 2002
2057    Tag: george-20020214-isabelle-bug
2058    Description:
2059    
2060    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2061    of Isabelle on the x86.
2062    
2063    From Allen:
2064    -----------
2065     I've found the problem:
2066    
2067         in ra-core.sml, I use the counter "blocked" to keep track of the
2068         true number of elements in the freeze queue.  When the counter goes
2069         to zero, I skip examining the queue.  But I've messed up the
2070         bookkeeping in combine():
2071    
2072             else ();
2073             case !ucol of
2074               PSEUDO => (if !cntv > 0 then
2075                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2076                                        ^^^^^^^^^^^^^^^^^^^^^^^
2077                      moveu := mergeMoveList(!movev, !moveu)
2078                     )
2079                  else ();
2080    
2081         combine() is called to coalesce two nodes u and v.
2082         I think I was thinking that if the move counts of u and v are both
2083         greater than zero then after they are coalesced then one node is
2084         removed from the freeze queue.  Apparently I was thinking that
2085         both u and v are of low degree, but that's clearly not necessarily true.
2086    
2087    
2088    02/12/2002:
2089        Here's the patch.  HOL now compiles.
2090    
2091        I don't know how this impact on performance (compile
2092        time or runtime).  This bug caused the RA (especially on the x86)
2093        to go thru the potential spill phase when there are still nodes on the
2094        freeze queue.
2095    
2096    
2097    
2098    
2099    ----------------------------------------------------------------------
2100    Name: Matthias Blume
2101    Date: 2002/02/13 22:40:00 EST
2102    Tag: blume-20020213-fptr-rtti
2103    Description:
2104    
2105    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2106    updates.
2107    
2108    ----------------------------------------------------------------------
2109    Name: Matthias Blume
2110    Date: 2002/02/13 16:41:00 EST
2111    Tag: blume-20020213-cmlpq
2112    Description:
2113    
2114    Added new priority queue export symbols (which have just been added to
2115    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2116    not compile and the installer would choke.)
2117    
2118    ----------------------------------------------------------------------
2119    Name: Matthias Blume
2120    Date: 2002/02/13 16:15:00 EST
2121    Tag: blume-20020213-various
2122    Description:
2123    
2124    1. More tweaks to ml-nlffigen:
2125    
2126       - better internal datastructures (resulting in slight speedup)
2127       - "-match" option requires exact match
2128       - "localized" gensym counters (untagged structs/unions nested within
2129         other structs/unions or within typedefs get a fresh counter; their
2130         tag will be prefixed by a concatenation of their parents' tags)
2131       - bug fixes (related to calculation of transitive closure of types
2132         to be included in the output)
2133    
2134    2. Minor Basis updates:
2135    
2136       - added implementations for List.collate and Option.app
2137    
2138    ----------------------------------------------------------------------
2139    Name: Matthias Blume
2140    Date: 2002/02/11 15:55:00 EST
2141    Tag: blume-20020211-gensym
2142    Description:
2143    
2144    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2145    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2146    names (ML structure names that correspond to unnamed C structs, unions,
2147    and enums), so that separate runs of ml-nlffigen do not clash.
2148    
2149    ----------------------------------------------------------------------
2150    Name: Matthias Blume
2151    Date: 2002/02/11 12:05:00 EST
2152    Tag: blume-20020211-gensml
2153    Description:
2154    
2155    A quick fix for a problem with GenSML (in the pgraph-util library):
2156    Make generation of toplevel "local" optional.  (Strictly speaking,
2157    signature definitions within "local" are not legal SML.)
2158    
2159    Other than that: updates to INSTALL and cm/TODO.
2160    
2161    ----------------------------------------------------------------------
2162    Name: Matthias Blume
2163    Date: 2002/02/08 15:00:00 EST
2164    Tag: blume-20020208-uniquepid
2165    Description:
2166    
2167    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2168    
2169    1. The installer (config/install.sh) has gotten smarter:
2170    
2171         - Configuration options are a bit easier to specify now
2172           (in config/targets).
2173         - Bug in recognizing .tar.bz2 files fixed.
2174         - Installer automatically resolves dependencies between
2175           configuration options (e.g., if you ask for eXene, you will
2176           also get cml -- regardless whether you asked for it or not).
2177         - Installer can run in "quieter mode" by setting the environment
2178           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2179           "completely silent", though.
2180         - Build HashCons library as part of smlnj-lib.
2181    
2182    2. A new scheme for assigning persistent identifiers to compilation
2183       units (and, by extension, to types etc.) has been put into place.
2184       This fixes a long-standing bug where types and even dynamic values
2185       can get internally confused, thereby compromising type safety
2186       (abstraction) and dynamic correctness.  See
2187    
2188         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2189    
2190       for an example of how things could go wrong until now.
2191    
2192       The downside of the new scheme is that pids are not quite as
2193       persistent as they used to be: CM will generate a fresh pid
2194       for every compilation unit that it thinks it sees for the first
2195       time.  That means that if you compile starting from a clean, fresh
2196       source tree at two different times, you end up with different
2197       binaries.
2198    
2199       Cutoff recompilation, however, has not been compromised because
2200       CM keeps pid information in special caches between runs.
2201    
2202    ----------------------------------------------------------------------
2203    Name: Lal George
2204    Date: 2002/02/07 15:34:13 EST 2002
2205    Tag: <none>
2206    Description:
2207    
2208    Compilers that generate assembly code may produce  global labels
2209    whose value is resolved at link time. The various peephole optimization
2210    modules did not take this in account.
2211    
2212    TODO. The Labels.addrOf function should really return an option
2213    type so that clients are forced to deal with this issue, rather
2214    than an exception being raised.
2215    
2216    ----------------------------------------------------------------------
2217    Name: Lal George
2218    Date: 2002/02/06 13:55:02 EST
2219    Tag: george-20020206-ra-breakup
2220    Description:
2221    
2222    1. A bug fix from Allen.
2223    
2224        A typo causes extra fstp %st(0)'s to be generated at compensation
2225        edges, which might cause stack underflow traps at runtime.  This
2226        occurs in fft where there are extraneous fstps right before the 'into'
2227        trap instruction (in this case they are harmless since none of the
2228        integers overflow.)
2229    
2230    2. Pulled out various utility modules that were embedded in the modules
2231       of the register allocator. I need these modules for other purposes, but
2232       they are not complete enough to put into a library (just yet).
2233    ----------------------------------------------------------------------
2234    Name: Matthias Blume
2235    Date: 2002/01/31 16:05:00 EST
2236    Tag: blume-20020131-sparc-ccalls
2237    Description:
2238    
2239    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2240       of extra stack space by mistake.  Fixed.
2241    
2242    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2243    
2244    ----------------------------------------------------------------------
2245    Name: Allen Leung
2246    Date: 2002/01/30
2247    Tag: leunga-20020130-nowhere-bug-fix
2248    Description:
2249    
2250       MLRISC bug fixes:
2251       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2252       2. MachineInt.fromString was negating its value.
2253    
2254    ----------------------------------------------------------------------
2255    Name: Matthias Blume
2256    Date: 2002/01/29
2257    Tag: blume-20020129-INSTALL
2258    Description:
2259    
2260    - Added somewhat detailed installation instructions (file INSTALL).
