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 656, Fri Jun 9 03:39:04 2000 UTC revision 1498, Tue May 25 20:59:00 2004 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: 2004/05/25 16:00:00 CDT
18    Tag: blume-20040525-group-owner
19    Description:
20    
21    CM now ignores (but still accepts) the "owner" information in group
22    descriptions.  The owner of a group is its next enclosing
23    library. Each group must have a unique owner.  (There is a virtual
24    "toplevel" library that own groups which are not nested within a real
25    library.)  Previously, each group had to explicitly declare its owner,
26    and CM would check that such a declaration is correct.  The new scheme
27    is to have CM check that for each group there is precisely one owning
28    library.
29    
30    The advantage of the new scheme is that the programmer no longer needs
31    to maintain the somewhat annoying owner information.  The downside is
32    that CM cannot enforce the ownership rule across multiple runs of
33    CM.make.  Fortunately, enclosing the same group in two different
34    libraries A and B which are not part of the same program does not
35    cause real problems.
36    
37    ----------------------------------------------------------------------
38    Name: Matthias Blume (blume (at) tti - c (dot) org)
39    Date: 2004/05/20 16:00:00 CDT
40    Tag: blume-20040520-win32
41    Description:
42    
43    Made the win32 version work again.  (Strangely, a misplaced comma had
44    slipped into win32-process.c which prevented the runtime from being
45    compiled correctly.)
46    
47    Also, included a minor addition to ml-build.bat analogous to what was
48    done in blume-20040519-ml-build.
49    
50    ----------------------------------------------------------------------
51    Name: Matthias Blume (blume (at) tti - c (dot) org)
52    Date: 2004/05/19 22:10:00 CDT
53    Tag: blume-20040519-ml-build
54    Description:
55    
56    Arranged for ml-build to clean up after itself a little bit better.
57    The script generates a temporary SML source file and compiles it using
58    CM, so CM generates metadata (GUID, SKEL, objectfile) for it.  It now
59    gets rid of those at the end, so they don't accumulate under .cm.
60    
61    This required a minor change to install.sh because the name of the
62    metadata directory (default: .cm) is actually configurable at
63    installation time.
64    
65    ----------------------------------------------------------------------
66    Name: Matthias Blume (blume (at) tti - c (dot) org)
67    Date: 2004/05/18 15:50:00 CDT
68    Tag: blume-20040518-mkreader
69    Description:
70    
71    Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
72    implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
73    
74    ----------------------------------------------------------------------
75    Name: Matthias Blume (blume (at) tti - c (dot) org)
76    Date: 2004/05/11 14:35:00 CDT
77    Tag: blume-20040511-win32sock
78    Description:
79    
80    Added previously missing support for many socket-related functions
81    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
82    for the voluminous patch!
83    
84    (I have not tested this patch under win32 yet.)
85    
86    Here is David's e-mail:
87    
88    Hi,
89    
90    Attached to this email you find a diff against sml/nj 110.45
91    that will enable socket support under Windows.
92    
93    To apply the patch (using unix or cygwin)
94    1) gunzip runtime.diff.gz
95    2) "cd" into "src/runtime" in the source tree of a fresh
96       110.45 installation.
97    3) patch -p 1 < [your/path/to]runtime.diff
98    
99    The code compiles fine but has NOT yet been extensively tested.
100    I only ran a few tests for basic socket client functionality
101    (which worked fine).  Especially the functions that use ioctl
102    are not tested at all and might not work (see below).
103    
104    I implemented this since we want to move to a newer version of sml/nj
105    but need socket support in order to use it.  This is the first time I
106    even had a look at the sml/nj source,  so please review my changes
107    before making this part of the distribution!  Here are a few issues
108    that I think might be better for someone to solve who is more
109    familiar with the sml/nj source (and socket programming):
110    
111    - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
112      exception since I could not figure out what the windows equivalent
113      of these functions is
114    
115    - In sockets-osdep.h there are a some #include statements that are
116      only used in a few files that include sockets-osdep.h
117    
118    - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
119      does not process its return value since I don't know how to
120      report an error upwards.
121    
122    - It would probably be good to have a call to WSACleanup() when
123      the library is unloaded (if there is such a possibility).
124      Otherwise I think Windows will take care of this automatically
125      when the process finishes.
126    
127    - I used ioctlsocket() as a replacement for ioctl() but I have
128      no idea if that is actually the proper replacement on Windows.
129    
130    - All these issues are marked in the code by "FIXME" comments.
131    
132    We use sml/nj extensively in our products and are quite happy
133    with it.  I hope this contribution will help you.
134    
135    Keep up the good work!
136    
137    David
138    
139    ----------------------------------------------------------------------
140    Name: Matthias Blume (blume (at) tti - c (dot) org)
141    Date: 2004/05/11 14:20:00 CDT
142    Tag: blume-20040511-installml
143    Description:
144    
145    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
146    the patch.)
147    
148    ----------------------------------------------------------------------
149    Name: Matthias Blume (blume (at) tti - c (dot) org)
150    Date: 2004/05/11 14:05:00 CDT
151    Tag: blume-20040511-nlffi-netbsd
152    Description:
153    
154    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
155    the patch.)
156    
157    ----------------------------------------------------------------------
158    Name: Matthias Blume (blume (at) tti - c (dot) org)
159    Date: 2004/05/11 12:05:00 CDT
160    Tag: blume-20040511-exports
161    Description:
162    
163    As per request by Adam Chlipala <adam@hcoop.net>, extended various
164    export lists in compiler-related .cm-files.
165    
166    ----------------------------------------------------------------------
167    Name: Matthias Blume (blume (at) tti - c (dot) org)
168    Date: 2004/05/11 11:35:00 CDT
169    Tag: blume-20040511-allsource
170    Description:
171    
172    The installer now honors the "src-smlnj" target again, although its meaning
173    has changed from "all sources required for the compiler" to "all sources
174    the installer knows about".  In other words, if you enable "src-smlnj"
175    in the "targets" file, then the installer will pull in sources for
176    everything.  (Notice that this refers to source code only.  Compiled
177    code is still only installed for modules that were requested explicitly
178    or which are required for other modules that were requested explicitly.)
179    
180    ----------------------------------------------------------------------
181    Name: Matthias Blume (blume (at) tti - c (dot) org)
182    Date: 2004/04/23 17:40:00 CDT
183    Tag: blume-20040423-ieee-scan
184    Description:
185    
186    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
187    in the exponent calculation we get INF or ZERO (depending on the mantissa
188    and the sign of the exponent).
189    
190    ----------------------------------------------------------------------
191    Name: Matthias Blume (blume (at) tti - c (dot) org)
192    Date: 2004/04/23 10:40:00 CDT
193    Tag: blume-20040423-ml-build
194    Description:
195    
196    The ml-build script now terminates with a non-0 status when something
197    goes wrong.
198    
199    ----------------------------------------------------------------------
200    Name: Matthias Blume (blume (at) tti - c (dot) org)
201    Date: 2004/04/22 16:35:00 CDT
202    Tag: blume-20040422-Option
203    Description:
204    
205    Made exception Option to be the same as exception Option.Option
206    (as it should be).
207    
208    ----------------------------------------------------------------------
209    Name: Allen Leung (leunga (at) reservoir (dot) com)
210    Date: 2004/03/19 14:40:00 EST
211    Tag: leunga-20040319-cygwin-nlffi
212    Description:
213    
214    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
215    of SML/NJ.  The problem is that
216    
217       lib = dlopen(NULL, ...)
218       f   = dlsym(lib, "malloc");
219    
220    does not work on Windows unless we explicitly export symbols
221    such as 'malloc' during linking.  We fixed this by explicitly
222    exporting the required symbols with the magic gcc incantation:
223    
224        -Wl,--export-all cygwin.def
225    
226    where cygwin.def is a file containing all the symbols that
227    we wish to export.
228    
229    I suspect this is a Windows problem and we'll have to
230    do the same (somehow with windows compilers) when
231    we build the native win32 version with the system
232    calls LoadLibrary/GetProcAddress.
233    
234    ----------------------------------------------------------------------
235    Name: Matthias Blume (blume (at) tti - c (dot) org)
236    Date: 2004/03/04 16:35:00 CST
237    Tag: blume-20040304-intinf-fmt
238    Description:
239    
240    Fixed problem with IntInf.fmt (sign would show up on the right instead
241    of on the left for BIN, OCT, and HEX).
242    
243    ----------------------------------------------------------------------
244    Name: Matthias Blume (blume (at) tti - c (dot) org)
245    Date: 2004/03/04 11:25:00 CST
246    Tag: blume-20040304-symlinks
247    Description:
248    
249    Fixed problem with installer script (unix only) where bin/ml-yacc and
250    friends pointed (via symlinks) to absolute locations instead of just
251    .run-sml.  This was reported by Vesa A Norrman.
252    
253    ----------------------------------------------------------------------
254    Name: Matthias Blume (blume (at) tti - c (dot) org)
255    Date: 2004/02/13 14:50:00 CST
256    Tag: Release_110_45
257    Description:
258    
259    New working version (110.45). New bootfiles.
260    
261    ----------------------------------------------------------------------
262    Name: Matthias Blume (blume (at) tti - c (dot) org)
263    Date: 2004/01/26 15:15:15 CST
264    Tag: blume-20040126-toplevel
265    Description:
266    
267    Improved handling of exceptions at the interactive toplevel.
268    
269    ----------------------------------------------------------------------
270    Name: Matthias Blume (blume (at) tti - c (dot) org)
271    Date: 2004/01/26 11:25:00
272    Tag: blume-20040126-app
273    Description:
274    
275    Type of top-level "app" corrected.
276    Added code for setting vp_limitPtrMask to Win32-specific runtime.
277    
278    ----------------------------------------------------------------------
279    Name: Matthias Blume (blume (at) tti - c (dot) org)
280    Date: 2003/11/18 17:10 CST
281    Tag: blume-20031118-basis-fiddle
282    Description:
283    
284    - changed Timer interface to what might become the spec
285    - POSIX_FLAGS -> BIT_FLAGS according to spec
286    - some other minor discrepancies wrt. spec eliminated
287    
288    ----------------------------------------------------------------------
289    Name: Matthias Blume (blume (at) tti - c (dot) org)
290    Date: 2003/11/06 12:00:00 CST
291    Tag: Release_110_44
292    Description:
293    
294    New working version (110.44). New bootfiles.
295    
296    ----------------------------------------------------------------------
297    Name: Matthias Blume (blume (at) tti - c (dot) org)
298    Date: 2003/11/04 11:50:00 CST
299    Tag: blume-20031104-move-libraries
300    Description:
301    
302    Eliminated the "dont_move_libraries" directive in config/targets.
303    (The mechanism was broken and could not be fixed easily.  Moreover,
304    there does not seem to be any reason not to move all libraries into
305    lib during installation.  I originally implemented this directive as a
306    backward-compatibility feature when I first introduced the new CM.
307    Now that things have been stable for a long time and going back to the
308    old CM is not an option, there is no reason to keep it around.)
309    
310    ----------------------------------------------------------------------
311    Name: Matthias Blume (blume (at) tti - c (dot) org)
312    Date: 2003/11/03 16:00:00 CST
313    Tag: blume-20031103-installdir
314    Description:
315    
316    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
317    Richards for pointing out the problem and providing the solution.)
318    
319    ----------------------------------------------------------------------
320    Name: Matthias Blume (blume (at) tti - c (dot) org)
321    Date: 2003/10/01 17:05:00 CDT
322    Tag: blume-20031001-lal-mlrisc
323    Description:
324    
325    MLRISC bug fix from Lal.
326    
327    ----------------------------------------------------------------------
328    Name: Matthias Blume (blume (at) tti - c (dot) org)
329    Date: 2003/09/30 16:10:00 CDT
330    Tag: blume-20030930-primio-bat
331    Description:
332    
333    1. Added openVector, nullRd, and nullWr to PRIM_IO.
334    2. Improved .bat files (for Win32 port) to make things work under Win95.
335       (thanks to Aaron S. Hawley for this one)
336    
337    ----------------------------------------------------------------------
338    Name: Matthias Blume (blume (at) tti - c (dot) org)
339    Date: 2003/09/26 16:05:00 CDT
340    Tag: blume-20030926-wrappriv
341    Description:
342    
343    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
344    
345    ----------------------------------------------------------------------
346    Name: Matthias Blume (blume (at) tti - c (dot) org)
347    Date: 2003/09/26 15:00:00 CDT
348    Tag: blume-20030926-110_43_3
349    Description:
350    
351    - additional cleanup
352    - version number bump, NEW BOOTFILES
353    
354    ----------------------------------------------------------------------
355    Name: Matthias Blume (blume (at) tti - c (dot) org)
356    Date: 2003/09/26 12:00:00 CDT
357    Tag: blume-20030926-ppautoload
358    Description:
359    
360    I modified the read-eval-print loop so that the autoloader gets
361    invoked whenever the prettyprinter tries to look up a symbol that
362    is not currently defined in the toplevel environment but which
363    appears in CM's autoload registry.  As a result, we see far fewer of
364    those ?.Foo.Bar.xxx names in the prettyprinter's output.
365    
366    In addition to this I tried to clean up some pieces of the Basis
367    implementation (e.g., Socket, Word8Array) in order to prevent other
368    instances of these ?.Foo.Bar.xxx names from being printed.
369    
370    The mechanism that picks names for types still needs some work, though.
371    (Right now it seems that if there is a type A.t which is defined to
372    be B.u, but B is unavailable at toplevel, then A.t gets printed as
373    "?.B.u" although the perhaps more sensible solution would be to use
374    "A.t" in this case.  In other words, the prettyprinter should follow
375    a chain of DEFtycs not farther than there are corresponding toplevel
376    names in the current environment.)
377    
378    ----------------------------------------------------------------------
379    Name: Matthias Blume (blume (at) tti - c (dot) org)
380    Date: 2003/09/24 16:31:00 CDT
381    Tag: blume-20030924-installer
382    Description:
383    
384    Another installer tweak:  All the ML code for the installer is now
385    compiled during CMB.make and put into a little library called
386    $smlnj/installer.cm.  The installation then simply invokes
387    
388       sml -m $smlnj/installer.cm
389    
390    and everything happens automagically.
391    
392    Win32:  ML code senses value of environment variable SMLNJ_HOME.
393    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
394            and BINDIR.
395    
396    The new scheme guarantees that the ML code responsible for the installation
397    is in sync with the APIs of the main system.  Also, the installer is
398    somewhat faster because the installer script is precompiled.
399    
400    ----------------------------------------------------------------------
401    Name: Matthias Blume (blume (at) tti - c (dot) org)
402    Date: 2003/09/24 15:35:00 CDT
403    Tag: blume-20030924-synsock
404    Description:
405    
406    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
407    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
408    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
409    implementations of non-blocking operations from CML's Socket
410    structure.
411    
412    ----------------------------------------------------------------------
413    Name: Matthias Blume (blume (at) tti - c (dot) org)
414    Date: 2003/09/24 15:10:05 CDT
415    Tag: blume-20030924-sockets
416    Description:
417    
418    1. Fixed SOCKET API and implementation to match Basis spec.
419       This required changing the internal representation of sockets to one
420       that remembers (for each socket file descriptor) whether it is currently
421       blocking or non-blocking.  This state is maintained lazily (i.e., a system
422       call is made only if the state actually needs to change).