2261    - Fixed curl-detection bug in config/install.sh.
2262    - It is now possible to select the URL getter using the URLGETTER
2263      environment variable:
2264    
2265          not set / "unknown"      --> automatic detection (script tries wget,
2266                                       curl, and lynx)
2267          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2268                                       how to properly invoke them)
2269          other                    --> use $URLGETTER directly, it must take
2270                                       precisely two command-line arguments
2271                                       (source URL and destination file name)
2272    
2273    ----------------------------------------------------------------------
2274    Name: Matthias Blume
2275    Date: 2002/01/28
2276    Tag: blume-20020128-sparc-ccalls
2277    Description:
2278    
2279    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2280    - Make use of the allocParam argument in sparc-c-calls.
2281    
2282    ----------------------------------------------------------------------
2283    Name: Matthias Blume
2284    Date: 2002/01/28
2285    Tag: blume-20020128-allocParam
2286    Description:
2287    
2288    John Reppy:  Changes c-calls API to accept client-callback for
2289    allocating extra stack space.
2290    me: Corresponding changes to mlriscGen (using a dummy argument that
2291        does not change the current behavior).
2292    
2293    ----------------------------------------------------------------------
2294    Name: Matthias Blume
2295    Date: 2002/01/28 12:00:00
2296    Tag: Release_110_38
2297    Description:
2298    
2299    This time for real!!!
2300    
2301    ----------------------------------------------------------------------
2302    Name: Matthias Blume
2303    Date: 2002/01/28 10:56:00 EST
2304    Tag: blume-20020128-retraction
2305    Description:
2306    
2307    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
2308       with blume-Release_110_38-retracted.)
2309    
2310    1. Fixed a problem with incorrect rounding modes in real64.sml.
2311       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2312    
2313    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2314       and enums fixed.  The naming of corresponding ML identifiers should
2315       now be consistent again.
2316    
2317    ----------------------------------------------------------------------
2318    Name: Allen Leung
2319    Date: 2002/01/27
2320    Tag: leunga-20020127-nowhere
2321    Description:
2322    
2323       Added a target called nowhere in the configuration scripts.
2324       Enabling this will build the MLRISC 'nowhere' tool (for translating
2325       programs with where-clauses into legal SML code) during installation.
2326    
2327    ----------------------------------------------------------------------
2328    Name: Matthias Blume
2329    Date: 2002/01/25 21:27:00 EST
2330    Tag: blume-Release_110_38-retracted
2331    Description:
2332    
2333    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
2334    
2335    README will be added later.
2336    
2337    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
2338    (Release_110_38) removed.  Reason: Last-minute bug fixes.
2339    
2340    ----------------------------------------------------------------------
2341    Name: Matthias Blume
2342    Date: 2002/01/25
2343    Tag: blume-20020125-ffi
2344    Description:
2345    
2346    A large number of tweaks and improvements to ml-nlffi-lib and
2347    ml-nlffigen:
2348    
2349       - ML represenation types have been streamlined
2350       - getter and setter functions work with concrete values, not abstract
2351         ones where possible
2352       - ml-nlffigen command line more flexible (see README file there)
2353       - some bugs have been fixed (hopefully)
2354    
2355    ----------------------------------------------------------------------
2356    Name: Lal George
2357    Date: 2002/01/24
2358    Tag: george-20020124-risc-ra-interface
2359    Description:
2360    
2361       There is a dramatic simplification in the interface to the
2362       register allocator for RISC architectures as a result of making
2363       parallel copy instructions explicit.
2364    
2365    ----------------------------------------------------------------------
2366    Name: Matthias Blume
2367    Date: 2002/01/22
2368    Tag: blume-20020122-x86-ccalls
2369    Description:
2370    
2371    Bug fix for c-calls on x86 (having to do with how char- and
2372    short-arguments are being handled).
2373    
2374    ----------------------------------------------------------------------
2375    Name: Matthias Blume
2376    Date: 2002/01/21
2377    Tag: blume-20020121-ff
2378    Description:
2379    
2380    Another day of fiddling with the FFI...
2381    
2382    1. Bug fix/workaround:  CKIT does not complain about negative array
2383       dimensions, so ml-nlffigen has to guard itself against this possibility.
2384       (Otherwise a negative dimension would send it into an infinite loop.)
2385    
2386    2. Some of the abstract types (light objects, light pointers, most "base"
2387       types) in structure C are now eqtypes.
2388    
2389    3. Added constructors and test functions for NULL function pointers.
2390    
2391    ----------------------------------------------------------------------
2392    Name: Matthias Blume
2393    Date: 2002/01/18
2394    Tag: blume-20020118-ready-for-new-release
2395    Description:
2396    
2397    Made config/srcarchiveurl point to a new place.  (Will provide boot
2398    files shortly.)
2399    
2400    Maybe we christen this to be 110.38?
2401    
2402    ----------------------------------------------------------------------
2403    Name: Matthias Blume
2404    Date: 2002/01/18
2405    Tag: blume-20020118-more-ffifiddle
2406    Description:
2407    
2408    Today's FFI fiddling:
2409    
2410      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
2411        C.Set.* that use concrete (MLRep.*) arguments and results instead
2412        of abstract ones.
2413    
2414      - Provided word-style bit operations etc. for "int" representation
2415        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
2416        and Long.
2417    
2418    ----------------------------------------------------------------------
2419    Name: Matthias Blume
2420    Date: 2002/01/18
2421    Tag: blume-20020118-use-x86-fp
2422    Description:
2423    
2424    Now that x86-fast-fp seems to be working, I turned it back on again
2425    by default.  (Seems to work fine now, even with the FFI.)
2426    
2427    Other than that, I added some documentation about the FFI to
2428    src/ml-nlffigen/README and updated the FFI test examples in
2429    src/ml-nlffi-lib/Tests/*.
2430    
2431    ----------------------------------------------------------------------
2432    Name: Allen Leung
2433    Date: 2002/01/17
2434    Tag: leunga-20020117-x86-fast-fp-call
2435    Description:
2436    
2437       1. Fixed a problem with handling return fp values when x86's fast fp
2438          mode is turned on.
2439    
2440       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
2441          of %f32.
2442    
2443       3. Added a constructor INT32lit to the ast of MLRISC tools.
2444    
2445    ----------------------------------------------------------------------
2446    Name: Matthias Blume
2447    Date: 2002/01/16
2448    Tag: blume-20020116-ffifiddle
2449    Description:
2450    
2451    More fiddling with the FFI interface:
2452    
2453     - Make constness 'c instead of rw wherever possible.  This eliminates
2454       the need for certain explicit coercions.  (However, due to ML's
2455       value polymorphism, there will still be many cases where explicit
2456       coercions are necessary.  Phantom types are not the whole answer
2457       to modeling a subtyping relationship in ML.)
2458    
2459     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
2460    
2461     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
2462       clumsy workaround for varargs problem.)
2463    
2464    ----------------------------------------------------------------------
2465    Name: Lal George
2466    Date: 2002/01/15
2467    Tag: <none>
2468    Description:
2469    
2470    1. Since COPY instructions are no longer native to the architecture,
2471       a generic functor can be used to implement the expandCopies function.