423    
424    2. OS-specific details of sockets were moved into separate files, thus
425       making it possible to unify the bulk of the socket implementations
426       between Unix and Win32.
427    
428    3. CML's socket API changed accordingly.
429       (Note that we need to remove non-blocking functions from this API
430       since they are redundant in the case of CML!)
431    
432    4. CML's socket implementation now makes use of non-blocking functions
433       provided by Basis, thus removing all OS-dependent code from this part
434       of CML.
435    
436    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
437    
438    ----------------------------------------------------------------------
439    Name: Matthias Blume (blume (at) tti - c (dot) org)
440    Date: 2003/09/22 12:10:00 CDT
441    Tag: blume-20030922-110_43_2
442    Description:
443    
444    Made a new interim version and bootfiles for developer's bootstrapping
445    convenience.
446    
447       110.43.2 -- NEW BOOTFILES
448    
449    ----------------------------------------------------------------------
450    Name: Matthias Blume (blume (at) tti - c (dot) org)
451    Date: 2003/09/19 15:55:00 CDT
452    Tag: blume-20030919-cmdir
453    Description:
454    
455    1. new-install.sh -> install.sh
456    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
457    3. tweaked installer so that another name instead of .cm can be chosen
458       at install time (by setting the CM_DIR_ARC environment variable
459       during installation); once installation is complete, the name is
460       fixed
461    
462    ----------------------------------------------------------------------
463    Name: Matthias Blume (blume (at) tti - c (dot) org)
464    Date: 2003/09/18 16:00:00 CDT
465    Tag: blume-20030918-110_43_1
466    Description:
467    
468    Made a new interim version and bootfiles for developer's bootstrapping
469    convenience.
470    
471       110.43.1 -- NEW BOOTFILES
472    
473    ----------------------------------------------------------------------
474    Name: Matthias Blume (blume (at) tti - c (dot) org)
475    Date: 2003/09/18 15:20:00 CDT
476    Tag: blume-20030918-misc
477    Description:
478    
479    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
480       this seems to be controversial at the moment) and used those in
481       Posix.ProcEnv.times.
482    
483    2. Added Time.{from,to}Nanoseconds to Time.
484    
485    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
486       For example, fromLargeInt never needs to look at more than 3 "big
487       digits" to get its 53 bits of precision.
488    
489    ----------------------------------------------------------------------
490    Name: Matthias Blume (blume (at) tti - c (dot) org)
491    Date: 2003/09/17 16:30:00 CDT
492    Tag: blume-20030917-real32-slices
493    Description:
494    
495    Added an entry to the primitive environment
496    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
497    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
498    
499    Removed some of the "magic" constants in real64.sml and replaced them
500    with code that generates these values from their corresponding
501    integer counterparts.
502    
503    Made all(?) the slice-related changes to the Basis and made everything
504    compile again...
505    
506    ----------------------------------------------------------------------
507    Name: Matthias Blume (blume (at) tti - c (dot) org)
508    Date: 2003/09/15 17:45:00 CDT
509    Tag: blume-20030915-rbase
510    Description:
511    
512    Fixed bug in Real.fromLargeInt.
513    
514    ----------------------------------------------------------------------
515    Name: Matthias Blume (blume (at) tti - c (dot) org)
516    Date: 2003/09/13 18:11:00 CDT
517    Tag: blume-20030913-libinstall
518    Description:
519    
520    Minor bugfix in config/libinstall (set anchor with path to
521    standalone tool after installing it, otherwise libraries that
522    need ml-lex or ml-yacc won't compile the first time the installer
523    runs).
524    
525    ----------------------------------------------------------------------
526    Name: Matthias Blume (blume (at) tti - c (dot) org)
527    Date: 2003/09/12 11:45:00 CDT
528    Tag: blume-20030912-various
529    Description:
530    
531     - fixed bug in Real.toLargeInt
532     - fixed bug in Posix.ProcEnv.times
533     - changed inputLine functions to return an option
534     - minor installer improvements / bugfixes
535     - changed default @SMLalloc parameter for x86/celeron to 64k
536    
537    ----------------------------------------------------------------------
538    Name: Matthias Blume (blume (at) tti - c (dot) org)
539    Date: 2003/09/09 22:00:00 CDT
540    Tag: Release_110_43
541    Description:
542    
543        New working release 110.43.  New bootfiles.
544    
545    ----------------------------------------------------------------------
546    Name: Matthias Blume (blume (at) tti - c (dot) org)
547    Date: 2003/09/09 19:20:00 CDT
548    Tag: blume-20030909-installer
549    Description:
550    
551    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
552    Modified config/install.bat to take advantage of it.  Also modified
553    config/install.sh (and called it config/new-install.sh) to take advantage
554    of it on Unix systems.  (The SML code is (supposed to be) platform-
555    independent.)
556    
557    The installer can now install everything under Win32
558    as well as under *nix as long as it compiles.
559    
560    Other changes:
561    
562      - made CML compile again under Win32
563      - made eXene compile under Win32 (by providing a fake structure UnixSock
564        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
565      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
566        same as type int;  under Win32 it isn't
567      - fixed some slice-related problems in the win32-specific parts of CML
568      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
569        (under Win32, the volume name is case-insensitive, and the
570        OS.Path code compares volume names for equality)
571    
572    ----------------------------------------------------------------------
573    Name: Matthias Blume (blume (at) tti - c (dot) org)
574    Date: 2003/09/08 11:55:00 CDT
575    Tag: blume-20030908-fullpath
576    Description:
577    
578    Made Win32 version of OS.FileSys.fullPath return current directory
579    when given an empty string.  This is what the spec says, and incidentally,
580    CM depends on it.  (CM otherwise goes into an infinite loop in certain
581    cases when presented with the name of a non-existing .cm file.)
582    
583    ----------------------------------------------------------------------
584    Name: Matthias Blume (blume (at) tti - c (dot) org)
585    Date: 2003/09/04 16:30:00 CDT
586    Tag: blume-20030905-slices-etc
587    Description:
588    
589      1. Changed interface to vectors and arrays in Basis to match
590         (draft) Basis spec.
591      2. Added signatures and implementations of slices according to
592         Basis spec.
593      3. Edited source code throughout the system to make it compile again
594         under 1. and 2.  (In some cases code had to be added to have it
595         match the new signatures.)
596      4. MLRISC should be backward-compatible: the copies of the originals
597         of files that needed to change under 3. were retained, the .cm files
598         check the compiler version number and use old versions when
599         appropriate.
600      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
601         dirstream -> string option (in accordance with Basis spec).
602      6. When generating code that counts lines, ml-lex used function
603         CharVector.foldli, taking advantage of its old interface.
604         This has been replaced with the corresponding code from
605         CharVectorSlice.  (html-lex must be re-lexed!)
606      7. BitArray in smlnj-lib/Util has been extended/modified to match the
607         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
608      8. Removed temporary additions (fromInternal, toInternal) from the
609         (now obsolete) IntInf in smlnj-lib/Util.
610      9. Cleaned up structure Byte.
611     10. Added localOffset, scan, and fromString to Date (according to spec).
612         Cleaned/corrected implementation of Date.
613         (Still need to check for correctness; implement better canonicalizeDate.)
614     11. Added "scan" to signature IEEE_REAL.
615     12. Some improvements to IntInf [in particular: efficiency-hack for
616         mod and rem when second operand is 2 (for parity checks).]
617     13. Changed representation of type Time.time, using a single IntInf.int
618         value counting microseconds.  This considerably simplified the
619         implementation of structure Time.  We now support negative time
620         values; scan and fromString handle signs.
621     14. Functor PrimIO now takes two additional arguments (VectorSlice and
622         ArraySlice).
623    
624    ----------------------------------------------------------------------
625    Name: Matthias Blume (blume (at) tti - c (dot) org)
626    Date: 2003/08/28 17:00:00 CDT
627    Tag: blume-20030828-intinf
628    Description:
629    
630    This is a major update which comes with a version number bump
631    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
632    and an implementation of IntInf in the Basis.
633    
634    There are a fairly large number of related changes and updates throughout
635    the system:
636    
637    Basis:
638            - Implemented IntInf.
639            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
640            - Added some missing Real64 operations, most notably Real.toLargeInt.
641            - Added FixedInt as a synonym for Int32.
642    
643    compiler:
644            * Added support for a built-in intinf type.
645              - literals
646              - pattern matching
647              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
648              - overloading on literals and operations
649    
650            This required adding a primitive type intinf, some additional
651            primops, and implementations for several non-trivial intinf
652            operations in Core.  (The intinf type is completely abstract
653            to the compiler; all operations get delegated back to the Core.)
654    
655            * Intinf equality is handled by polyequal.  However, the compiler
656            does not print its usual warning in this case (since polyequal
657            is the right thing to do there).
658    
659            * Improved the organization of structure InlineT.
660    
661            * A word about conversion primops:
662            If conversions involving intinf do not cancel out during
663            CPS contract, then the compiler must insert calls to Core functions.
664            Since all core access must be resolved already during the FLINT
665            translate phase, it would be too late a the time of CPS contract
666            to add new Core calls.  For this reason, conversion primops
667            for intinf carry two arguments:  1. the numeric argument that
668            they are supposed to convert, and 2. the Core function that
669            can help with this conversion if necessary.  If CPS contract
670            eliminates a primop, then the associated Core function becomes
671            dead and goes away.  Intinf conversion primops that do not get
672            eliminated by CPS contract get rewritten into calls of their
673            core functions by a separate, new phase.
674    
675    interactive system:
676            - Control.Print.intinfDepth controls max length of intinf constants
677            being printed. (Analogous to Control.Print.stringDepth.)
678            - Cleanup in printutil and pputil: got rid of unused stuff and
679            duplicates; replaced some of the code with code that makes better
680            use of library functionality.
681    
682    CM:
683            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
684            are no longer silent.
685    
686    CKIT:
687            Fixed mismatched uses of Int32 and LargeInt.  I always decided
688            in favor of LargeInt -- which is now the same as IntInf.
689            CKIT-knowledgable people should check whether this is what's
690            intended and otherwise change things back to using Int32 or
691            FixedInt.
692    
693    Throughout the code:
694            Started using IntInf.int literals and built-in operations
695            (e.g., comparison with 0) where this seems appropriate.
696    
697    
698    ----------------------------------------------------------------------
699    Name: Dave MacQueen (dbm@cs.uchicago.edu)
700    Date: 2003/08/13 11:36:00 CDT
701    Tag: dbm-20030813-mcz-merge1
702    Description:
703    
704    Merging changes from the mcz-branch development branch into trunk.
705    These changes involve replacement of the emulated old prettyprinter
706    interface with direct use of the SML/NJ Lib PP library, and fixing
707    of a couple of bugs (895, 1186) relating to error messages.  A new
708    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
709    has been added.
710    
711    ----------------------------------------------------------------------
712    Name: Matthias Blume (blume (at) tti - c (dot) org)
713    Date: 2003/08/11 15:45:00 CDT
714    Tag: blume-20030811-windows
715    Description:
716    
717    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
718    
719        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
720    
721    This patch restores SML/NJ's ability to run under win32.  There are a
722    number of changes, including fixes for several bugs that had gone
723    unnoticed until now:
724    
725      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
726        trying to run the win32 version from within cygwin.)
727    
728      - There are a number of simple .bat scripts that substitute for their
729        corresponding Unix shell-scripts.  (See below.)
730    
731      - The internals of ml-build have been modified slightly.  The main
732        difference is that instead of calling ".link-sml" (or link-sml.bat)
733        using OS.Process.system, the ML process delegates this task back
734        to the script.  Otherwise problems arise in mixed environments such
735        as Cygwin where scripts look and work like Unix scripts, but
736        where OS.Process.system cannot run them.
737    
738      - In CM, the srcpath pickler used native pathname syntax -- which
739        is incorrect in the case of cross-compilation.  The new pickle format
740        is independent of platform-specific naming conventions.
741    
742      - Path configuration files (such as lib/pathconfig) can now choose
743        between native and standard syntax.  Placing a line of the form
744    
745            standard!
746    
747        into the file causes all subsequent paths to be interpreted using
748        CM standard pathname syntax (= Unix conventions); a line
749    
750            native!
751    
752        switches back to native style.  This was needed so that
753        path config files can be written portably, see src/system/pathconfig.
754    
755      - Runtime system:
756    
757          - win32-filesys.c:  get_file_time and set_file_time now
758            access modification time, not creation time.
759    
760          - I/O code made aware of new array representation.
761    
762          - Bug fixes in X86.prim.masm.
763    
764       - src/system/makeml made aware of win32. (For use under cygwin
765            and other Unix-environments for windows.)
766    
767       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
768         which caused BinIO.inputAll to fail consistently.
769    
770    .bat scripts:
771    
772       Windows .bat scripts assume that SMLNJ_HOME is defined.
773    
774       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
775         applications (sml, ml-yacc, ml-lex).
776       - ml-build.bat: analogous to ml-build.
777       - config\install.bat:  Analogous to config/install.sh.  This requires
778         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
779         (nmake etc. must be on the path, and vcvars32 must have been run.)
780         Moreover, sources for ml-lex and ml-yacc need to exist under src,
781         and the bootfile hierarchy must have been unpacked under
782         sml.boot.x86-win32.
783         The script is very primitive and does a poor job at error checking.
784         It only installs the base system, ml-lex, and ml-yacc.  No other
785         libraries are being installed (i.e., you get only those that
786         are part of the compiler.)
787       - link-sml.bat:  analogous to .link-sml, but not currently used
788    
789    Unrelated bug fixes:
790    
791      - ml-nlffigen now exports structures ST_* corresponding to incomplete
792        types.
793      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
794        otherwise.)
795    
796    ----------------------------------------------------------------------
797    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
798    Date: 2003/06/17
799    Tag: macqueen-20030617-bug895
800    Description:
801    
802    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
803    Tag will be used for new development branch (mcz-branch) for
804    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
805    
806    ----------------------------------------------------------------------
807    Name: Matthias Blume (blume (at) tti - c (dot) org)
808    Date: 2003/05/27 16:55:00 CDT
809    Tag: blume-20030527-polyeq
810    Description:
811    
812    Tried to eliminated most cases of polymorphic equality.
813    
814    ----------------------------------------------------------------------
815    Name: Matthias Blume (blume (at) tti - c (dot) org)
816    Date: 2003/05/21 17:45:00 CDT
817    Tag: blume-20030517-complete
818    Description:
819    
820    Two changes:
821    
822      1. Added a flag for controlling whether non-exhaustive bindings will
823         be treated as errors (default is false).
824      2. Cleaned up the *entire* source tree so that CMB.make goes through
825         without a single non-exhaustive match- or bind warning.
826    
827    ----------------------------------------------------------------------
828    Name: Matthias Blume (blume (at) tti - c (dot) org)
829    Date: 2003/05/17 10:20:00 CDT
830    Tag: blume-20030517-absyn
831    Description:
832    
833    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
834    
835       This mainly affects the quality of error messages.  However, some
836       of the code is now more straightforward than before. (Treatment of
837       the above four constructs in translate.sml is much simpler than
838       the "macro-expansion" that was going on before.  Plus, the mach-
839       compiler no longer gets invoked just to be able to compile an
840       if-expression.)
841    
842    2. The ErrorMsg.Error exception is now caught and absorbed by the
843       interactive loop.
844    
845    ----------------------------------------------------------------------
846    Name: Allen Leung
847    Date: 2003/05/16 13:05:00 CDT
848    Tag: leunga-20030516-cygwin-runtime
849    Description:
850    
851        Ported the runtime system to cygwin, which uses the unix
852    x86-unix bin files.  Missing/buggy features:
853    
854       o getnetbyname, getnetbyaddr: these functions seem to be missing in
855         the Cygwin library.
856       o Ctrl-C handling may be flaky.
857       o Windows system calls and Windows I/O are not supported.
858    
859       A new set of binfiles is located at:
860    
861         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
862    
863       This is only needed for bootstrapping the cygwin version of smlnj.
864       Other x86 versions can use the existing binfiles.
865    
866    ----------------------------------------------------------------------
867    Name: Matthias Blume
868    Date: 2003/04/08 15:42:00 CDT
869    Tag: blume-20030408-listpair
870    Description:
871    
872    1. Added a target 'mlrisc' to installer.
873    
874    2. Added missing elements to structure ListPair.
875    
876    ----------------------------------------------------------------------
877    Name: Allen Leung
878    Date: 2003/01/07 10:40:00 EST
879    Tag: leunga-20030107-int-rem
880    Description:
881    
882    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
883    The arguments to the SUBL instruction were swapped.
884    
885    ----------------------------------------------------------------------
886    Name: Matthias Blume
887    Date: 2002/12/12 16:25:00 EST
888    Tag: blume-20021212-risc-ra
889    Description:
890    
891    Fixed a serious bug in the rewrite code for FP spilling/reloading that
892    sent the RA into an infinite loop when floating point registers get
893    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
894    110.37 and 110.38.)
895    There was another set of potential problems related to the handling of
896    MLRISC annotations (but those did not yet cause real problems, apparently).
897    
898    ----------------------------------------------------------------------
899    Name: Matthias Blume
900    Date: 2002/12/06 22:40:00 EST
901    Tag: blume-20021206-cm-fileid
902    Description:
903    
904    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
905    This fixes the problem of CM getting confused by files that suddenly
906    change their identity (e.g., by getting unlinked and recreated by some
907    text editor such as vi).  There might be a better/cheaper/cleaner way
908    of doing this, but for now this will have to do.
909    
910    ----------------------------------------------------------------------
911    Name: Matthias Blume
912    Date: 2002/10/28 09:50:00 EST
913    Tag: blume-20021028-typecheck
914    Description:
915    
916    Exported structure Typecheck from $smlnj/viscomp/core.cm.
917    
918    ----------------------------------------------------------------------
919    Name: Matthias Blume
920    Date: 2002/10/17 09:10:00 EDT
921    Tag: Release_110_42
922    Description:
923    
924    In good old tradition, there has been a slight hiccup so that we have
925    to patch 110.42 after the fact.   The old release tag has been replaced
926    (see below).
927    
928    The change solves a problem with two competing approaches the
929    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
930    each other's way.
931    
932    This change only affects the runtime system code and the installer script.
933    (No new bootfiles.)
934    
935    ----------------------------------------------------------------------
936    Name: Matthias Blume
937    Date: 2002/10/16 12:00:00 EDT
938    Tag: Release_110_42_removed
939    Description:
940    
941    New working release.  New bootfiles.
942    
943    ----------------------------------------------------------------------
944    Name: Matthias Blume
945    Date: 2002/10/10 13:10:00 EDT
946    Tag: blume-20021010-ppc-divs
947    Description:
948    
949    The mltree operator DIVS must be implemented with an overflow check on
950    the PPC because the hardware indicates divide-by-zero using "overflow" as
951    well.
952    
953    ----------------------------------------------------------------------
954    Name: Matthias Blume
955    Date: 2002/07/23 11:45:00
956    Tag: blume-20020723-smlnj-home
957    Description:
958    
959    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
960    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
961    CM_PATHCONFIG is also set), the path configuration file is assumed
962    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
963    the entire tree to some other place and everything will "just work".
964    
965    (Companion commands such as ml-build and ml-makedepend also sense this
966    variable.)
967    
968    ----------------------------------------------------------------------
969    Name: Matthias Blume
970    Date: 2002/07/12 21:19:00 EDT
971    Tag: blume-20020712-liveness
972    Description:
973    
974    Exported two useful "step" functions from liveness module (MLRISC).
975    
976    ----------------------------------------------------------------------
977    Name: Matthias Blume
978    Date: 2002/07/05 16:00 EDT
979    Tag: Release_110_41
980    Description:
981    
982    New working release.  New bootfiles.
983    
984    ----------------------------------------------------------------------
985    Name: Matthias Blume
986    Date: 2002/07/05 10:25:00 EDT
987    Tag: blume-20020705-btimp
988    Description:
989    
990    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
991    other clients can set up backtracing support.
992    
993    ----------------------------------------------------------------------
994    Name: Matthias Blume
995    Date: 2002/06/25 17:23:00 EDT
996    Tag: blume-20020625-fpmax
997    Description:
998    
999    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
1000    case.  (The sense of the isNaN test was reversed -- which made min and
1001    max always return their first argument.)
1002    
1003    ----------------------------------------------------------------------
1004    Name: Matthias Blume
1005    Date: 2002/06/11
1006    Tag: blume-20020611-unixpath
1007    Description:
1008    
1009    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
1010    
1011    ----------------------------------------------------------------------
1012    Name: Matthias Blume
1013    Date: 2002/06/10 16:35:00 EDT
1014    Tag: blume-20020610-ieeereal
1015    Description:
1016    
1017    I back-ported my implementation of IEEEReal.fromString from the
1018    idlbasis-devel branch so that we can test it.
1019    
1020    Another small change is that ppDec tries to give more information
1021    than just "<sig>" in the case of functors.  However, this code is
1022    broken in some mysterious way if the functor's body's signature
1023    has not been declared by ascription but gets inferred from the
1024    implementation.  This needs fixing...
1025    
1026    ----------------------------------------------------------------------
1027    Name: Matthias Blume
1028    Date: 2002/05/31
1029    Tag: blume-20020531-btrace-mode
1030    Description:
1031    
1032    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
1033    the wayside when I switched over to using Controls everywhere.)
1034    
1035    ----------------------------------------------------------------------
1036    Name: Lal George
1037    Date: 2002/05/23 12:21:40 EDT
1038    Tag: george-20020523-visual-labels
1039    Description:
1040    
1041    Labels are now displayed in the graphical output to make
1042    the fall-through and target blocks obvious.
1043    
1044    ----------------------------------------------------------------------
1045    Name: Matthias Blume
1046    Date: 2002/05/22 11:03:00 EDT
1047    Tag: blume-20020522-shrink
1048    Description:
1049    
1050    John tweaked yesterday's fix for 1131 to handle an out-of-memory
1051    situation that comes up when allocating huge arrays.
1052    
1053    ----------------------------------------------------------------------
1054    Name: Matthias Blume
1055    Date: 2002/05/21 16:00:00 EDT
1056    Tag: Release_110_40
1057    Description:
1058    
1059    New working release (110.40).  New bootfiles.
1060    
1061    [Also: John Reppy fixed GC bug 1131.]
1062    
1063    ----------------------------------------------------------------------
1064    Name: Matthias Blume
1065    Date: 2002/05/21 12:35:00 EDT
1066    Tag: blume-20020521-cmdoc
1067    Description:
1068    
1069    CM documentation update.
1070    
1071    ----------------------------------------------------------------------
1072    Name: Matthias Blume
1073    Date: 2002/05/21 10:55:00 EDT
1074    Tag: blume-20020521-misc
1075    Description:
1076    
1077    - John tweaked runtime to be silent on heap export (except when
1078      GC messages are on).
1079    - I added a few more things (cross-compiling versions of CMB) to
1080      config/preloads (as suggestions).
1081    
1082    ----------------------------------------------------------------------
1083    Name: Matthias Blume
1084    Date: 2002/05/20 22:25:00 EDT
1085    Tag: blume-20020520-controls
1086    Description:
1087    
1088    - Added ControlUtil structure to control-lib.cm.
1089    - Use it throughout.
1090    - Used Controls facility to define MLRISC controls (as opposed to
1091      registering MLRISC control ref cells with Controls after the fact)
1092    - Fixed messed-up controls priorities.
1093    
1094    * Removed again all the stuff from config/preloads that one wouldn't
1095      be able to preload at the time the initial heap image is built.
1096      (Many libraries, e.g., CML, do not exist yet at this time.  The
1097       only libraries that can be preloaded via config/preloads are those
1098       that come bundled with the bootfiles.)
1099    
1100    ----------------------------------------------------------------------
1101    Name: Matthias Blume
1102    Date: 2002/05/20 10:59:00 EDT
1103    Tag: blume-20020520-preloads
1104    Description:
1105    
1106    Added a lot of commented-out suggestions for things to be included
1107    in config/preloads.
1108    
1109    ----------------------------------------------------------------------
1110    Name: Allen Leung
1111    Date: 2002/05/18 14:20:00 EDT
1112    Tag: leunga-20020518-mdl
1113    Description:
1114    
1115    o Made the mdl tool stuff compile and run again.
1116    o I've disabled all the stuff that depends on RTL specifications; they
1117      are all badly broken anyway.
1118    
1119    ----------------------------------------------------------------------
1120    Name: Matthias Blume
1121    Date: 2002/05/17 16:49:00 EDT
1122    Tag: blume-20020517-controls
1123    Description:
1124    
1125    0. John Reppy made several modifications to the SML/NJ library.
1126       In particular, there is a shiny new controls-lib.cm.
1127    
1128    1. Pushed new controls interface through compiler so that everything
1129       compiles again.
1130    
1131    2. Added FormatComb and FORMAT_COMB to the CML version of the
1132       SML/NJ library (so that CML compiles again).
1133    
1134    3. Modified init scripts because XXX_DEFAULT environment variables
1135       are no longer with us.  (Boot-time initialization is now done
1136       using the same environment variables that are also used for
1137       startup-time initialization of controls.)
1138    
1139    ----------------------------------------------------------------------
1140    Name: Lal George
1141    Date: 2002/05/15 09:20:10 EDT
1142    Tag: george-20020515-pseudo-op-decls
1143    Description:
1144    
1145            All pseudo-ops emitted before the first segment declaration
1146    such as TEXT, DATA, and BSS directives are assumed to be global
1147    declarations and are emitted first in the assembly file. This is
1148    useful in a number of situations where one has pseudo-ops that are not
1149    specific to any segment, and also works around the constraint that one
1150    cannot have client pseudo-ops in the TEXT segment.
1151    
1152            Because no segment is associated with these declarations it is
1153    an error to allocate any space or objects before the first segment
1154    directive and an exception will be raised. However, we cannot make
1155    this check for client pseudo-ops.
1156    
1157            These top level declarations are a field in the CFG graph_info.
1158    In theory you can continue to add to this field after the CFG has been
1159    built -- provided you know what you are doing;-)
1160    
1161    ----------------------------------------------------------------------
1162    Name: Matthias Blume
1163    Date: 2002/05/13 16:40:00 EDT
1164    Tag: blume-20020513-pp-etc
1165    Description:
1166    
1167    A few minor bugfixes:
1168    
1169      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1170        (Bogus printouts for redefined bindings still occur.  Compiler
1171        bug should no longer occur now.  We need to redo the prettyprinter
1172        from scratch.)
1173    
1174      - CM pathname printer now also adds escape sequences for ( and )
1175    
1176      - commend and docu fixes for ml-nlffi
1177    
1178    ----------------------------------------------------------------------
1179    Name: Matthias Blume
1180    Date: 2002/05/10 16:40:00 EDT
1181    Tag: blume-20020510-erg-textio
1182    Description:
1183    
1184    Applied the following bugfix provided by Emden Gansner:
1185    
1186        Output is corrupted when outputSubstr is used rather than output.
1187    
1188        The problem occurs when a substring
1189    
1190            ss = (s, dataStart, dataLen)
1191    
1192        where dataStart > 0, fills a stream buffer with avail bytes left.
1193        avail bytes of s, starting at index dataStart, are copied into the
1194        buffer, the buffer is flushed, and then the remaining dataLen-avail
1195        bytes of ss are copied into the beginning of the buffer. Instead of
1196        starting this copy at index dataStart+avail in s, the current code
1197        starts the copy at index avail.
1198    
1199        Fix:
1200        In text-io-fn.sml, change line 695 from
1201             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1202        to
1203             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1204    
1205    ----------------------------------------------------------------------
1206    Name: Matthias Blume
1207    Date: 2002/04/12 13:55:00 EDT
1208    Tag: blume-20020412-assyntax
1209    Description:
1210    
1211    1. Grabbed newer assyntax.h from the XFree86 project.
1212    2. Fiddled with how to compile X86.prim.asm without warnings.
1213    3. (Very) Minor cleanup in CM.
1214    
1215    ----------------------------------------------------------------------
1216    Name: Matthias Blume
1217    Date: 2002/04/01 (no joke!) 17:07:00 EST
1218    Tag: blume-20020401-x86div
1219    Description:
1220    
1221    Added full support for div/mod/rem/quot on the x86, using the machine
1222    instruction's two results (without clumsily recomputing the remainder)
1223    directly where appropriate.
1224    
1225    Some more extensive power-of-two support was added to the x86 instruction
1226    selector (avoiding expensive divs, mods, and muls where they can be
1227    replaced with cheaper shifts and masks).  However, this sort of thing
1228    ought to be done earlier, e.g., within the CPS optimizer so that
1229    all architectures benefit from it.
1230    
1231    The compiler compiles to a fixed point, but changes might be somewhat
1232    fragile nevertheless.  Please, report any strange things that you might
1233    see wrt. div/mod/quot/rem...
1234    
1235    ----------------------------------------------------------------------
1236    Name: Matthias Blume
1237    Date: 2002/03/29 17:22:00
1238    Tag: blume-20020329-div
1239    Description:
1240    
1241    Fixed my broken div/mod logic.  Unfortunately, this means that the
1242    inline code for div/mod now has one more comparison than before.
1243    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1244    The problem was with quotient = 0, because that alone does not tell
1245    us which way the rounding went.  One then has to look at whether
1246    remainder and divisor have the same sign...  :(
1247    
1248    Anyway, I replaced the bootfiles with fresh ones...
1249    
1250    ----------------------------------------------------------------------
1251    Name: Matthias Blume
1252    Date: 2002/03/29 14:10:00 EST
1253    Tag: blume-20020329-inlprims
1254    Description:
1255    
1256    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1257    
1258    Primops have changed. This means that the bin/boot-file formats have
1259    changed as well.
1260    
1261    To make sure that there is no confusion, I made a new version.
1262    
1263    
1264    CHANGES:
1265    
1266    * removed REMT from mltree (remainder should never overflow).
1267    
1268    * added primops to deal with divisions of all flavors to the frontend
1269    
1270    * handled these primops all the way through so they map to their respective
1271      MLRISC support
1272    
1273    * used these primops in the implementation of Int, Int32, Word, Word32
1274    
1275    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1276    
1277    * parameterized INLMIN, INLMAX, and INLABS by a numkind
1278    
1279    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1280      floating point
1281    
1282    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1283      and Real (but Real.abs maps to a separate floating-point-only primop)
1284    
1285    
1286    TODO items:
1287    
1288    * Hacked Alpha32 instruction selection, disabling the selection of REMx
1289      instructions because the machine instruction encoder cannot handle
1290      them.  (Hppa, PPC, and Sparc instruction selection did not handle
1291      REM in the first place, and REM is supported by the x86 machine coder.)
1292    
1293    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1294      selection phase.  (The two can be streamlined because the hardware
1295      delivers both quotient and remainder at the same time anyway.)