2472    
2473    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
2474       TEXT segment.
2475    
2476    ----------------------------------------------------------------------
2477    Name: Matthias Blume
2478    Date: 2002/01/15
2479    Tag: blume-20020115-ffiupdates
2480    Description:
2481    
2482    1. Fix for bug resulting in single-precision float values being returned
2483       incorrectly from FFI calls.
2484    
2485    2. Small modifications to C FFI API:
2486    
2487        - memory-allocation routines return straight objects (no options)
2488          and raise an exception in out-of-memory situations
2489        - unsafe extensions to cast between function pointers and pointers
2490          from/to ints
2491        - added structure C_Debug as an alternative to structure C where
2492          pointer-dereferencing (|*| and |*!) always check for null-pointers
2493        - added open_lib' to DynLinkage;  open_lib' works like open_lib
2494          but also takes a (possibly empty) list of existing library handles
2495          that the current library depends on
2496    
2497    ----------------------------------------------------------------------
2498    Name: Matthias Blume
2499    Date: 2002/01/10
2500    Tag: blume-20020110-newffigen
2501    Description:
2502    
2503    1. Updates to portable graph code.
2504    
2505    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
2506       more scalable now so that even huge interfaces such as the one
2507       for GTK compile in finite time and space. :-)
2508       See src/ml-nlffigen/README for details on what's new.
2509    
2510    ----------------------------------------------------------------------
2511    Name: Lal George
2512    Date: 2001/01/09 14:31:35 EST 2002
2513    Tag: george-20011206-rm-native-copy
2514    Description:
2515    
2516            Removed the native COPY and FCOPY instructions
2517            from all the architectures and replaced it with the
2518            explicit COPY instruction from the previous commit.
2519    
2520            It is now possible to simplify many of the optimizations
2521            modules that manipulate copies. This has not been
2522            done in this change.
2523    
2524    ----------------------------------------------------------------------
2525    Name: Lal George
2526    Date: 2001/12/06 16:50:13 EST 2001
2527    Tag: george-20011206-mlrisc-instruction
2528    Description:
2529    
2530    Changed the representation of instructions from being fully abstract
2531    to being partially concrete. That is to say:
2532    
2533      from
2534            type instruction
2535    
2536      to
2537            type instr                              (* machine instruction *)
2538    
2539            datatype instruction =
2540                LIVE of {regs: C.cellset, spilled: C.cellset}
2541              | KILL of {regs: C.cellset, spilled: C.cellset}
2542              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
2543              | ANNOTATION of {i: instruction, a: Annotations.annotation}
2544              | INSTR of instr
2545    
2546    This makes the handling of certain special instructions that appear on
2547    all architectures easier and uniform.
2548    
2549    LIVE and KILL say that a list of registers are live or killed at the
2550    program point where they appear. No spill code is generated when an
2551    element of the 'regs' field is spilled, but the register is moved to
2552    the 'spilled' (which is present, more for debugging than anything else).
2553    
2554    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
2555    We used to generate:
2556    
2557            DEFFREG f1
2558            f1 := f2 + f3
2559            trapb
2560    
2561    but now generate:
2562    
2563            f1 := f2 + f3
2564            trapb
2565            LIVE {regs=[f1,f2,f3], spilled=[]}
2566    
2567    Furthermore, the DEFFREG (hack) required that all floating point instruction
2568    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
2569    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
2570    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
2571    this architecture).
2572    
2573    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
2574    all the architectures. This will result in further simplification of the
2575    register allocator that must be aware of them for coalescing purposes, and
2576    will also simplify certain aspects of the machine description that provides
2577    callbacks related to parallel copies.
2578    
2579    ANNOTATION should be obvious, and now INSTR represents the honest to God
2580    machine instruction set!
2581    
2582    The <arch>/instructions/<arch>Instr.sml files define certain utility
2583    functions for making porting easier -- essentially converting upper case
2584    to lower case. All machine instructions (of type instr) are in upper case,
2585    and the lower case form generates an MLRISC instruction. For example on
2586    the alpha we have:
2587    
2588      datatype instr =
2589         LDA of {r:cell, b:cell, d:operand}
2590       | ...
2591    
2592      val lda : {r:cell, b:cell, d:operand} -> instruction
2593        ...
2594    
2595    where lda is just (INSTR o LDA), etc.
2596    
2597    ----------------------------------------------------------------------
2598    Name: Matthias Blume
2599    Date: 2001/11/22 21:40:00 EST
2600    Tag: Release_110_37
2601    Description:
2602    
2603    Release 110.37.  This time for real.
2604    
2605    ----------------------------------------------------------------------
2606    Name: Matthias Blume
2607    Date: 2001/11/21 16:35:00 EST
2608    Tag: blume-20011121-foot-in-mouth
2609    Description:
2610    
2611    Removed the "Release_110_37" tag because of a serious bug.
2612    This will be re-tagged once the bug is fixed.
2613    
2614    ----------------------------------------------------------------------
2615    Name: Matthias Blume
2616    Date: 2001/11/21 16:14:00 EST
2617    Tag: blume-20011121-forgottenfile
2618    Description:
2619    
2620    Forgot to add a file.  (Just a .tex-file -- part of
2621    the CM manual source.)
2622    
2623    ----------------------------------------------------------------------
2624    Name: Matthias Blume
2625    Date: 2001/11/21 16:10:00 EST
2626    Tag: blume-20011121-invalid_110_37
2627    Description:
2628    
2629    Note: I removed the original tag "Release_110_37" from this commit
2630          because we found a serious bug in all non-x86 backends.
2631          - Matthias
2632    
2633    1. Modifications to the SML/NJ code generator and to the runtime system
2634       so that code object name strings are directly inserted into code
2635       objects at code generation time.  The only business the runtime system
2636       has with this is now to read the name strings on occasions.
2637       (The encoding of the name string has also changed somewhat.)
2638    
2639    2. CM now implements a simple "set calculus" for specifying export lists.
2640       In particular, it is now possible to refer to the export lists of
2641       other libraries/groups/sources and form unions as well as differences.
2642       See the latest CM manual for details.
2643    
2644    3. An separate notion of "proxy" libraries has again be eliminated from
2645       CM's model.  (Proxy libraries are now simply a special case of using
2646       the export list calculus.)
2647    
2648    4. Some of the existing libraries now take advantage of the new set
2649       calculus.
2650       (Notice that not all libraries have been converted because some
2651       of the existing .cm-files are supposed to be backward compatible
2652       with 110.0.x.)
2653    
2654    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
2655       "General.exnMessage"!  The former relies on a certain hook to be
2656       initialized, and that often does not happen in the stand-alone case.)
2657    
2658    ----------------------------------------------------------------------
2659    Name: Lal George
2660    Date: 2001/11/21  13:56:18 EST
2661    Tag: george-2001121-pseudo-ops
2662    Description:
2663    
2664      Implemented a complete redesign of MLRISC pseudo-ops. Now there
2665      ought to never be any question of incompatabilities with
2666      pseudo-op syntax expected by host assemblers.
2667    
2668      For now, only modules supporting GAS syntax are implemented
2669      but more should follow, such as MASM, and vendor assembler
2670      syntax, e.g. IBM as, Sun as, etc.