1296    
1297    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1298      (Currently the behavior is inconsistent both across architectures and
1299      wrt. the draft Basis spec.)
1300    
1301    * Word8 should eventually be handled natively, too.
1302    
1303    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1304      as if there currently is no execution path that could trigger it in
1305      SML/NJ.  (The assumptions underlying functions arith and promotable do not
1306      hold for things like multiplication and division.)
1307    
1308    ----------------------------------------------------------------------
1309    Name: Matthias Blume
1310    Date: 2002/03/27 16:27:00 EST
1311    Tag: blume-20020327-mlrisc-divisions
1312    Description:
1313    
1314    Added support for all four division operations (ML's div, mod, quot,
1315    and rem) to MLRISC.  In the course of doing so, I also rationalized
1316    the naming (no more annoying switch-around of DIV and QUOT), by
1317    parameterizing the operation by div_rounding_mode (which can be either
1318    DIV_TO_ZERO or DIV_TO_NEGINF).
1319    
1320    The generic MLTreeGen functor takes care of compiling all four
1321    operations down to only round-to-zero div.
1322    
1323    Missing pieces:
1324    
1325      * Doing something smarter than relying on MLTreeGen on architectures
1326        like, e.g., the x86 where hardware division delivers both quotient and
1327        remainder at the same time.  With this, the implementation of the
1328        round-to-neginf operations could be further streamlined.
1329    
1330      * Remove inlining support for div/mod/rem from the frontend and replace it
1331        with primops that get carried through to the backend.  Do this for all
1332        int and word types.
1333    
1334    ----------------------------------------------------------------------
1335    Name: Matthias Blume
1336    Date: 2002/03/25 17:25:00 EST
1337    Tag: blume-20020325-divmod
1338    Description:
1339    
1340    I improved (hopefully without breaking them) the implementation of Int.div,
1341    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1342    advantage of the following observations:
1343    
1344      Let  q = x quot y      r = x rem y
1345           d = x div  y      m = x mod y
1346    
1347    where "quot" is the round-to-zero version of integer division that
1348    hardware usually provides.  Then we have:
1349    
1350         r = x - q * y        where neither the * nor the - will overflow
1351         d = if q >= 0 orelse x = q * y then q else q - 1
1352                              where neither the * nor the - will overflow
1353         m = if q >= 0 orelse r = 0 then r else r + y
1354                              where the + will not overflow
1355    
1356    This results in substantial simplification of the generated code.
1357    The following table shows the number of CFG nodes and edges generated
1358    for
1359            fun f (x, y) = x OPER y
1360            (* with OPER \in div, mod, quot, rem *)
1361    
1362    
1363        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1364        --------------------------------------------------------
1365         div |         24 |         39 |         12 |         16
1366         mod |         41 |         71 |         12 |         16
1367        quot |          8 |         10 |          8 |         10
1368         rem |         10 |         14 |          8 |         10
1369    
1370    
1371    ----------------------------------------------------------------------
1372    Name: Matthias Blume
1373    Date: 2002/03/25 22:06:00 EST
1374    Tag: blume-20020325-cprotobug
1375    Description:
1376    
1377    Fixed a bug in cproto (c prototype decoder).
1378    
1379    ----------------------------------------------------------------------
1380    Name: Matthias Blume
1381    Date: 2002/03/25 16:00:00 EST
1382    Tag: blume-20020325-raw-primops
1383    Description:
1384    
1385    I did some cleanup to Allen's new primop code and
1386    replaced yesterday's bootfiles with new ones.
1387    (But they are stored in the same place.)
1388    
1389    ----------------------------------------------------------------------
1390    Name: Matthias Blume
1391    Date: 2002/03/24 22:40:00 EST
1392    Tag: blume-20020324-bootfiles
1393    Description:
1394    
1395    Made the bootfiles that Allen asked for.
1396    
1397    ----------------------------------------------------------------------
1398    Name: Allen Leung
1399    Date: 2002/03/23 15:50:00 EST
1400    Tag: leunga-20020323-flint-cps-rcc-primops
1401    Description:
1402    
1403      1. Changes to FLINT primops:
1404    
1405        (* make a call to a C-function;
1406         * The primop carries C function prototype information and specifies
1407         * which of its (ML-) arguments are floating point. C prototype
1408         * information is for use by the backend, ML information is for
1409         * use by the CPS converter. *)
1410      | RAW_CCALL of { c_proto: CTypes.c_proto,
1411                       ml_args: ccall_type list,
1412                       ml_res_opt: ccall_type option,
1413                       reentrant : bool
1414                     } option
1415       (* Allocate uninitialized storage on the heap.
1416        * The record is meant to hold short-lived C objects, i.e., they
1417        * are not ML pointers.  With the tag, the representation is
1418        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1419        *)
1420      | RAW_RECORD of {tag:bool,sz:int}
1421      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1422    
1423      2.  These CPS primops are now overloaded:
1424    
1425           rawload of {kind:numkind}
1426           rawstore of {kind:numkind}
1427    
1428          The one argument form is:
1429    
1430             rawload {kind} address
1431    
1432          The two argument form is:
1433    
1434             rawload {kind} [ml object, byte-offset]
1435    
1436      3. RAW_CCALL/RCC now takes two extra arguments:
1437    
1438         a. The first is whether the C call is reentrant, i.e., whether
1439            ML state should be saved and restored.
1440         b. The second argument is a string argument specifying the name of
1441            library and the C function.
1442    
1443         These things are currently not handled in the code generator, yet.
1444    
1445      4. In CProto,
1446    
1447         An encoding type of "bool" means "ml object" and is mapped into
1448         C prototype of PTR.  Note that "bool" is different than "string",
1449         even though "string" is also mapped into PTR, because "bool"
1450         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1451    
1452      5. Pickler/unpicker
1453    
1454         Changed to handle RAW_RECORD and newest RAW_CCALL
1455    
1456      6. MLRiscGen,
1457    
1458         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1459         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1460            in the file CodeGen/cpscompile/cps-c-calls.sml
1461    
1462      7. Added the conditional move operator
1463    
1464             condmove of branch
1465    
1466         to cps.  Generation of this is still buggy so it is currently
1467         disabled.
1468    
1469    ----------------------------------------------------------------------
1470    Name: Lal George
1471    Date: 2002/03/22 14:18:25 EST
1472    Tag: george-20020322-cps-branch-prob
1473    Description:
1474    
1475    Implemented the Ball-Larus branch prediction-heuristics, and
1476    incorporated graphical viewers for control flow graphs.
1477    
1478    Ball-Larus Heuristics:
1479    ---------------------
1480    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1481    
1482    By design it uses the Dempster-Shafer theory for combining
1483    probabilities.  For example, in the function:
1484    
1485        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1486    
1487    the ball-larus heuristics predicts that the n=0 is unlikely
1488    (OH-heuristic), and the 'then' branch is unlikely because of the
1489    RH-heuristic -- giving the 'then' branch an even lower combined
1490    probability using the Dempster-Shafer theory.
1491    
1492    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1493    probability of the 'then' branch because of the loop in the else
1494    branch.
1495    
1496    
1497    Graphical Viewing:
1498    ------------------
1499    I merely plugged in Allen's graphical viewers into the compiler. The
1500    additional code is not much. At the top level, saying:
1501    
1502            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1503    
1504    will display the graphical view of the control flow graph just before
1505    back-patching.  daVinci must be in your path for this to work. If
1506    daVinci is not available, then the default viewer can be changed
1507    using:
1508    
1509            Control.MLRISC.getString "viewer"
1510    
1511    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1512    course, these viewers must be in your path.
1513    
1514    The above will display the compilation unit at the level of clusters,
1515    many of which are small, boring, and un-interesting. Also setting:
1516    
1517            Control.MLRISC.getInt "cfg-graphical-view_size"
1518    
1519    will display clusters that are larger than the value set by the above.
1520    
1521    
1522    ----------------------------------------------------------------------
1523    Name: Matthias Blume
1524    Date: 2002/03/21 22:20:00 EST
1525    Tag: blume-20020321-kmp-bugfix
1526    Description:
1527    
1528    Changed the interface to the KMP routine in PreString and fixed
1529    a minor bug in one place where it was used.
1530    
1531    ----------------------------------------------------------------------
1532    Name: Allen Leung
1533    Date: 2002/03/21 20:30:00 EST
1534    Tag: leunga-20020321-cfg
1535    Description:
1536    
1537      Fixed a potential problem in cfg edge splitting.
1538    
1539    ----------------------------------------------------------------------
1540    Name: Allen Leung
1541    Date: 2002/03/21 17:15:00 EST
1542    Tag: leunga-20020321-x86-fp-cfg
1543    Description:
1544    
1545      1. Recoded the buggy parts of x86-fp.
1546    
1547         a. All the block reordering code has been removed.
1548            We now depend on the block placement phases to do this work.
1549    
1550         b. Critical edge splitting code has been simplified and moved into the
1551            CFG modules, as where they belong.
1552    
1553         Both of these were quite buggy and complex.  The code is now much, much
1554         simpler.
1555    
1556      2. X86 backend.
1557    
1558         a. Added instructions for 64-bit support.  Instruction selection for
1559            64-bit has not been committed, however, since that
1560            requires changes to MLTREE which haven't been approved by
1561            Lal and John.
1562    
1563         b. Added support for FUCOMI and FUCOMIP when generating code for
1564            PentiumPro and above.  We only generate these instructions in
1565            the fast-fp mode.
1566    
1567         c. Added cases for JP and JNP in X86FreqProps.
1568    
1569      3. CFG
1570    
1571         CFG now has a bunch of methods for edge splitting and merging.
1572    
1573      4. Machine description.
1574    
1575         John's simplification of MLTREE_BASIS.fcond broke a few machine
1576         description things:
1577    
1578         rtl-build.{sig,sml} and hppa.mdl fixed.
1579    
1580         NOTE: the machine description stuff in the repository is still broken.
1581               Again, I can't put my fixes in because that involves
1582               changes to MLTREE.
1583    
1584    ----------------------------------------------------------------------
1585    Name: Matthias Blume
1586    Date: 2002/03/20 15:55:00 EST
1587    Tag: blume-20020320-kmp
1588    Description:
1589    
1590    Implemented Knuth-Morris-Pratt string matching in PreString and used
1591    it for String.isSubstring, Substring.isSubstring, and
1592    Substring.position.
1593    
1594    (Might need some stress-testing.  Simple examples worked fine.)
1595    
1596    ----------------------------------------------------------------------
1597    Name: Matthias Blume
1598    Date: 2002/03/19 16:37:00 EST
1599    Tag: blume-20020319-witnesses
1600    Description:
1601    
1602    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1603    
1604    This implements a generic mechanism for changing constness qualifiers
1605    anywhere within big C types without resorting to outright "casts".
1606    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1607    modify the constness at the outermost level.)
1608    The implementation of "convert" is based on the idea of "witness"
1609    values -- values that are not used by the operation but whose types
1610    "testify" to their applicability.  On the implementation side, "convert"
1611    is simply a projection (returning its second curried argument).  With
1612    cross-module inlining, it should not result in any machine code being
1613    generated.
1614    
1615    ----------------------------------------------------------------------
1616    Name: Matthias Blume
1617    Date: 2002/03/15 16:40:00 EST
1618    Tag: blume-20020315-basis
1619    Description:
1620    
1621    Provided (preliminary?) implementations for
1622    
1623      {String,Substring}.{concatWith,isSuffix,isSubstring}
1624    
1625    and
1626    
1627      Substring.full
1628    
1629    Those are in the Basis spec but they were missing in SML/NJ.
1630    
1631    ----------------------------------------------------------------------
1632    Name: Matthias Blume
1633    Date: 2002/03/14 21:30:00 EST
1634    Tag: blume-20020314-controls
1635    Description:
1636    
1637    Controls:
1638    ---------
1639    
1640    1. Factored out the recently-added Controls : CONTROLS stuff and put
1641       it into its own library $/controls-lib.cm.  The source tree for
1642       this is under src/smlnj-lib/Controls.
1643    
1644    2. Changed the names of types and functions in this interface, so they
1645       make a bit more "sense":
1646    
1647          module -> registry
1648          'a registry -> 'a group
1649    
1650    3. The interface now deals in ref cells only.  The getter/setter interface
1651       is (mostly) gone.
1652    
1653    4. Added a function that lets one register an already-existing ref cell.
1654    
1655    5. Made the corresponding modifications to the rest of the code so that
1656       everything compiles again.
1657    
1658    6. Changed the implementation of Controls.MLRISC back to something closer
1659       to the original.  In particular, this module (and therefore MLRISC)
1660       does not depend on Controls.  There now is some link-time code in
1661       int-sys.sml that registers the MLRISC controls with the Controls
1662       module.
1663    
1664    CM:
1665    ---
1666    
1667      * One can now specify the lambda-split aggressiveness in init.cmi.
1668    
1669    ----------------------------------------------------------------------
1670    Name: Allen Leung
1671    Date: 2002/03/13 17:30:00 EST
1672    Tag: leunga-20020313-x86-fp-unary
1673    Description:
1674    
1675    Bug fix for:
1676    
1677    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1678    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1679    > - fun f(x,(y,z)) = Real.~ y;
1680    > [autoloading]
1681    > [autoloading done]
1682    >       fchsl   (%eax), 184(%esp)
1683    > Error: MLRisc bug: X86MCEmitter.emitInstr
1684    >
1685    > uncaught exception Error
1686    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1687    
1688    The problem was that the code generator did not generate any fp registers
1689    in this case, and the ra didn't know that it needed to run the X86FP phase to
1690    translate the pseudo fp instruction.   This only happened with unary fp
1691    operators in certain situations.
1692    
1693    ----------------------------------------------------------------------
1694    Name: Matthias Blume
1695    Date: 2002/03/13 14:00:00 EST
1696    Tag: blume-20020313-overload-etc
1697    Description:
1698    
1699    1. Added _overload as a synonym for overload for backward compatibility.
1700       (Control.overloadKW must be true for either version to be accepted.)
1701    
1702    2. Fixed bug in install script that caused more things to be installed
1703       than what was requested in config/targets.
1704    
1705    3. Made CM aware of the (_)overload construct so that autoloading
1706       works.
1707    
1708    ----------------------------------------------------------------------
1709    Name: Matthias Blume
1710    Date: 2002/03/12 22:03:00 EST
1711    Tag: blume-20020312-url
1712    Description:
1713    
1714    Forgot to update BOOT and srcarchiveurl.
1715    
1716    ----------------------------------------------------------------------
1717    Name: Matthias Blume
1718    Date: 2002/03/12 17:30:00 EST
1719    Tag: blume-20020312-version110392
1720    Description:
1721    
1722    Yet another version number bump (because of small changes to the
1723    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1724    
1725    Changes:
1726    
1727      The new pid generation scheme described a few weeks ago was overly
1728      complicated.  I implemented a new mechanism that is simpler and
1729      provides a bit more "stability":  Once CM has seen a compilation
1730      unit, it keeps its identity constant (as long as you do not delete
1731      those crucial CM/GUID/* files).  This means that when you change
1732      an interface, compile, then go back to the old interface, and
1733      compile again, you arrive at the original pid.