2671    
2672    ----------------------------------------------------------------------
2673    Name: Matthias Blume
2674    Date: 2001/11/14 11:52:00 EST
2675    Tag: blume-20011114-srcname
2676    Description:
2677    
2678    1. Routed the name of the current source file to mlriscgen where it
2679       should be directly emitted into the code object.  (This last part
2680       is yet to be done.)
2681    
2682    2. Some cleanup of the pgraph code to make it match the proposal that
2683       I put out the other day.  (The proposal notwithstanding, things are
2684       still in flux here.)
2685    
2686    ----------------------------------------------------------------------
2687    Name: Lal George
2688    Date: 2001/11/14 09:44:04 EST
2689    Tag:
2690    Description:
2691    
2692      Fix for a backpatching bug reported by Allen.
2693    
2694      Because the boundary between short and long span-dependent
2695      instructions is +/- 128, there are an astounding number of
2696      span-dependent instructions whose size is over estimated.
2697    
2698      Allen came up with the idea of letting the size of span
2699      dependent instructions be non-monotonic, for a maxIter
2700      number of times, after which the size must be monotonically
2701      increasing.
2702    
2703      This table shows the number of span-dependent instructions
2704      whose size was over-estimated as a function of maxIter, for the
2705      file Parse/parse/ml.grm.sml:
2706    
2707         maxIter            # of instructions:
2708            10                      687
2709            20                      438
2710            30                      198
2711            40                        0
2712    
2713      In compiling the compiler, there is no significant difference in
2714      compilation speed between maxIter=10 and maxIter=40. Actually,
2715      my measurements showed that maxIter=40 was a tad faster than
2716      maxIter=10! Also 96% of the  files in the compiler reach a fix
2717      point within 13 iterations, so fixing maxIter at 40, while high,
2718      is okay.
2719    
2720    ----------------------------------------------------------------------
2721    Name: Matthias Blume
2722    Date: 2001/10/31 15:25:00 EST
2723    Tag: blume-20011031-pgraph
2724    Description:
2725    
2726    CKIT:
2727    * Changed the "Function" constructor of type Ast.ctype to carry optional
2728      argument identifiers.
2729    * Changed the return type of TypeUtil.getFunction accordingly.
2730    * Type equality ignores the argument names.
2731    * TypeUtil.composite tries to preserve argument names but gives up quickly
2732      if there is a mismatch.
2733    
2734    installation script:
2735    * attempts to use "curl" if available (unless "wget" is available as well)
2736    
2737    CM:
2738    * has an experimental implementation of "portable graphs" which I will
2739      soon propose as an implementation-independent library format
2740    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
2741    
2742    NLFFI-LIB:
2743    * some cleanup (all cosmetic)
2744    
2745    NLFFIGEN:
2746    * temporarily disabled the mechanism that suppresses ML output for
2747      C definitions whose identifiers start with an underscore character
2748    * generate val bindings for enum constants
2749    * user can request that only one style (light or heavy) is being used;
2750      default is to use both (command-line arguments: -heavy and -light)
2751    * fixed bug in handling of function types involving incomplete pointers
2752    * generate ML entry points that take record arguments (i.e., using
2753      named arguments) for C functions that have a prototype with named
2754      arguments
2755      (see changes to CKIT)
2756    
2757    ----------------------------------------------------------------------
2758    Name: Allen Leung
2759    Date: 2001/10/27 20:34:00 EDT
2760    Tag: leunga-20011027-x86-fast-fp-call
2761    Description:
2762    
2763       Fixed the bug described in blume-20010920-slowfp.
2764    
2765       The fix involves
2766          1. generating FCOPYs in FSTP in ia32-svid
2767          2. marking a CALL with the appropriate annotation
2768    
2769    ----------------------------------------------------------------------
2770    Name: Matthias Blume
2771    Date: 2001/10/16 11:32:00 EDT
2772    Tag: blume-20011016-netbsd
2773    Description:
2774    
2775    Underscore patch from Chris Richards (fixing problem with compiling
2776    runtime system under recent NetBSD).
2777    
2778    ----------------------------------------------------------------------
2779    Name: Allen Leung
2780    Date: 2001/10/12 17:18:32 EDT 2001
2781    Tag: leung-20011012-x86-printflowgraph
2782    Description:
2783    
2784    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
2785    
2786    ----------------------------------------------------------------------
2787    Name: Lal George
2788    Date: 2001/10/11 23:51:34 EDT
2789    Tag: george-20011011-too-many-instrs
2790    Description:
2791    
2792    The representation of a program point never expected to see more
2793    than 65536 instructions in a basic block!
2794    
2795    ----------------------------------------------------------------------
2796    Name: Lal George
2797    Date: 2001/10/09 09:41:37 EDT
2798    Tag: george-20011008-mlrisc-labels
2799    Description:
2800    
2801    Changed the machine description files to support printing of
2802    local and global labels in assembly code, based on host assembler
2803    conventions.
2804    
2805    ----------------------------------------------------------------------
2806    Name: Matthias Blume
2807    Date: 2001/09/25 15:25:00 EDT
2808    Tag: blume-20010925-exninfo
2809    Description:
2810    
2811    I provided a non-hook implementation of exnName (at the toplevel) and
2812    made the "dummy" implementation of exnMessage (at the toplevel) more
2813    useful: if nothing gets "hooked in", then at least you are going to
2814    see the exception name and a message indicating why you don't see more.
2815    
2816    [For the time being, programs that need exnMessage and want to use
2817    ml-build should either use General.exnMessage (strongly recommended) or
2818    refer to structure General at some other point so that CM sees a
2819    static dependency.]
2820    
2821    [Similar remarks go for "print" and "use":  If you want to use their
2822    functionality in stand-alone programs generated by ml-build, then use
2823    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
2824    
2825    ----------------------------------------------------------------------
2826    Name: Matthias Blume
2827    Date: 2001/09/20 17:28:00 EDT
2828    Tag: blume-20010920-slowfp
2829    Description:
2830    
2831    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
2832    
2833    ----------------------------------------------------------------------
2834    Name: Matthias Blume
2835    Date: 2001/09/20 17:20:00 EDT
2836    Tag: blume-20010920-canonicalpaths
2837    Description:
2838    
2839    0. Updated the BOOT file (something that I forgot to do earlier).
2840    
2841    1. Small internal change to CM so that it avoids "/../" in filenames
2842       as much as possible (but only where it is safe).
2843    
2844    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
2845       that arguments that contain delimiters are passed through correctly.
2846       This change also means that all "special" arguments of the form
2847       @SMLxxx... must come first.
2848    
2849    3. Changed install script to put relative anchor names for tool commands
2850       into pathconfig.