1734    
1735      There now also is a mechanism that instructs CM to use the plain
1736      environment hash as a module's pid (effectively making its GUID
1737      the empty string).  For this, "noguid" must be specified as an
1738      option to the .sml file in question within its .cm file.
1739      This is most useful for code that is being generated by tools such
1740      as ml-nlffigen (because during development programmers tend to
1741      erase the tool's entire output directory tree including CM's cached
1742      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1743      revert to the old, broken behavior of SML/NJ, but in specific cases
1744      where there is no danger of interface confusion, its use is ok
1745      (I think).
1746    
1747      ml-nlffigen by default generates "noguid" annotations.  They can be
1748      turned off by specifying -guid in its command line.
1749    
1750    ----------------------------------------------------------------------
1751    Name: Lal George
1752    Date: 2002/03/12 12 14:42:36 EST
1753    Tag: george-20020312-frequency-computation
1754    Description:
1755    
1756    Integrated jump chaining and static block frequency into the
1757    compiler. More details and numbers later.
1758    
1759    ----------------------------------------------------------------------
1760    Name: Lal George
1761    Date: 2002/03/11 11 22:38:53 EST
1762    Tag: george-20020311-jump-chain-elim
1763    Description:
1764    
1765    Tested the jump chain elimination on all architectures (except the
1766    hppa).  This is on by default right now and is profitable for the
1767    alpha and x86, however, it may not be profitable for the sparc and ppc
1768    when compiling the compiler.
1769    
1770    The gc test will typically jump to a label at the end of the cluster,
1771    where there is another jump to an external cluster containing the actual
1772    code to invoke gc. This is to allow factoring of common gc invocation
1773    sequences. That is to say, we generate:
1774    
1775            f:
1776               testgc
1777               ja   L1      % jump if above to L1
1778    
1779            L1:
1780               jmp L2
1781    
1782    
1783    After jump chain elimination the 'ja L1' instructions is converted to
1784    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1785    up being implemented in their long form (if L2 is far away) using:
1786    
1787            jbe     L3      % jump if below or equal to L3
1788            jmp     L2
1789         L3:
1790            ...
1791    
1792    
1793    For large compilation units L2  may be far away.
1794    
1795    
1796    ----------------------------------------------------------------------
1797    Name: Matthias Blume
1798    Date: 2002/03/11 13:30:00 EST
1799    Tag: blume-20020311-mltreeeval
1800    Description:
1801    
1802    A functor parameter was missing.
1803    
1804    ----------------------------------------------------------------------
1805    Name: Allen Leung
1806    Date: 2002/03/11 10:30:00 EST
1807    Tag: leunga-20020311-runtime-string0
1808    Description:
1809    
1810       The representation of the empty string now points to a
1811    legal null terminated C string instead of unit.  It is now possible
1812    to convert an ML string into C string with InlineT.CharVector.getData.
1813    This compiles into one single machine instruction.
1814    
1815    ----------------------------------------------------------------------
1816    Name: Allen Leung
1817    Date: 2002/03/10 23:55:00 EST
1818    Tag: leunga-20020310-x86-call
1819    Description:
1820    
1821       Added machine generation for CALL instruction (relative displacement mode)
1822    
1823    ----------------------------------------------------------------------
1824    Name: Matthias Blume
1825    Date: 2002/03/08 16:05:00
1826    Tag: blume-20020308-entrypoints
1827    Description:
1828    
1829    Version number bumped to 110.39.1.  NEW BOOTFILES!
1830    
1831    Entrypoints: non-zero offset into a code object where execution should begin.
1832    
1833    - Added the notion of an entrypoint to CodeObj.
1834    - Added reading/writing of entrypoint info to Binfile.
1835    - Made runtime system bootloader aware of entrypoints.
1836    - Use the address of the label of the first function given to mlriscGen
1837      as the entrypoint.  This address is currently always 0, but it will
1838      not be 0 once we turn on block placement.
1839    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
1840      with entry points) from mlriscGen.
1841    
1842    ----------------------------------------------------------------------
1843    Name: Allen Leung
1844    Date: 2002/03/07 20:45:00 EST
1845    Tag: leunga-20020307-x86-cmov
1846    Description:
1847    
1848       Bug fixes for CMOVcc on x86.
1849    
1850       1. Added machine code generation for CMOVcc
1851       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1852       3. CMOVcc cannot have an immediate operand as argument.
1853    
1854    ----------------------------------------------------------------------
1855    Name: Matthias Blume
1856    Date: 2002/03/07 16:15:00 EST
1857    Tag: blume-20020307-controls
1858    Description:
1859    
1860    This is a very large but mostly boring patch which makes (almost)
1861    every tuneable compiler knob (i.e., pretty much everything under
1862    Control.* plus a few other things) configurable via both the command
1863    line and environment variables in the style CM did its configuration
1864    until now.
1865    
1866    Try starting sml with '-h' (or, if you are brave, '-H')
1867    
1868    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1869    implements the underlying generic mechanism.
1870    
1871    The interface to some of the existing such facilities has changed somewhat.
1872    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1873    (The getFoo interface is still there for backward-compatibility, but its
1874    use is deprecated.)
1875    
1876    The ml-build script passes -Cxxx=yyy command-line arguments through so
1877    that one can now twiddle the compiler settings when using this "batch"
1878    compiler.
1879    
1880    TODO items:
1881    
1882    We should go through and throw out all controls that are no longer
1883    connected to anything.  Moreover, we should go through and provide
1884    meaningful (and correct!) documentation strings for those controls
1885    that still are connected.
1886    
1887    Currently, multiple calls to Controls.new are accepted (only the first
1888    has any effect).  Eventually we should make sure that every control
1889    is being made (via Controls.new) exactly once.  Future access can then
1890    be done using Controls.acc.
1891    
1892    Finally, it would probably be a good idea to use the getter-setter
1893    interface to controls rather than ref cells.  For the time being, both
1894    styles are provided by the Controls module, but getter-setter pairs are
1895    better if thread-safety is of any concern because they can be wrapped.
1896    
1897    *****************************************
1898    
1899    One bug fix: The function blockPlacement in three of the MLRISC
1900    backpatch files used to be hard-wired to one of two possibilities at
1901    link time (according to the value of the placementFlag).  But (I
1902    think) it should rather sense the flag every time.
1903    
1904    *****************************************
1905    
1906    Other assorted changes (by other people who did not supply a HISTORY entry):
1907    
1908    1. the cross-module inliner now works much better (Monnier)
1909    2. representation of weights, frequencies, and probabilities in MLRISC
1910       changed in preparation of using those for weighted block placement
1911       (Reppy, George)
1912    
1913    ----------------------------------------------------------------------
1914    Name: Lal George
1915    Date: 2002/03/07 14:44:24 EST 2002
1916    Tag: george-20020307-weighted-block-placement
1917    
1918    Tested the weighted block placement optimization on all architectures
1919    (except the hppa) using AMPL to generate the block and edge frequencies.
1920    Changes were required in the machine properties to correctly
1921    categorize trap instructions. There is an MLRISC flag
1922    "weighted-block-placement" that can be used to enable weighted block
1923    placement, but this will be ineffective without block/edge
1924    frequencies (coming soon).
1925    
1926    
1927    ----------------------------------------------------------------------
1928    Name: Lal George
1929    Date: 2002/03/05 17:24:48 EST
1930    Tag: george-20020305-linkage-cluster
1931    
1932    In order to support the block placement optimization, a new cluster
1933    is generated as the very first cluster (called the linkage cluster).
1934    It contains a single jump to the 'real' entry point for the compilation
1935    unit. Block placement has no effect on the linkage cluster itself, but
1936    all the other clusters  have full freedom in the manner in which they
1937    reorder blocks or functions.
1938    
1939    On the x86 the typical linkage code that is generated is:
1940       ----------------------
1941            .align 2
1942       L0:
1943            addl    $L1-L0, 72(%esp)
1944            jmp     L1
1945    
1946    
1947            .align  2
1948       L1:
1949       ----------------------
1950    
1951    72(%esp) is the memory location for the stdlink register. This
1952    must contain the address of the CPS function being called. In the
1953    above example, it contains the address of  L0; before
1954    calling L1 (the real entry point for the compilation unit), it
1955    must contain the address for L1, and hence
1956    
1957            addl $L1-L0, 72(%esp)
1958    
1959    I have tested this on all architectures except the hppa.The increase
1960    in code size is of course negligible
1961    
1962    ----------------------------------------------------------------------
1963    Name: Allen Leung
1964    Date: 2002/03/03 13:20:00 EST
1965    Tag: leunga-20020303-mlrisc-tools
1966    
1967      Added #[ ... ] expressions to mlrisc tools
1968    
1969    ----------------------------------------------------------------------
1970    Name: Matthias Blume
1971    Date: 2002/02/27 12:29:00 EST
1972    Tag: blume-20020227-cdebug
1973    Description:
1974    
1975    - made types in structure C and C_Debug to be equal
1976    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1977    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1978    
1979    ----------------------------------------------------------------------
1980    Name: Matthias Blume
1981    Date: 2002/02/26 12:00:00 EST
1982    Tag: blume-20020226-ffi
1983    Description:
1984    
1985    1. Fixed a minor bug in CM's "noweb" tool:
1986       If numbering is turned off, then truly don't number (i.e., do not
1987       supply the -L option to noweb).  The previous behavior was to supply
1988       -L'' -- which caused noweb to use the "default" line numbering scheme.
1989       Thanks to Chris Richards for pointing this out (and supplying the fix).
1990    
1991    2. Once again, I reworked some aspects of the FFI:
1992    
1993       A. The incomplete/complete type business:
1994    
1995       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
1996         gone!
1997       - ML types representing an incomplete type are now *equal* to
1998         ML types representing their corresponding complete types (just like
1999         in C).  This is still safe because ml-nlffigen will not generate
2000         RTTI for incomplete types, nor will it generate functions that
2001         require access to such RTTI.   But when ML code generated from both
2002         incomplete and complete versions of the C type meet, the ML types
2003         are trivially interoperable.
2004    
2005         NOTE:  These changes restore the full generality of the translation
2006         (which was previously lost when I eliminated functorization)!
2007    
2008       B. Enum types:
2009    
2010       - Structure C now has a type constructor "enum" that is similar to
2011         how the "su" constructor works.  However, "enum" is not a phantom
2012         type because each "T enum" has values (and is isomorphic to
2013         MLRep.Signed.int).
2014       - There are generic access operations for enum objects (using
2015         MLRep.Signed.int).
2016       - ml-nlffigen will generate a structure E_foo for each "enum foo".
2017         * The structure contains the definition of type "mlrep" (the ML-side
2018         representation type of the enum).  Normally, mlrep is the same
2019         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
2020         then mlrep will be defined as a datatype -- thus facilitating
2021         pattern matching on mlrep values.
2022         ("-ec" will be suppressed if there are duplicate values in an
2023          enumeration.)
2024         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
2025         will be generated for each C enum constant xxx.
2026         * Conversion functions m2i and i2m convert between mlrep and
2027         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
2028         * Coversion functions c and ml convert between mlrep and "tag enum".
2029         * Access functions (get/set) fetch and store mlrep values.
2030       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
2031         enumerations are merged into one single enumeration represented by
2032         structure E_'.
2033    
2034    ----------------------------------------------------------------------
2035    Name: Allen Leung
2036    Date: 2002/02/25 04:45:00 EST
2037    Tag: leunga-20020225-cps-spill
2038    
2039    This is a new implementation of the CPS spill phase.
2040    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
2041    In case of problems, replace it with the old file spill.sml
2042    
2043    The current compiler runs into some serious performance problems when
2044    constructing a large record.  This can happen when we try to compile a
2045    structure with many items.  Even a very simple structure like the following
2046    makes the compiler slow down.
2047    
2048        structure Foo = struct
2049           val x_1 = 0w1 : Word32.int
2050           val x_2 = 0w2 : Word32.int
2051           val x_3 = 0w3 : Word32.int
2052           ...
2053           val x_N = 0wN : Word32.int
2054        end
2055    
2056    The following table shows the compile time, from N=1000 to N=4000,
2057    with the old compiler:
2058    
2059    N
2060    1000   CPS 100 spill                           0.04u  0.00s  0.00g
2061           MLRISC ra                               0.06u  0.00s  0.05g
2062              (spills = 0 reloads = 0)
2063           TOTAL                                   0.63u  0.07s  0.21g
2064    
2065    1100   CPS 100 spill                           8.25u  0.32s  0.64g
2066           MLRISC ra                               5.68u  0.59s  3.93g
2067              (spills = 0 reloads = 0)
2068           TOTAL                                   14.71u  0.99s  4.81g
2069    
2070    1500   CPS 100 spill                           58.55u  2.34s  1.74g
2071           MLRISC ra                               5.54u  0.65s  3.91g
2072              (spills = 543 reloads = 1082)
2073           TOTAL                                   65.40u  3.13s  6.00g
2074    
2075    2000   CPS 100 spill                           126.69u  4.84s  3.08g
2076           MLRISC ra                               0.80u  0.10s  0.55g
2077              (spills = 42 reloads = 84)
2078           TOTAL                                   129.42u  5.10s  4.13g
2079    
2080    3000   CPS 100 spill                           675.59u  19.03s  11.64g
2081           MLRISC ra                               2.69u  0.27s  1.38g
2082              (spills = 62 reloads = 124)
2083           TOTAL                                   682.48u  19.61s  13.99g
2084    
2085    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
2086           MLRISC ra                               4.96u  0.27s  2.72g
2087              (spills = 85 reloads = 170)
2088           TOTAL                                   2375.26u  57.21s  48.00g
2089    
2090    As you can see the old cps spill module suffers from some serious
2091    performance problem.  But since I cannot decipher the old code fully,
2092    instead of patching the problems up, I'm reimplementing it
2093    with a different algorithm.  The new code is more modular,
2094    smaller when compiled, and substantially faster
2095    (O(n log n) time and O(n) space).  Timing of the new spill module:
2096    
2097    4000  CPS 100 spill                           0.02u  0.00s  0.00g
2098          MLRISC ra                               0.25u  0.02s  0.15g
2099             (spills=1 reloads=3)
2100          TOTAL                                   7.74u  0.34s  1.62g
2101    
2102    Implementation details:
2103    
2104    As far as I can tell, the purpose of the CPS spill module is to make sure the
2105    number of live variables at any program point (the bandwidth)
2106    does not exceed a certain limit, which is determined by the
2107    size of the spill area.
2108    
2109    When the bandwidth is too large, we decrease the register pressure by
2110    packing live variables into spill records.  How we achieve this is
2111    completely different than what we did in the old code.
2112    
2113    First, there is something about the MLRiscGen code generator
2114    that we should be aware of:
2115    
2116    o MLRiscGen performs code motion!
2117    
2118       In particular, it will move floating point computations and
2119       address computations involving only the heap pointer to
2120       their use sites (if there is only a single use).
2121       What this means is that if we have a CPS record construction
2122       statement
2123    
2124           RECORD(k,vl,w,e)
2125    
2126       we should never count the new record address w as live if w
2127       has only one use (which is often the case).
2128    
2129       We should do something similar to floating point, but the transformation
2130       there is much more complex, so I won't deal with that.
2131    
2132    Secondly, there are now two new cps primops at our disposal:
2133    
2134     1. rawrecord of record_kind option
2135        This pure operator allocates some uninitialized storage from the heap.
2136        There are two forms:
2137    
2138         rawrecord NONE [INT n]  allocates a tagless record of length n
2139         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2140                                     and initializes the tag.
2141    
2142     2. rawupdate of cty
2143          rawupdate cty (v,i,x)
2144          Assigns to x to the ith component of record v.
2145          The storelist is not updated.
2146    
2147    We use these new primops for both spilling and increment record construction.
2148    
2149     1. Spilling.
2150    
2151        This is implemented with a linear scan algorithm (but generalized
2152        to trees).  The algorithm will create a single spill record at the
2153        beginning of the cps function and use rawupdate to spill to it,
2154        and SELECT or SELp to reload from it.  So both spills and reloads
2155        are fine-grain operations.  In contrast, in the old algorithm
2156        "spills" have to be bundled together in records.
2157    
2158        Ideally, we should sink the spill record construction to where
2159        it is needed.  We can even split the spill record into multiple ones
2160        at the places where they are needed.  But CPS is not a good
2161        representation for global code motion, so I'll keep it simple and
2162        am not attempting this.
2163    
2164     2. Incremental record construction (aka record splitting).
2165    
2166        Long records with many component values which are simulatenously live
2167        (recall that single use record addresses are not considered to
2168         be live) are constructed with rawrecord and rawupdate.
2169        We allocate space on the heap with rawrecord first, then gradually
2170        fill it in with rawupdate.  This is the technique suggested to me
2171        by Matthias.
2172    
2173        Some restrictions on when this is applicable:
2174        1. It is not a VECTOR record.  The code generator currently does not handle
2175           this case. VECTOR record uses double indirection like arrays.
2176        2. All the record component values are defined in the same "basic block"
2177           as the record constructor.  This is to prevent speculative
2178           record construction.
2179    
2180    ----------------------------------------------------------------------
2181    Name: Allen Leung
2182    Date: 2002/02/22 01:02:00 EST
2183    Tag: leunga-20020222-mlrisc-tools
2184    
2185    Minor bug fixes in the parser and rewriter
2186    
2187    ----------------------------------------------------------------------
2188    Name: Allen Leung
2189    Date: 2002/02/21 20:20:00 EST
2190    Tag: leunga-20020221-peephole
2191    
2192    Regenerated the peephole files.  Some contained typos in the specification
2193    and some didn't compile because of pretty printing bugs in the old version
2194    of 'nowhere'.
2195    
2196    ----------------------------------------------------------------------
2197    Name: Allen Leung
2198    Date: 2002/02/19 20:20:00 EST
2199    Tag: leunga-20020219-mlrisc-tools
2200    Description:
2201    
2202       Minor bug fixes to the mlrisc-tools library:
2203    
2204       1.  Fixed up parsing colon suffixed keywords
2205       2.  Added the ability to shut the error messages up
2206       3.  Reimplemented the pretty printer and fixed up/improved
2207           the pretty printing of handle and -> types.
2208       4.  Fixed up generation of literal symbols in the nowhere tool.
2209       5.  Added some SML keywords to to sml.sty
2210    
2211    ----------------------------------------------------------------------
2212    Name: Matthias Blume
2213    Date: 2002/02/19 16:20:00 EST
2214    Tag: blume-20020219-cmffi
2215    Description:
2216    
2217    A wild mix of changes, some minor, some major:
2218    
2219    * All C FFI-related libraries are now anchored under $c:
2220        $/c.cm      --> $c/c.cm
2221        $/c-int.cm  --> $c/internals/c-int.cm
2222        $/memory.cm --> $c/memory/memory.cm
2223    
2224    * "make" tool (in CM) now treats its argument pathname slightly
2225      differently:
2226        1. If the native expansion is an absolute name, then before invoking
2227           the "make" command on it, CM will apply OS.Path.mkRelative
2228           (with relativeTo = OS.FileSys.getDir()) to it.
2229        2. The argument will be passed through to subsequent phases of CM
2230           processing without "going native".  In particular, if the argument
2231           was an anchored path, then "make" will not lose track of that anchor.
2232    
2233    * Compiler backends now "know" their respective C calling conventions
2234      instead of having to be told about it by ml-nlffigen.  This relieves
2235      ml-nlffigen from one of its burdens.
2236    
2237    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
2238    
2239    * Export C_DEBUG and C_Debug from $c/c.cm.
2240    
2241    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
2242      subtyping relationship between incomplete pointers and their complete
2243      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
2244      with the convention of instantiating 'o with ('t, 'c) obj whenever
2245      the pointer target type is complete.  In the incomplete case, 'o
2246      will be instantiated with some "'c iobj" -- a type obtained by
2247      using one of the functors PointerToIncompleteType or PointerToCompleteType.
2248    
2249      Operations that work on both incomplete and complete pointer types are
2250      typed as taking an 'o ptr while operations that require the target to
2251      be known are typed as taking some ('t, 'c) obj ptr.
2252    
2253      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
2254      where void is an eqtype without any values.  This makes it possible
2255      to work on voidptr values using functions meant to operate on light
2256      incomplete pointers.
2257    
2258    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
2259      been vastly simplified.
2260    
2261    ----------------------------------------------------------------------
2262    Name: Matthias Blume
2263    Date: 2002/02/19 10:48:00 EST
2264    Tag: blume-20020219-pqfix
2265    Description:
2266    
2267    Applied Chris Okasaki's bug fix for priority queues.
2268    
2269    ----------------------------------------------------------------------
2270    Name: Matthias Blume
2271    Date: 2002/02/15 17:05:00
2272    Tag: Release_110_39
2273    Description:
2274    
2275    Last-minute retagging is becoming a tradition... :-(
2276    
2277    This is the working release 110.39.
2278    
2279    ----------------------------------------------------------------------
2280    Name: Matthias Blume
2281    Date: 2002/02/15 16:00:00 EST
2282    Tag: Release_110_39-orig
2283    Description:
2284    
2285    Working release 110.39.  New bootfiles.
2286    
2287    (Update: There was a small bug in the installer so it wouldn't work
2288    with all shells.  So I retagged. -Matthias)
2289    
2290    ----------------------------------------------------------------------
2291    Name: Matthias Blume
2292    Date: 2002/02/15 14:17:00 EST
2293    Tag: blume-20020215-showbindings
2294    Description:
2295    
2296    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
2297    the latter can be useful for exploring what bindings are available at
2298    the interactive prompt.  (The first function returns only the list
2299    of symbols that are really bound, the second prints those but also the
2300    ones that CM's autoloading mechanism knows about.)
2301    
2302    ----------------------------------------------------------------------
2303    Name: Matthias Blume
2304    Date: 2002/02/15 12:08:00 EST
2305    Tag: blume-20020215-iptrs
2306    Description:
2307    
2308    Two improvements to ml-nlffigen:
2309    
2310      1. Write files only if they do not exist or if their current contents
2311         do not coincide with what's being written.  (That is, avoid messing
2312         with the time stamps unless absolutely necessary.)
2313    
2314      2. Implement a "repository" mechanism for generated files related
2315         to "incomplete pointer types".   See the README file for details.
2316    
2317    ----------------------------------------------------------------------
2318    Name: Matthias Blume
2319    Date: 2002/02/14 11:50:00 EST
2320    Tag: blume-20020214-quote
2321    Description:
2322    
2323    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
2324    because of the new and improved tag generation scheme.  (Thanks to Allen
2325    Leung for pointing it out.)
2326    
2327    ----------------------------------------------------------------------
2328    Name: Lal George
2329    Date: 2002/02/14 09:55:27 EST 2002
2330    Tag: george-20020214-isabelle-bug
2331    Description:
2332    
2333    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2334    of Isabelle on the x86.
2335    
2336    From Allen:
2337    -----------
2338     I've found the problem:
2339    
2340         in ra-core.sml, I use the counter "blocked" to keep track of the
2341         true number of elements in the freeze queue.  When the counter goes
2342         to zero, I skip examining the queue.  But I've messed up the
2343         bookkeeping in combine():
2344    
2345             else ();
2346             case !ucol of
2347               PSEUDO => (if !cntv > 0 then
2348                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2349                                        ^^^^^^^^^^^^^^^^^^^^^^^
2350                      moveu := mergeMoveList(!movev, !moveu)
2351                     )
2352                  else ();
2353    
2354         combine() is called to coalesce two nodes u and v.
2355         I think I was thinking that if the move counts of u and v are both
2356         greater than zero then after they are coalesced then one node is
2357         removed from the freeze queue.  Apparently I was thinking that
2358         both u and v are of low degree, but that's clearly not necessarily true.
2359    
2360    
2361    02/12/2002:
2362        Here's the patch.  HOL now compiles.
2363    
2364        I don't know how this impact on performance (compile
2365        time or runtime).  This bug caused the RA (especially on the x86)
2366        to go thru the potential spill phase when there are still nodes on the
2367        freeze queue.
2368    
2369    
2370    
2371    
2372    ----------------------------------------------------------------------
2373    Name: Matthias Blume
2374    Date: 2002/02/13 22:40:00 EST
2375    Tag: blume-20020213-fptr-rtti
2376    Description:
2377    
2378    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2379    updates.
2380    
2381    ----------------------------------------------------------------------
2382    Name: Matthias Blume
2383    Date: 2002/02/13 16:41:00 EST
2384    Tag: blume-20020213-cmlpq
2385    Description:
2386    
2387    Added new priority queue export symbols (which have just been added to
2388    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2389    not compile and the installer would choke.)
2390    
2391    ----------------------------------------------------------------------
2392    Name: Matthias Blume
2393    Date: 2002/02/13 16:15:00 EST
2394    Tag: blume-20020213-various
2395    Description:
2396    
2397    1. More tweaks to ml-nlffigen:
2398    
2399       - better internal datastructures (resulting in slight speedup)
2400       - "-match" option requires exact match
2401       - "localized" gensym counters (untagged structs/unions nested within
2402         other structs/unions or within typedefs get a fresh counter; their
2403         tag will be prefixed by a concatenation of their parents' tags)
2404       - bug fixes (related to calculation of transitive closure of types
2405         to be included in the output)
2406    
2407    2. Minor Basis updates:
2408    
2409       - added implementations for List.collate and Option.app
2410    
2411    ----------------------------------------------------------------------
2412    Name: Matthias Blume
2413    Date: 2002/02/11 15:55:00 EST
2414    Tag: blume-20020211-gensym
2415    Description:
2416    
2417    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2418    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2419    names (ML structure names that correspond to unnamed C structs, unions,
2420    and enums), so that separate runs of ml-nlffigen do not clash.
2421    
2422    ----------------------------------------------------------------------
2423    Name: Matthias Blume
2424    Date: 2002/02/11 12:05:00 EST
2425    Tag: blume-20020211-gensml
2426    Description:
2427    
2428    A quick fix for a problem with GenSML (in the pgraph-util library):
2429    Make generation of toplevel "local" optional.  (Strictly speaking,
2430    signature definitions within "local" are not legal SML.)
2431    
2432    Other than that: updates to INSTALL and cm/TODO.
2433    
2434    ----------------------------------------------------------------------
2435    Name: Matthias Blume
2436    Date: 2002/02/08 15:00:00 EST
2437    Tag: blume-20020208-uniquepid
2438    Description:
2439    
2440    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2441    
2442    1. The installer (config/install.sh) has gotten smarter:
2443    
2444         - Configuration options are a bit easier to specify now
2445           (in config/targets).
2446         - Bug in recognizing .tar.bz2 files fixed.
2447         - Installer automatically resolves dependencies between
2448           configuration options (e.g., if you ask for eXene, you will
2449           also get cml -- regardless whether you asked for it or not).
2450         - Installer can run in "quieter mode" by setting the environment
2451           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2452           "completely silent", though.
2453         - Build HashCons library as part of smlnj-lib.
2454    
2455    2. A new scheme for assigning persistent identifiers to compilation
2456       units (and, by extension, to types etc.) has been put into place.
2457       This fixes a long-standing bug where types and even dynamic values
2458       can get internally confused, thereby compromising type safety
2459       (abstraction) and dynamic correctness.  See
2460    
2461         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2462    
2463       for an example of how things could go wrong until now.
2464    
2465       The downside of the new scheme is that pids are not quite as
2466       persistent as they used to be: CM will generate a fresh pid
2467       for every compilation unit that it thinks it sees for the first
2468       time.  That means that if you compile starting from a clean, fresh
2469       source tree at two different times, you end up with different
2470       binaries.
2471    
2472       Cutoff recompilation, however, has not been compromised because
2473       CM keeps pid information in special caches between runs.
2474    
2475    ----------------------------------------------------------------------
2476    Name: Lal George
2477    Date: 2002/02/07 15:34:13 EST 2002
2478    Tag: <none>
2479    Description:
2480    
2481    Compilers that generate assembly code may produce  global labels
2482    whose value is resolved at link time. The various peephole optimization
2483    modules did not take this in account.
2484    
2485    TODO. The Labels.addrOf function should really return an option
2486    type so that clients are forced to deal with this issue, rather
2487    than an exception being raised.
2488    
2489    ----------------------------------------------------------------------
2490    Name: Lal George
2491    Date: 2002/02/06 13:55:02 EST
2492    Tag: george-20020206-ra-breakup
2493    Description:
2494    
2495    1. A bug fix from Allen.
2496    
2497        A typo causes extra fstp %st(0)'s to be generated at compensation
2498        edges, which might cause stack underflow traps at runtime.  This
2499        occurs in fft where there are extraneous fstps right before the 'into'
2500        trap instruction (in this case they are harmless since none of the
2501        integers overflow.)
2502    
2503    2. Pulled out various utility modules that were embedded in the modules
2504       of the register allocator. I need these modules for other purposes, but
2505       they are not complete enough to put into a library (just yet).
2506    ----------------------------------------------------------------------
2507    Name: Matthias Blume
2508    Date: 2002/01/31 16:05:00 EST
2509    Tag: blume-20020131-sparc-ccalls
2510    Description:
2511    
2512    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2513       of extra stack space by mistake.  Fixed.
2514    
2515    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2516    
2517    ----------------------------------------------------------------------
2518    Name: Allen Leung
2519    Date: 2002/01/30
2520    Tag: leunga-20020130-nowhere-bug-fix
2521    Description:
2522    
2523       MLRISC bug fixes:
2524       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2525       2. MachineInt.fromString was negating its value.
2526    
2527    ----------------------------------------------------------------------
2528    Name: Matthias Blume
2529    Date: 2002/01/29
2530    Tag: blume-20020129-INSTALL
2531    Description:
2532    
2533    - Added somewhat detailed installation instructions (file INSTALL).
2534    - Fixed curl-detection bug in config/install.sh.
2535    - It is now possible to select the URL getter using the URLGETTER
2536      environment variable:
2537    
2538          not set / "unknown"      --> automatic detection (script tries wget,
2539                                       curl, and lynx)
2540          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2541                                       how to properly invoke them)
2542          other                    --> use $URLGETTER directly, it must take
2543                                       precisely two command-line arguments
2544                                       (source URL and destination file name)
2545    
2546    ----------------------------------------------------------------------
2547    Name: Matthias Blume
2548    Date: 2002/01/28
2549    Tag: blume-20020128-sparc-ccalls
2550    Description:
2551    
2552    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2553    - Make use of the allocParam argument in sparc-c-calls.