2851    
2852    ----------------------------------------------------------------------
2853    Name: Matthias Blume
2854    Date: 2001/09/18 15:35:00 EDT
2855    Tag: blume-20010918-readme11036
2856    Description:
2857    
2858    Added README files.
2859    
2860    ----------------------------------------------------------------------
2861    Name: Matthias Blume
2862    Date: 2001/09/18 11:45:00 EDT
2863    Tag: Release_110_36 (retag)
2864    Description:
2865    
2866    Fixed mistake in config/preloads. Retagged as 110.36.
2867    
2868    ----------------------------------------------------------------------
2869    Name: Matthias Blume
2870    Date: 2001/09/18 09:40:00 EDT
2871    Tag: Release_110_36_orig (tag changed)
2872    Description:
2873    
2874    New version (110.36).  New bootfiles.
2875    
2876    ----------------------------------------------------------------------
2877    Name: Matthias Blume
2878    Date: 2001/09/14 16:15:00 EDT
2879    Tag: blume-20010914-x86fastfp
2880    Description:
2881    
2882    John committed some changes that Allen made, in particular a (hopefully)
2883    correctly working version of the x86-fp module.
2884    
2885    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
2886    flag to "true".  Everything seems to compile to a fixpoint ok, and
2887    "mandelbrot" speeds up by about 15%.
2888    
2889    ----------------------------------------------------------------------
2890    Name: Matthias Blume
2891    Date: 2001/09/13 11:20:00 EDT
2892    Tag: blume-20010913-minimal
2893    Description:
2894    
2895    1. Stefan Monnier's patch to fix a miscompilation problem that
2896       was brought to light by John Reppy's work on Moby.
2897    
2898    2. Implemented a minimal "structure Compiler" that contains just
2899       "version" and "architecture".  The minimal version will be
2900       available when the full version is not.  This is for backward-
2901       compatibility with code that wants to test Compiler.version.
2902    
2903    ----------------------------------------------------------------------
2904    Name: Matthias Blume
2905    Date: 2001/08/28 14:03:00 EDT
2906    Tag: blume-20010828-ml-lex
2907    Description:
2908    
2909    Fix for bug 1581, received from Neophytos Michael.
2910    
2911    ----------------------------------------------------------------------
2912    Name: Matthias Blume
2913    Date: 2001/08/27 11:20:00 EDT
2914    Tag: blume-20010827-readme11035
2915    Description:
2916    
2917    Fleshed out the README file for 110.35.
2918    
2919    ----------------------------------------------------------------------
2920    Name: Matthias Blume
2921    Date: 2001/08/24 17:10:00 EDT
2922    Tag: Release_110_35
2923    Description:
2924    
2925    New version number (110.35).  New bootfiles.
2926    
2927    ----------------------------------------------------------------------
2928    Name: Lal George
2929    Date: 2001/08/24 13:47:18 EDT 2001
2930    Tag: george-20010824-MLRISC-graphs
2931    Description:
2932    
2933     removed clusters from MLRISC completely and replaced with graphs.
2934    
2935    ----------------------------------------------------------------------
2936    Name: Matthias Blume
2937    Date: 2001/08/23 17:50:00 EDT
2938    Tag: blume-20010823-toplevel
2939    Description:
2940    
2941    - some reorganization of the code that implements various kinds of
2942      environments in the compiler (static, dynamic, symbolic, combined)
2943    - re-implemented the EnvRef module so that evalStream works properly
2944      (if the stream contains references to "use", "CM.make", etc.)
2945    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
2946    
2947    ----------------------------------------------------------------------
2948    Name: Matthias Blume
2949    Date: 2001/08/20 15:50 EDT
2950    Tag: blume20010820-slipup
2951    Description:
2952    
2953    I forgot to commit a few files.  Here they are...
2954    
2955    ----------------------------------------------------------------------
2956    Name: Matthias Blume
2957    Date: 2001/08/20 15:35:00 EDT
2958    Tag: blume-20010820-debugprof
2959    Description:
2960    
2961    !!!! NEW BOOTFILES !!!!
2962    
2963    This is another round of reorganizing the compiler sources.  This
2964    time the main goal was to factor out all the "instrumentation"
2965    passes (for profiling and backtracing) into their own library.
2966    The difficulty was to do it in such a way that it does not depend
2967    on elaborate.cm but only on elabdata.cm.
2968    
2969    Therefore there have been further changes to both elaborate.cm and
2970    elabdata.cm -- more "generic" things have been moved from the former
2971    to the latter.  As a result, I was forced to split the assignment
2972    of numbers indicating "primtyc"s into two portions: SML-generic and
2973    SML/NJ-specific.  Since it would have been awkward to maintain,
2974    I bit the bullet and actually _changed_ the mapping between these
2975    numbers and primtycs.  The bottom line of this is that you need
2976    a new set of bin- and bootfiles.
2977    
2978    I have built new bootfiles for all architectures, so doing a fresh
2979    checkout and config/install.sh should be all you need.
2980    
2981    The newly created library's name is
2982    
2983        $smlnj/viscomp/debugprof.cm
2984    
2985    and its sources live under
2986    
2987        src/compiler/DebugProf
2988    
2989    ----------------------------------------------------------------------
2990    Name: Matthias Blume
2991    Date: 2001/08/15 17:15:00 EDT
2992    Tag: blume-20010815-compreorg
2993    Description:
2994    
2995    This is a first cut at reorganizing the CM libraries that make up the
2996    core of the compiler.  The idea is to separate out pieces that could
2997    be used independently by tools, e.g., the parser, the typechecker, etc.
2998    
2999    The current status is a step in this direction, but it is not quite
3000    satisfactory yet.  Expect more changes in the future.
3001    
3002    Here is the current (new) organization...
3003    
3004        What used to be $smlnj/viscomp/core.cm is now divided into
3005        six CM libraries:
3006    
3007             $smlnj/viscomp/basics.cm
3008                           /parser.cm
3009                           /elabdata.cm
3010                           /elaborate.cm
3011                           /execute.cm
3012                           /core.cm
3013    
3014        The CM files for these libraries live under src/system/smlnj/viscomp.
3015        All these libraries are proxy libraries that contain precisely
3016        one CM library component.  Here are the locations of the components
3017        (all within the src/compiler tree):
3018    
3019             Basics/basics.cm
3020             Parse/parser.cm
3021             ElabData/elabdata.cm
3022             Elaborator/elaborate.cm
3023             Execution/execute.cm
3024             core.cm
3025    
3026         [This organization is the same that has been used already
3027         for a while for the architecture-specific parts of the visible
3028         compiler and for the old version of core.cm.]
3029    
3030         As you will notice, many source files have been moved from their
3031         respective original locations to a new home in one of the above
3032         subtrees.
3033    
3034         The division of labor between the new libraries is the following:
3035    
3036             basics.cm:
3037                - Simple, basic definitions that pertain to many (or all) of
3038                  the other libraries.
3039             parser.cm:
3040                - The SML parser, producing output of type Ast.dec.
3041                - The type family for Ast is also defined and exported here.
3042             elabdata.cm:
3043                - The datatypes that describe input and output of the elaborator.
3044                  This includes types, absyn, and static environments.
3045             elaborator.cm:
3046                - The SML/NJ type checker and elaborator.
3047                  This maps an Ast.dec (with a given static environment) to
3048                  an Absyn.dec (with a new static environment).
3049                - This libraries implements certain modules that used to be
3050                  structures as functors (to remove dependencies on FLINT).
3051             execute.cm:
3052                - Everything having to do with executing binary code objects.
3053                - Dynamic environments.
3054             core.cm:
3055                - SML/NJ-specific instantiations of the elaborator and MLRISC.
3056                - Top-level modules.