2554    
2555    ----------------------------------------------------------------------
2556    Name: Matthias Blume
2557    Date: 2002/01/28
2558    Tag: blume-20020128-allocParam
2559    Description:
2560    
2561    John Reppy:  Changes c-calls API to accept client-callback for
2562    allocating extra stack space.
2563    me: Corresponding changes to mlriscGen (using a dummy argument that
2564        does not change the current behavior).
2565    
2566    ----------------------------------------------------------------------
2567    Name: Matthias Blume
2568    Date: 2002/01/28 12:00:00
2569    Tag: Release_110_38
2570    Description:
2571    
2572    This time for real!!!
2573    
2574    ----------------------------------------------------------------------
2575    Name: Matthias Blume
2576    Date: 2002/01/28 10:56:00 EST
2577    Tag: blume-20020128-retraction
2578    Description:
2579    
2580    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
2581       with blume-Release_110_38-retracted.)
2582    
2583    1. Fixed a problem with incorrect rounding modes in real64.sml.
2584       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2585    
2586    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2587       and enums fixed.  The naming of corresponding ML identifiers should
2588       now be consistent again.
2589    
2590    ----------------------------------------------------------------------
2591    Name: Allen Leung
2592    Date: 2002/01/27
2593    Tag: leunga-20020127-nowhere
2594    Description:
2595    
2596       Added a target called nowhere in the configuration scripts.
2597       Enabling this will build the MLRISC 'nowhere' tool (for translating
2598       programs with where-clauses into legal SML code) during installation.
2599    
2600    ----------------------------------------------------------------------
2601    Name: Matthias Blume
2602    Date: 2002/01/25 21:27:00 EST
2603    Tag: blume-Release_110_38-retracted
2604    Description:
2605    
2606    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
2607    
2608    README will be added later.
2609    
2610    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
2611    (Release_110_38) removed.  Reason: Last-minute bug fixes.
2612    
2613    ----------------------------------------------------------------------
2614    Name: Matthias Blume
2615    Date: 2002/01/25
2616    Tag: blume-20020125-ffi
2617    Description:
2618    
2619    A large number of tweaks and improvements to ml-nlffi-lib and
2620    ml-nlffigen:
2621    
2622       - ML represenation types have been streamlined
2623       - getter and setter functions work with concrete values, not abstract
2624         ones where possible
2625       - ml-nlffigen command line more flexible (see README file there)
2626       - some bugs have been fixed (hopefully)
2627    
2628    ----------------------------------------------------------------------
2629    Name: Lal George
2630    Date: 2002/01/24
2631    Tag: george-20020124-risc-ra-interface
2632    Description:
2633    
2634       There is a dramatic simplification in the interface to the
2635       register allocator for RISC architectures as a result of making
2636       parallel copy instructions explicit.
2637    
2638    ----------------------------------------------------------------------
2639    Name: Matthias Blume
2640    Date: 2002/01/22
2641    Tag: blume-20020122-x86-ccalls
2642    Description:
2643    
2644    Bug fix for c-calls on x86 (having to do with how char- and
2645    short-arguments are being handled).
2646    
2647    ----------------------------------------------------------------------
2648    Name: Matthias Blume
2649    Date: 2002/01/21
2650    Tag: blume-20020121-ff
2651    Description:
2652    
2653    Another day of fiddling with the FFI...
2654    
2655    1. Bug fix/workaround:  CKIT does not complain about negative array
2656       dimensions, so ml-nlffigen has to guard itself against this possibility.
2657       (Otherwise a negative dimension would send it into an infinite loop.)
2658    
2659    2. Some of the abstract types (light objects, light pointers, most "base"
2660       types) in structure C are now eqtypes.
2661    
2662    3. Added constructors and test functions for NULL function pointers.
2663    
2664    ----------------------------------------------------------------------
2665    Name: Matthias Blume
2666    Date: 2002/01/18
2667    Tag: blume-20020118-ready-for-new-release
2668    Description:
2669    
2670    Made config/srcarchiveurl point to a new place.  (Will provide boot
2671    files shortly.)
2672    
2673    Maybe we christen this to be 110.38?
2674    
2675    ----------------------------------------------------------------------
2676    Name: Matthias Blume
2677    Date: 2002/01/18
2678    Tag: blume-20020118-more-ffifiddle
2679    Description:
2680    
2681    Today's FFI fiddling:
2682    
2683      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
2684        C.Set.* that use concrete (MLRep.*) arguments and results instead
2685        of abstract ones.
2686    
2687      - Provided word-style bit operations etc. for "int" representation
2688        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
2689        and Long.
2690    
2691    ----------------------------------------------------------------------
2692    Name: Matthias Blume
2693    Date: 2002/01/18
2694    Tag: blume-20020118-use-x86-fp
2695    Description:
2696    
2697    Now that x86-fast-fp seems to be working, I turned it back on again
2698    by default.  (Seems to work fine now, even with the FFI.)
2699    
2700    Other than that, I added some documentation about the FFI to
2701    src/ml-nlffigen/README and updated the FFI test examples in
2702    src/ml-nlffi-lib/Tests/*.
2703    
2704    ----------------------------------------------------------------------
2705    Name: Allen Leung
2706    Date: 2002/01/17
2707    Tag: leunga-20020117-x86-fast-fp-call
2708    Description:
2709    
2710       1. Fixed a problem with handling return fp values when x86's fast fp
2711          mode is turned on.
2712    
2713       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
2714          of %f32.
2715    
2716       3. Added a constructor INT32lit to the ast of MLRISC tools.
2717    
2718    ----------------------------------------------------------------------
2719    Name: Matthias Blume
2720    Date: 2002/01/16
2721    Tag: blume-20020116-ffifiddle
2722    Description:
2723    
2724    More fiddling with the FFI interface:
2725    
2726     - Make constness 'c instead of rw wherever possible.  This eliminates
2727       the need for certain explicit coercions.  (However, due to ML's
2728       value polymorphism, there will still be many cases where explicit
2729       coercions are necessary.  Phantom types are not the whole answer
2730       to modeling a subtyping relationship in ML.)
2731    
2732     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
2733    
2734     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
2735       clumsy workaround for varargs problem.)
2736    
2737    ----------------------------------------------------------------------
2738    Name: Lal George
2739    Date: 2002/01/15
2740    Tag: <none>
2741    Description:
2742    
2743    1. Since COPY instructions are no longer native to the architecture,
2744       a generic functor can be used to implement the expandCopies function.
2745    
2746    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
2747       TEXT segment.
2748    
2749    ----------------------------------------------------------------------
2750    Name: Matthias Blume
2751    Date: 2002/01/15
2752    Tag: blume-20020115-ffiupdates
2753    Description:
2754    
2755    1. Fix for bug resulting in single-precision float values being returned
2756       incorrectly from FFI calls.
2757    
2758    2. Small modifications to C FFI API:
2759    
2760        - memory-allocation routines return straight objects (no options)
2761          and raise an exception in out-of-memory situations
2762        - unsafe extensions to cast between function pointers and pointers
2763          from/to ints
2764        - added structure C_Debug as an alternative to structure C where
2765          pointer-dereferencing (|*| and |*!) always check for null-pointers
2766        - added open_lib' to DynLinkage;  open_lib' works like open_lib
2767          but also takes a (possibly empty) list of existing library handles
2768          that the current library depends on
2769    
2770    ----------------------------------------------------------------------
2771    Name: Matthias Blume
2772    Date: 2002/01/10
2773    Tag: blume-20020110-newffigen
2774    Description:
2775    
2776    1. Updates to portable graph code.
2777    
2778    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
2779       more scalable now so that even huge interfaces such as the one
2780       for GTK compile in finite time and space. :-)
2781       See src/ml-nlffigen/README for details on what's new.
2782    
2783    ----------------------------------------------------------------------
2784    Name: Lal George
2785    Date: 2001/01/09 14:31:35 EST 2002
2786    Tag: george-20011206-rm-native-copy
2787    Description:
2788    
2789            Removed the native COPY and FCOPY instructions
2790            from all the architectures and replaced it with the
2791            explicit COPY instruction from the previous commit.
2792    
2793            It is now possible to simplify many of the optimizations
2794            modules that manipulate copies. This has not been
2795            done in this change.
2796    
2797    ----------------------------------------------------------------------
2798    Name: Lal George
2799    Date: 2001/12/06 16:50:13 EST 2001
2800    Tag: george-20011206-mlrisc-instruction
2801    Description:
2802    
2803    Changed the representation of instructions from being fully abstract
2804    to being partially concrete. That is to say:
2805    
2806      from
2807            type instruction
2808    
2809      to
2810            type instr                              (* machine instruction *)
2811    
2812            datatype instruction =
2813                LIVE of {regs: C.cellset, spilled: C.cellset}
2814              | KILL of {regs: C.cellset, spilled: C.cellset}
2815              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
2816              | ANNOTATION of {i: instruction, a: Annotations.annotation}
2817              | INSTR of instr
2818    
2819    This makes the handling of certain special instructions that appear on
2820    all architectures easier and uniform.
2821    
2822    LIVE and KILL say that a list of registers are live or killed at the
2823    program point where they appear. No spill code is generated when an
2824    element of the 'regs' field is spilled, but the register is moved to
2825    the 'spilled' (which is present, more for debugging than anything else).
2826    
2827    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
2828    We used to generate:
2829    
2830            DEFFREG f1
2831            f1 := f2 + f3
2832            trapb
2833    
2834    but now generate:
2835    
2836            f1 := f2 + f3
2837            trapb
2838            LIVE {regs=[f1,f2,f3], spilled=[]}
2839    
2840    Furthermore, the DEFFREG (hack) required that all floating point instruction
2841    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
2842    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
2843    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
2844    this architecture).
2845    
2846    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
2847    all the architectures. This will result in further simplification of the
2848    register allocator that must be aware of them for coalescing purposes, and
2849    will also simplify certain aspects of the machine description that provides
2850    callbacks related to parallel copies.
2851    
2852    ANNOTATION should be obvious, and now INSTR represents the honest to God
2853    machine instruction set!
2854    
2855    The <arch>/instructions/<arch>Instr.sml files define certain utility
2856    functions for making porting easier -- essentially converting upper case
2857    to lower case. All machine instructions (of type instr) are in upper case,
2858    and the lower case form generates an MLRISC instruction. For example on
2859    the alpha we have:
2860    
2861      datatype instr =
2862         LDA of {r:cell, b:cell, d:operand}
2863       | ...
2864    
2865      val lda : {r:cell, b:cell, d:operand} -> instruction
2866        ...
2867    
2868    where lda is just (INSTR o LDA), etc.
2869    
2870    ----------------------------------------------------------------------
2871    Name: Matthias Blume
2872    Date: 2001/11/22 21:40:00 EST
2873    Tag: Release_110_37
2874    Description:
2875    
2876    Release 110.37.  This time for real.
2877    
2878    ----------------------------------------------------------------------
2879    Name: Matthias Blume
2880    Date: 2001/11/21 16:35:00 EST
2881    Tag: blume-20011121-foot-in-mouth
2882    Description:
2883    
2884    Removed the "Release_110_37" tag because of a serious bug.
2885    This will be re-tagged once the bug is fixed.
2886    
2887    ----------------------------------------------------------------------
2888    Name: Matthias Blume
2889    Date: 2001/11/21 16:14:00 EST
2890    Tag: blume-20011121-forgottenfile
2891    Description:
2892    
2893    Forgot to add a file.  (Just a .tex-file -- part of
2894    the CM manual source.)
2895    
2896    ----------------------------------------------------------------------
2897    Name: Matthias Blume
2898    Date: 2001/11/21 16:10:00 EST
2899    Tag: blume-20011121-invalid_110_37
2900    Description:
2901    
2902    Note: I removed the original tag "Release_110_37" from this commit
2903          because we found a serious bug in all non-x86 backends.
2904          - Matthias
2905    
2906    1. Modifications to the SML/NJ code generator and to the runtime system
2907       so that code object name strings are directly inserted into code
2908       objects at code generation time.  The only business the runtime system
2909       has with this is now to read the name strings on occasions.
2910       (The encoding of the name string has also changed somewhat.)
2911    
2912    2. CM now implements a simple "set calculus" for specifying export lists.
2913       In particular, it is now possible to refer to the export lists of
2914       other libraries/groups/sources and form unions as well as differences.
2915       See the latest CM manual for details.
2916    
2917    3. An separate notion of "proxy" libraries has again be eliminated from
2918       CM's model.  (Proxy libraries are now simply a special case of using
2919       the export list calculus.)
2920    
2921    4. Some of the existing libraries now take advantage of the new set
2922       calculus.
2923       (Notice that not all libraries have been converted because some
2924       of the existing .cm-files are supposed to be backward compatible
2925       with 110.0.x.)
2926    
2927    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
2928       "General.exnMessage"!  The former relies on a certain hook to be
2929       initialized, and that often does not happen in the stand-alone case.)
2930    
2931    ----------------------------------------------------------------------
2932    Name: Lal George
2933    Date: 2001/11/21  13:56:18 EST
2934    Tag: george-2001121-pseudo-ops
2935    Description:
2936    
2937      Implemented a complete redesign of MLRISC pseudo-ops. Now there
2938      ought to never be any question of incompatabilities with
2939      pseudo-op syntax expected by host assemblers.
2940    
2941      For now, only modules supporting GAS syntax are implemented
2942      but more should follow, such as MASM, and vendor assembler
2943      syntax, e.g. IBM as, Sun as, etc.
2944    
2945    ----------------------------------------------------------------------
2946    Name: Matthias Blume
2947    Date: 2001/11/14 11:52:00 EST
2948    Tag: blume-20011114-srcname
2949    Description:
2950    
2951    1. Routed the name of the current source file to mlriscgen where it
2952       should be directly emitted into the code object.  (This last part
2953       is yet to be done.)
2954    
2955    2. Some cleanup of the pgraph code to make it match the proposal that
2956       I put out the other day.  (The proposal notwithstanding, things are
2957       still in flux here.)
2958    
2959    ----------------------------------------------------------------------
2960    Name: Lal George
2961    Date: 2001/11/14 09:44:04 EST
2962    Tag:
2963    Description:
2964    
2965      Fix for a backpatching bug reported by Allen.
2966    
2967      Because the boundary between short and long span-dependent
2968      instructions is +/- 128, there are an astounding number of
2969      span-dependent instructions whose size is over estimated.
2970    
2971      Allen came up with the idea of letting the size of span
2972      dependent instructions be non-monotonic, for a maxIter
2973      number of times, after which the size must be monotonically
2974      increasing.
2975    
2976      This table shows the number of span-dependent instructions
2977      whose size was over-estimated as a function of maxIter, for the
2978      file Parse/parse/ml.grm.sml:
2979    
2980         maxIter            # of instructions:
2981            10                      687
2982            20                      438
2983            30                      198
2984            40                        0
2985    
2986      In compiling the compiler, there is no significant difference in
2987      compilation speed between maxIter=10 and maxIter=40. Actually,
2988      my measurements showed that maxIter=40 was a tad faster than
2989      maxIter=10! Also 96% of the  files in the compiler reach a fix
2990      point within 13 iterations, so fixing maxIter at 40, while high,
2991      is okay.
2992    
2993    ----------------------------------------------------------------------
2994    Name: Matthias Blume
2995    Date: 2001/10/31 15:25:00 EST
2996    Tag: blume-20011031-pgraph
2997    Description:
2998    
2999    CKIT:
3000    * Changed the "Function" constructor of type Ast.ctype to carry optional
3001      argument identifiers.