3057                - FLINT (this should eventually become its own library)
3058    
3059    Notes:
3060    
3061    I am not 100% happy with the way I separated the elaborator (and its
3062    data structures) from FLINT.  Two instances of the same problem:
3063    
3064        1. Data structures contain certain fields that carry FLINT-specific
3065           information.  I hacked around this using exn and the property list
3066           module from smlnj-lib.  But the fact that there are middle-end
3067           specific fields around at all is a bit annoying.
3068    
3069        2. The elaborator calculates certain FLINT-related information.  I tried
3070           to make this as abstract as I could using functorization, but, again,
3071           the fact that the elaborator has to perform calculations on behalf
3072           of the middle-end at all is not nice.
3073    
3074        3. Having to used exn and property lists is unfortunate because it
3075           weakens type checking.  The other alternative (parameterizing
3076           nearly *everything*) is not appealing, though.
3077    
3078    I removed the "rebinding =" warning hack because due to the new organization
3079    it was awkward to maintain it.  As a result, the compiler now issues some of
3080    these warnings when compiling init.cmi during bootstrap compilation. On
3081    the plus side, you also get a warning when you do, for example:
3082       val op = = Int32.+
3083    which was not the case up to now.
3084    
3085    I placed "assign" and "deref" into the _Core structure so that the
3086    code that deals with the "lazy" keyword can find them there.  This
3087    removes the need for having access to the primitive environment
3088    during elaboration.
3089    
3090    ----------------------------------------------------------------------
3091    Name: Matthias Blume
3092    Date: 2001/08/13
3093    Tag: blume-20010813-closures
3094    Description:
3095    
3096    This fix was sent to us by Zhong Shao.  It is supposed to improve the
3097    performance of certain loops by avoiding needless closure allocation.
3098    
3099    ----------------------------------------------------------------------
3100    Name: Lal George
3101    Date: 2001/07/31 10:03:23 EDT 2001
3102    Tag: george-20010731-x86-fmalloc
3103    Description: Fixed bug in x86 calls
3104    
3105        There was a bug where call instructions would mysteriously
3106        vanish. The call instruction had to be one that returned
3107        a floating point value.
3108    
3109    ----------------------------------------------------------------------
3110    Name: Lal George
3111    Date: 2001/07/19 16:36:29 EDT 2001
3112    Tag: george-20010719-simple-cells
3113    Description:
3114    
3115    I have dramatically simplified the interface for CELLS in MLRISC.
3116    
3117    In summary, the cells interface is broken up into three parts:
3118    
3119      1. CellsBasis : CELLS_BASIS
3120    
3121            CellsBasis is a top level structure and common for all
3122            architectures.  it contains the definitions of basic datatypes
3123            and utility  functions over these types.
3124    
3125      2. functor Cells() : CELLS
3126    
3127            Cells generates an interface for CELLS that incorporates the
3128            specific resources on the target architecture, such as the
3129            presence of special register classes, their number and size,
3130            and various useful substructures.
3131    
3132      3. <ARCH>CELLS
3133    
3134            e.g. SparcCells: SPARCCELLS
3135    
3136            <ARCH>CELLS usually contains additional bindings for special
3137            registers  on the architecture, such as:
3138    
3139                    val r0 : cell           (* register zero *)
3140                    val y : cell            (* Y register *)
3141                    val psr : cell          (* processor status register *)
3142                    ...
3143    
3144            The structure returned by applying the Cells functor is opened
3145            in this interface.
3146    
3147    The main implication of all this is that the datatypes for cells is
3148    split between CellsBasis and CELLS -- a fairly simple change for user
3149    code.
3150    
3151    In the old scheme the CELLS interface had a definitional binding of
3152    the form:
3153    
3154            signature CELLS = sig
3155    
3156               structure CellsBasis = CellsBasis
3157    
3158               ...
3159    
3160            end
3161    
3162    With all the sharing constraints that goes on in MLRISC, this old
3163    design  quickly leads to errors such as:
3164    
3165            "structure definition spec inside of sharing ... "
3166    
3167    
3168    and appears to require an unacceptable amount of sharing and where
3169    constraint hackery.
3170    
3171    I think this error message (the interaction of definitional specs and
3172    sharing) requires more explanation on our web page.
3173    
3174    ----------------------------------------------------------------------
3175    Name: Matthias Blume
3176    Date: 2001/07/19 15:00:00 EDT
3177    Tag: blume-20010719-libreorg
3178    Description:
3179    
3180    This update puts together a fairly extensive but straightforward change
3181    to the way the libraries that implement the interactive system are
3182    organized:
3183    
3184       The biggest change is the elimination of structure Compiler.  As a
3185       replacement for this structure, there is now a CM library
3186       (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
3187       that exports all the substructures of the original structure Compiler
3188       directly.  So instead of saying Compiler.Foo.bar one now simply
3189       says Foo.bar.  (The CM libraries actually export a collection of
3190       structures that is richer than the collection of substructures of
3191       structure Compiler.)
3192    
3193       To make the transition smooth, there is a separate library called
3194       $smlnj/compiler/compiler.cm which puts together and exports the
3195       original structure Compiler (or at least something very close to it).
3196    
3197       There are five members of the original structure Compiler
3198       that are not exported directly but which instead became members
3199       of a new structure Backend (described by signature BACKEND).  These are:
3200       structure Profile (: PROFILE), structure Compile (: COMPILE), structure
3201       Interact (: INTERACT), structure Machine (: MACHINE), and val
3202       architecture (: string).
3203    
3204       Structure Compiler.Version has become structure CompilerVersion.
3205    
3206       Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
3207       by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
3208       or x86, respectively.
3209       Each of these exports the same frontend structures that
3210       $smlnj/compiler.cm exports.  But they do not have a structure Backend
3211       and instead export some structure <Arch>Backend where <Arch> is Alpha32,
3212       Hppa, PPC, Sparc, or X86, respectively.
3213    
3214       Library $smlnj/compiler/all.cm exports the union of the exports of
3215       $smlnj/compiler/<arch>.cm
3216    
3217       There are no structures <Arch>Compiler anymore, use
3218       $smlnj/compiler/<arch>.cm instead.
3219    
3220       Library host-compiler-0.cm is gone.  Instead, the internal library
3221       that instantiates CM is now called cm0.cm.  Selection of the host
3222       compiler (backend) is no longer done here but. (Responsibility for it
3223       now lies with $smlnj/compiler/current.cm.  This seems to be more
3224       logical.)
3225    
3226       Many individual files have been moved or renamed.  Some files have
3227       been split into multiple files, and some "dead" files have been deleted.
3228    
3229    Aside from these changes to library organization, there are also changes
3230    to the way the code itself is organized:
3231    
3232       Structure Binfile has been re-implemented in such a way that it no
3233       longer needs any knowledge of the compiler.  It exclusively deals
3234       with the details of binfile layout.  It no longer invokes the
3235       compiler (for the purpose of creating new prospective binfile
3236       content), and it no longer has any knowledge of how to interpret
3237       pickles.
3238    
3239       Structure Compile (: COMPILE) has been stripped down to the bare
3240       essentials of compilation.  It no longer deals with linking/execution.
3241       The interface has been cleaned up considerably.
3242    
3243       Utility routines for dealing with linking and execution have been
3244       moved into their own substructures.
3245    
3246       (The ultimate goal of these changes is to provide a light-weight
3247       binfile loader/linker (at least for, e.g., stable libraries) that
3248       does not require CM or the compiler to be present.)
3249    
3250    CM documentation has been updated to reflect the changes to library
3251    organization.
3252    
3253    ----------------------------------------------------------------------
3254    Name: Matthias Blume
3255    Date: 2001/07/10 17:30:00 EDT
3256    Tag: Release_110_34
3257    Description:
3258    
3259    Minor tweak to 110.34 (re-tagged):
3260    
3261      - README.html file added to CVS repository
3262      - runtime compiles properly under FreeBSD 3.X and 4.X
3263    
3264    ----------------------------------------------------------------------
3265    Name: Matthias Blume
3266    Date: 2001/07/10 17:30:00 EDT
3267    Tag: Release_110_34
3268    Description:
3269    
3270    New version number (110.34). New bootfiles.
3271    
3272    ----------------------------------------------------------------------
3273    Name: Matthias Blume
3274    Date: 2001/07/09 16:00:00 EDT
3275    Tag: blume-20010709-more-varargs
3276    Description:
3277    
3278    I changed the handling of varargs in ml-nlffigen again:
3279    The ellipsis ... will now simply be ignored (with an accompanying warning).
3280    
3281    The immediate effect is that you can actually call a varargs function
3282    from ML -- but you can't actually supply any arguments beyond the ones
3283    specified explicitly.  (For example, you can call printf with its format
3284    string, but you cannot pass additional arguments.)
3285    
3286    This behavior is only marginally more useful than the one before, but
3287    it has the advantage that a function or, more importantly, a function
3288    type never gets dropped on the floor, thus avoiding follow-up problems with
3289    other types that refer to the offending one.
3290    
3291    ----------------------------------------------------------------------
3292    Name: Matthias Blume
3293    Date: 2001/07/09 11:25:00 EDT
3294    Tag: blume-20010709-varargs
3295    Description:
3296    
3297    1. ckit-lib.cm now exports structure Error
3298    2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
3299       with a warning accompanied by a source location.  Moreover, it
3300       merely skips the offending function or type and proceeds with the
3301       rest of its work.u  As a result, one can safely feed C code containing
3302       "..." to ml-nlffigen.
3303    3. There are some internal improvements to CM, providing slightly
3304       more general string substitutions in the tools subsystem.
3305    
3306    ----------------------------------------------------------------------
3307    Name: Matthias Blume
3308    Date: 2001/06/27 15:10:00 EDT
3309    Tag: blume-20010627-concur
3310    Description:
3311    
3312    Fixed a small bug in CM's handling of parallel compilation.
3313    (You could observe the bug by Control-C-interrupting an ordinary
3314    CMB.make or CM.stabilize and then attaching some compile servers.
3315    The result was that all of a sudden the previously interrupted
3316    compilation would continue on its own.  This was because of
3317    an over-optimization: CM did not bother to clean out certain queues
3318    when no servers were attached "anyway", resulting in the contents
3319    of these queues to grab control when new servers did get attached.)
3320    
3321    There is also another minor update to the CM manual.
3322    
3323    ----------------------------------------------------------------------
3324    Name: Matthias Blume
3325    Date: 2001/06/26 16:15:00 EDT
3326    Tag: blume-20010626-cmdoc
3327    Description:
3328    
3329    Minor typo fixed in CM manual (syntax diagram for libraries).
3330    
3331    ----------------------------------------------------------------------
3332    Name: Matthias Blume
3333    Date: 2001/06/25 22:55:00 EDT
3334    Tag: blume-20010625-x86pc
3335    Description:
3336    
3337    Fixed a nasty bug in the X86 assembly code that caused signal
3338    handlers to fail (crash) randomly.
3339    
3340    ----------------------------------------------------------------------
3341    Name: Matthias Blume
3342    Date: 2001/06/25 12:05:00 EDT
3343    Tag: blume-20010625-nlffigen
3344    Description:
3345    
3346    This update fixes a number of minor bugs in ml-nlffigen as reported by
3347    Nick Carter <nbc@andrew.cmu.edu>.
3348    
3349      1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
3350      2. Default names for generated files are now derived from the name of
3351         the C file *without its directory*.  In particular, this causes generated
3352         files to be placed locally even if the C file is in some system directory.
3353      3. Default names for generated signatures and structures are also derived
3354         from the C file name without its directory.  This avoids silly things
3355         like "structure GL/GL".
3356         (Other silly names are still possible because ml-nlffigen does not do
3357          a thorough check of whether generated names are legal ML identifiers.
3358          When in doubt, use command line arguments to force particular names.)
3359    
3360    ----------------------------------------------------------------------
3361    Name: Matthias Blume
3362    Date: 2001/06/21 12:25:00 EDT
3363    Tag: blume-20010621-eXene
3364    Description:
3365    
3366    eXene now compiles and (sort of) works again.
3367    
3368    The library name (for version > 110.33) is $/eXene.cm.
3369    
3370    I also added an new example in src/eXene/examples/nbody.  See the
3371    README file there for details.
3372    
3373    ----------------------------------------------------------------------
3374    Name: Matthias Blume
3375    Date: 2001/06/20 16:40:00 EDT
3376    Tag: blume-20010620-cml
3377    Description:
3378    
3379    CML now compiles and works again.
3380    
3381    Libraries (for version > 110.33):
3382    
3383      $cml/cml.cm            Main CML library.
3384      $cml/basis.cm          CML's version of $/basis.cm.
3385      $cml/cml-internal.cm   Internal helper library.
3386      $cml/core-cml.cm       Internal helper library.
3387      $cml-lib/trace-cml.cm  Tracing facility.
3388      $cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
3389    
3390    The installer (config/install.sh) has been taught how to properly
3391    install this stuff.
3392    
3393    ----------------------------------------------------------------------
3394    Name: Matthias Blume
3395    Date: 2001/06/19 17:55:00 EDT
3396    Tag: blume-20010619-instantiate
3397    Description:
3398    
3399    This un-breaks the fix for bug 1432.
3400    (The bug was originally fixed in 110.9 but I broke it again some
3401    time after that.)
3402    
3403    ----------------------------------------------------------------------
3404    Name: Matthias Blume
3405    Date: 2001/06/19 17:25:00 EDT
3406    Tag: blume-20010619-signals
3407    Description:
3408    
3409    This should (hopefully) fix the long-standing signal handling bug.
3410    (The runtime system was constructing a continuation record with an
3411    incorrect descriptor which would cause the GC to drop data on the floor...)
3412    
3413    ----------------------------------------------------------------------
3414    Name: Matthias Blume
3415    Date: 2001/06/15 15:05:00 EDT
3416    Tag: blume-20010615-moresparc
3417    Description:
3418    
3419    Here is a short late-hour update related to Sparc c-calls:
3420    
3421     -- made handling of double-word arguments a bit smarter
3422    
3423     -- instruction selection phase tries to collapse certain clumsily
3424        constructed ML-Trees; typical example:
3425    
3426            ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
3427    
3428        This currently has no further impact on SML/NJ since mlriscGen does
3429        not seem to generate such patterns in the first place, and c-calls
3430        (which did generate them in the beginning) has meanwhile been fixed
3431        so as to avoid them as well.
3432    
3433    ----------------------------------------------------------------------
3434    Name: Matthias Blume
3435    Date: 2001/06/15 15:05:00 EDT
3436    Tag: blume-20010615-sparc
3437    Description:
3438    
3439    The purpose of this update is to provide an implementation of NLFFI
3440    on Sparc machines.
3441    
3442    Here are the changes in detail:
3443    
3444       * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
3445       the Sparc implementation of the c-calls API.
3446       * The Sparc backend of SML/NJ has been modified to uniformely use %fp
3447       for accessing the ML frame.  Thus, we have a real frame pointer and
3448       can freely modify %sp without need for an omit-frame-ptr phase.
3449       The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
3450       this case.
3451       * ml-nlffigen has been taught to produce code for different architectures
3452       and calling conventions.
3453       * In a way similar to what was done in the x86 case, the Sparc
3454       backend uses its own specific extension to mltree.  (For example,
3455       it needs to be able to generate UNIMP instructions which are part
3456       of the calling convention.)
3457       * ml-nlffi-lib was reorganized to make it more modular (in particular,
3458       to make it easier to plug in new machine- and os-dependent parts).
3459    
3460    There are some other fairly unrelated bug fixes and cleanups as well:
3461    
3462       * I further hacked the .cm files for MLRISC tools (like MDLGen) so
3463       that they properly share their libraries with existing SML/NJ libraries.
3464       * I fixed a minor cosmetic bug in CM, supressing certain spurious
3465       follow-up error messages.
3466       * Updates to CM/CMB documentation.
3467    
3468    TODO items:
3469    
3470       * MLRISC should use a different register as its asmTemp on the Sparc.
3471         (The current %o2 is a really bad choice because it is part of the
3472         calling conventions, so things might interfere in unexpected ways.)
3473    
3474    ----------------------------------------------------------------------
3475    Name: Matthias Blume
3476    Date: 2001/06/07
3477    Tag: blume-20010607-calls
3478    Description:
3479    
3480    A number of internal changes related to C calls and calling conventions:
3481    
3482    1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
3483       number of bytes popped implicitly (by the callee).  In most cases
3484       this field is 0 but on x86/win32 it is some non-zero value.  This
3485       is information provided for the benefit of the "omit-frameptr" pass.
3486    2. The CALL instruction on the x86 carries a similar "pops" field.
3487       The instruction selection phase copies its value from the ML-Tree
3488       CALL statement.
3489    3. On all other architectures, the instruction selection phase checks
3490       whether "pops=0" and complains if not.
3491    4. The c-calls implementation for x86 now accepts two calling conventions:
3492       "ccall" and "stdcall".  When "ccall" is selected, the caller cleans
3493       up after the call and pops is set to 0.  For "stdcall", the caller
3494       does nothing, leaving the cleanup to the callee; pops is set to
3495       the number of bytes that were pushed onto the stack.
3496    5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
3497       distinguish between "ccall" and "stdcall".
3498    6. The UNIMP instruction has been added to the supported Sparc instruction
3499       set. (This is needed for implementing the official C calling convention
3500       on this architecture.)
3501    7. I fixed some of the .cm files under src/MLRISC/Tools to make them
3502       work with the latest CM.
3503    
3504    ----------------------------------------------------------------------
3505    Name: Matthias Blume
3506    Date: 2001/06/05 15:10:00 EDT
3507    Tag: blume-20010605-cm-index
3508    Description:
3509    
3510    0. The "lambdasplit" parameter for class "sml" in CM has been documented.
3511    
3512    1. CM can now generate "index files".  These are human-readable files
3513       that list on a per-.cm-file basis each toplevel symbol defined or
3514       imported.  The location of the index file for
3515       <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
3516       To enable index-file generation, set CM.Control.generate_index to true
3517       or export an environment-symbol: export CM_GENERATE_INDEX=true.
3518    
3519       The CM manual has been updated accordingly.
3520    
3521    2. I made some slight modifications to the c-calls API in MLRISC.
3522    
3523         a) There is now a callback to support saving/restoring of
3524            dedicated but caller-save registers around the actual call
3525            instruction.
3526         b) One can optionally specify a comment-annotation for the
3527            call instruction.
3528    
3529    3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
3530       (For example, the comment annotation shows the C prototype of
3531        the function being called.)
3532    
3533    ----------------------------------------------------------------------
3534    Name: Matthias Blume
3535    Date: 2001/06/01 13:30:00 EDT
3536    Tag: blume-20010601-nlffi-cleanup
3537    Description:
3538    
3539    This is mostly a cleanup of MLFFI stuff:
3540    
3541       - some signature files have been put into a more exposed place
3542       - the ugly 'f type parameter is gone (simplifies types tremendously!)
3543       - ml-nlffigen changed accordingly
3544       - tutorial updated
3545    
3546    Other changes:
3547    
3548       - author's affiliation in CM manual(s) updated
3549       - some more recognized keywords added to Allen's sml.sty
3550    
3551    ----------------------------------------------------------------------
3552    Name: Matthias Blume
3553    Date: 2001/05/25 15:30:00 EDT
3554    Tag: blume-20010525-iptr
3555    Description:
3556    
3557      - put the official 110.33-README (as it appears on the ftp server) under
3558        CVS
3559      - fixed a small bug related to incomplete pointer types in
3560        ml-nlffigen
3561      - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
3562        (it does not need the 'f type parameter)
3563    
3564    ----------------------------------------------------------------------
3565    Name: Matthias Blume
3566    Date: 2001/05/23 14:30:00 EDT
3567    Tag: Release_110_33
3568    Description:
3569    
3570    New version number (110.33).  New bootfiles.
3571    
3572    ----------------------------------------------------------------------
3573    Name: Matthias Blume
3574    Date: 2001/05/22 18:06:00 EDT
3575    Tag: blume-20010522-targets
3576    Description:
3577    
3578    Made install.sh use file config/targets.customized if it exists, falling
3579    back to config/targets if it doesn't.  This way one can have a customized
3580    version of the targets file without touching the "real thing", thus
3581    eliminating the constant fear of accidentally checking something bogus
3582    back into the CVS repository...  (File config/targets.customized must
3583    not be added to the repository!)
3584    
3585    ----------------------------------------------------------------------
3586    Name: Matthias Blume
3587    Date: 2001/05/22 16:30:00 EDT
3588    Tag: blume-20010522-minitut
3589    Description:
3590    
3591    1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
3592       struct returns.
3593    2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
3594       incomplete, preliminary documentation for NLFFI.
3595    
3596    ----------------------------------------------------------------------
3597    Name: Matthias Blume
3598    Date: 2001/05/14 11:30:00 EDT
3599    Tag: blume-20010514-script
3600    Description:
3601    
3602    Some bugs in install script fixed.
3603    
3604    In addition to that I also made a slight change to the NLFFI API:
3605    Functors generated by ml-nlffigen now take the dynamic library as a
3606    straight functor argument, not as a suspended one.  (The original
3607    functor code used to force the suspension right away anyway, so there
3608    was nothing gained by this complication of the interface.)
3609    
3610  </