3002    * Changed the return type of TypeUtil.getFunction accordingly.
3003    * Type equality ignores the argument names.
3004    * TypeUtil.composite tries to preserve argument names but gives up quickly
3005      if there is a mismatch.
3006    
3007    installation script:
3008    * attempts to use "curl" if available (unless "wget" is available as well)
3009    
3010    CM:
3011    * has an experimental implementation of "portable graphs" which I will
3012      soon propose as an implementation-independent library format
3013    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
3014    
3015    NLFFI-LIB:
3016    * some cleanup (all cosmetic)
3017    
3018    NLFFIGEN:
3019    * temporarily disabled the mechanism that suppresses ML output for
3020      C definitions whose identifiers start with an underscore character
3021    * generate val bindings for enum constants
3022    * user can request that only one style (light or heavy) is being used;
3023      default is to use both (command-line arguments: -heavy and -light)
3024    * fixed bug in handling of function types involving incomplete pointers
3025    * generate ML entry points that take record arguments (i.e., using
3026      named arguments) for C functions that have a prototype with named
3027      arguments
3028      (see changes to CKIT)
3029    
3030    ----------------------------------------------------------------------
3031    Name: Allen Leung
3032    Date: 2001/10/27 20:34:00 EDT
3033    Tag: leunga-20011027-x86-fast-fp-call
3034    Description:
3035    
3036       Fixed the bug described in blume-20010920-slowfp.
3037    
3038       The fix involves
3039          1. generating FCOPYs in FSTP in ia32-svid
3040          2. marking a CALL with the appropriate annotation
3041    
3042    ----------------------------------------------------------------------
3043    Name: Matthias Blume
3044    Date: 2001/10/16 11:32:00 EDT
3045    Tag: blume-20011016-netbsd
3046    Description:
3047    
3048    Underscore patch from Chris Richards (fixing problem with compiling
3049    runtime system under recent NetBSD).
3050    
3051    ----------------------------------------------------------------------
3052    Name: Allen Leung
3053    Date: 2001/10/12 17:18:32 EDT 2001
3054    Tag: leung-20011012-x86-printflowgraph
3055    Description:
3056    
3057    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
3058    
3059    ----------------------------------------------------------------------
3060    Name: Lal George
3061    Date: 2001/10/11 23:51:34 EDT
3062    Tag: george-20011011-too-many-instrs
3063    Description:
3064    
3065    The representation of a program point never expected to see more
3066    than 65536 instructions in a basic block!
3067    
3068    ----------------------------------------------------------------------
3069    Name: Lal George
3070    Date: 2001/10/09 09:41:37 EDT
3071    Tag: george-20011008-mlrisc-labels
3072    Description:
3073    
3074    Changed the machine description files to support printing of
3075    local and global labels in assembly code, based on host assembler
3076    conventions.
3077    
3078    ----------------------------------------------------------------------
3079    Name: Matthias Blume
3080    Date: 2001/09/25 15:25:00 EDT
3081    Tag: blume-20010925-exninfo
3082    Description:
3083    
3084    I provided a non-hook implementation of exnName (at the toplevel) and
3085    made the "dummy" implementation of exnMessage (at the toplevel) more
3086    useful: if nothing gets "hooked in", then at least you are going to
3087    see the exception name and a message indicating why you don't see more.
3088    
3089    [For the time being, programs that need exnMessage and want to use
3090    ml-build should either use General.exnMessage (strongly recommended) or
3091    refer to structure General at some other point so that CM sees a
3092    static dependency.]
3093    
3094    [Similar remarks go for "print" and "use":  If you want to use their
3095    functionality in stand-alone programs generated by ml-build, then use
3096    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
3097    
3098    ----------------------------------------------------------------------
3099    Name: Matthias Blume
3100    Date: 2001/09/20 17:28:00 EDT
3101    Tag: blume-20010920-slowfp
3102    Description:
3103    
3104    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
3105    
3106    ----------------------------------------------------------------------
3107    Name: Matthias Blume
3108    Date: 2001/09/20 17:20:00 EDT
3109    Tag: blume-20010920-canonicalpaths
3110    Description:
3111    
3112    0. Updated the BOOT file (something that I forgot to do earlier).
3113    
3114    1. Small internal change to CM so that it avoids "/../" in filenames
3115       as much as possible (but only where it is safe).
3116    
3117    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
3118       that arguments that contain delimiters are passed through correctly.
3119       This change also means that all "special" arguments of the form
3120       @SMLxxx... must come first.
3121    
3122    3. Changed install script to put relative anchor names for tool commands
3123       into pathconfig.
3124    
3125    ----------------------------------------------------------------------
3126    Name: Matthias Blume
3127    Date: 2001/09/18 15:35:00 EDT
3128    Tag: blume-20010918-readme11036
3129    Description:
3130    
3131    Added README files.
3132    
3133    ----------------------------------------------------------------------
3134    Name: Matthias Blume
3135    Date: 2001/09/18 11:45:00 EDT
3136    Tag: Release_110_36 (retag)
3137    Description:
3138    
3139    Fixed mistake in config/preloads. Retagged as 110.36.
3140    
3141    ----------------------------------------------------------------------
3142    Name: Matthias Blume
3143    Date: 2001/09/18 09:40:00 EDT
3144    Tag: Release_110_36_orig (tag changed)
3145    Description:
3146    
3147    New version (110.36).  New bootfiles.
3148    
3149    ----------------------------------------------------------------------
3150    Name: Matthias Blume
3151    Date: 2001/09/14 16:15:00 EDT
3152    Tag: blume-20010914-x86fastfp
3153    Description:
3154    
3155    John committed some changes that Allen made, in particular a (hopefully)
3156    correctly working version of the x86-fp module.
3157    
3158    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
3159    flag to "true".  Everything seems to compile to a fixpoint ok, and
3160    "mandelbrot" speeds up by about 15%.
3161    
3162    ----------------------------------------------------------------------
3163    Name: Matthias Blume
3164    Date: 2001/09/13 11:20:00 EDT
3165    Tag: blume-20010913-minimal
3166    Description:
3167    
3168    1. Stefan Monnier's patch to fix a miscompilation problem that
3169       was brought to light by John Reppy's work on Moby.
3170    
3171    2. Implemented a minimal "structure Compiler" that contains just
3172       "version" and "architecture".  The minimal version will be
3173       available when the full version is not.  This is for backward-
3174       compatibility with code that wants to test Compiler.version.
3175    
3176    ----------------------------------------------------------------------
3177    Name: Matthias Blume
3178    Date: 2001/08/28 14:03:00 EDT
3179    Tag: blume-20010828-ml-lex
3180    Description:
3181    
3182    Fix for bug 1581, received from Neophytos Michael.
3183    
3184    ----------------------------------------------------------------------
3185    Name: Matthias Blume
3186    Date: 2001/08/27 11:20:00 EDT
3187    Tag: blume-20010827-readme11035
3188    Description:
3189    
3190    Fleshed out the README file for 110.35.
3191    
3192    ----------------------------------------------------------------------
3193    Name: Matthias Blume
3194    Date: 2001/08/24 17:10:00 EDT
3195    Tag: Release_110_35
3196    Description:
3197    
3198    New version number (110.35).  New bootfiles.
3199    
3200    ----------------------------------------------------------------------
3201    Name: Lal George
3202    Date: 2001/08/24 13:47:18 EDT 2001
3203    Tag: george-20010824-MLRISC-graphs
3204    Description:
3205    
3206     removed clusters from MLRISC completely and replaced with graphs.
3207    
3208    ----------------------------------------------------------------------
3209    Name: Matthias Blume
3210    Date: 2001/08/23 17:50:00 EDT
3211    Tag: blume-20010823-toplevel
3212    Description:
3213    
3214    - some reorganization of the code that implements various kinds of
3215      environments in the compiler (static, dynamic, symbolic, combined)
3216    - re-implemented the EnvRef module so that evalStream works properly
3217      (if the stream contains references to "use", "CM.make", etc.)
3218    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
3219    
3220    ----------------------------------------------------------------------
3221    Name: Matthias Blume
3222    Date: 2001/08/20 15:50 EDT
3223    Tag: blume20010820-slipup
3224    Description:
3225    
3226    I forgot to commit a few files.  Here they are...
3227    
3228    ----------------------------------------------------------------------
3229    Name: Matthias Blume
3230    Date: 2001/08/20 15:35:00 EDT
3231    Tag: blume-20010820-debugprof
3232    Description:
3233    
3234    !!!! NEW BOOTFILES !!!!
3235    
3236    This is another round of reorganizing the compiler sources.  This
3237    time the main goal was to factor out all the "instrumentation"
3238    passes (for profiling and backtracing) into their own library.
3239    The difficulty was to do it in such a way that it does not depend
3240    on elaborate.cm but only on elabdata.cm.
3241    
3242    Therefore there have been further changes to both elaborate.cm and
3243    elabdata.cm -- more "generic" things have been moved from the former
3244    to the latter.  As a result, I was forced to split the assignment
3245    of numbers indicating "primtyc"s into two portions: SML-generic and
3246    SML/NJ-specific.  Since it would have been awkward to maintain,
3247    I bit the bullet and actually _changed_ the mapping between these
3248    numbers and primtycs.  The bottom line of this is that you need
3249    a new set of bin- and bootfiles.
3250    
3251    I have built new bootfiles for all architectures, so doing a fresh
3252    checkout and config/install.sh should be all you need.
3253    
3254    The newly created library's name is
3255    
3256        $smlnj/viscomp/debugprof.cm
3257    
3258    and its sources live under
3259    
3260        src/compiler/DebugProf
3261    
3262    ----------------------------------------------------------------------
3263    Name: Matthias Blume
3264    Date: 2001/08/15 17:15:00 EDT
3265    Tag: blume-20010815-compreorg
3266    Description:
3267    
3268    This is a first cut at reorganizing the CM libraries that make up the
3269    core of the compiler.  The idea is to separate out pieces that could
3270    be used independently by tools, e.g., the parser, the typechecker, etc.
3271    
3272    The current status is a step in this direction, but it is not quite
3273    satisfactory yet.  Expect more changes in the future.
3274    
3275    Here is the current (new) organization...
3276    
3277        What used to be $smlnj/viscomp/core.cm is now divided into
3278        six CM libraries:
3279    
3280             $smlnj/viscomp/basics.cm
3281                           /parser.cm
3282                           /elabdata.cm
3283                           /elaborate.cm
3284                           /execute.cm
3285                           /core.cm
3286    
3287        The CM files for these libraries live under src/system/smlnj/viscomp.
3288        All these libraries are proxy libraries that contain precisely
3289        one CM library component.  Here are the locations of the components
3290        (all within the src/compiler tree):
3291    
3292             Basics/basics.cm
3293             Parse/parser.cm
3294             ElabData/elabdata.cm
3295             Elaborator/elaborate.cm
3296             Execution/execute.cm
3297             core.cm
3298    
3299         [This organization is the same that has been used already
3300         for a while for the architecture-specific parts of the visible
3301         compiler and for the old version of core.cm.]
3302    
3303         As you will notice, many source files have been moved from their
3304         respective original locations to a new home in one of the above
3305         subtrees.
3306    
3307         The division of labor between the new libraries is the following:
3308    
3309             basics.cm:
3310                - Simple, basic definitions that pertain to many (or all) of
3311                  the other libraries.
3312             parser.cm:
3313                - The SML parser, producing output of type Ast.dec.
3314                - The type family for Ast is also defined and exported here.
3315             elabdata.cm:
3316                - The datatypes that describe input and output of the elaborator.
3317                  This includes types, absyn, and static environments.
3318             elaborator.cm:
3319                - The SML/NJ type checker and elaborator.
3320                  This maps an Ast.dec (with a given static environment) to
3321                  an Absyn.dec (with a new static environment).
3322                - This libraries implements certain modules that used to be
3323                  structures as functors (to remove dependencies on FLINT).
3324             execute.cm:
3325                - Everything having to do with executing binary code objects.
3326                - Dynamic environments.
3327             core.cm:
3328                - SML/NJ-specific instantiations of the elaborator and MLRISC.
3329                - Top-level modules.
3330                - FLINT (this should eventually become its own library)
3331    
3332    Notes:
3333    
3334    I am not 100% happy with the way I separated the elaborator (and its
3335    data structures) from FLINT.  Two instances of the same problem:
3336    
3337        1. Data structures contain certain fields that carry FLINT-specific
3338           information.  I hacked around this using exn and the property list
3339           module from smlnj-lib.  But the fact that there are middle-end
3340           specific fields around at all is a bit annoying.
3341    
3342        2. The elaborator calculates certain FLINT-related information.  I tried
3343           to make this as abstract as I could using functorization, but, again,
3344           the fact that the elaborator has to perform calculations on behalf
3345           of the middle-end at all is not nice.
3346    
3347        3. Having to used exn and property lists is unfortunate because it
3348           weakens type checking.  The other alternative (parameterizing
3349           nearly *everything*) is not appealing, though.
3350    
3351    I removed the "rebinding =" warning hack because due to the new organization
3352    it was awkward to maintain it.  As a result, the compiler now issues some of
3353    these warnings when compiling init.cmi during bootstrap compilation. On
3354    the plus side, you also get a warning when you do, for example:
3355       val op = = Int32.+
3356    which was not the case up to now.
3357    
3358    I placed "assign" and "deref" into the _Core structure so that the
3359    code that deals with the "lazy" keyword can find them there.  This
3360    removes the need for having access to the primitive environment
3361    during elaboration.
3362    
3363    ----------------------------------------------------------------------
3364    Name: Matthias Blume
3365    Date: 2001/08/13
3366    Tag: blume-20010813-closures
3367    Description:
3368    
3369    This fix was sent to us by Zhong Shao.  It is supposed to improve the
3370    performance of certain loops by avoiding needless closure allocation.
3371    
3372    ----------------------------------------------------------------------
3373    Name: Lal George
3374    Date: 2001/07/31 10:03:23 EDT 2001
3375    Tag: george-20010731-x86-fmalloc
3376    Description: Fixed bug in x86 calls
3377    
3378        There was a bug where call instructions would mysteriously
3379        vanish. The call instruction had to be one that returned
3380        a floating point value.
3381    
3382    ----------------------------------------------------------------------
3383    Name: Lal George
3384    Date: 2001/07/19 16:36:29 EDT 2001
3385    Tag: george-20010719-simple-cells
3386    Description:
3387    
3388    I have dramatically simplified the interface for CELLS in MLRISC.
3389    
3390    In summary, the cells interface is broken up into three parts:
3391    
3392      1. CellsBasis : CELLS_BASIS
3393    
3394            CellsBasis is a top level structure and common for all
3395            architectures.  it contains the definitions of basic datatypes
3396            and utility  functions over these types.
3397    
3398      2. functor Cells() : CELLS
3399    
3400            Cells generates an interface for CELLS that incorporates the
3401            specific resources on the target architecture, such as the
3402            presence of special register classes, their number and size,
3403            and various useful substructures.
3404    
3405      3. <ARCH>CELLS
3406    
3407            e.g. SparcCells: SPARCCELLS
3408    
3409            <ARCH>CELLS usually contains additional bindings for special
3410            registers  on the architecture, such as:
3411    
3412                    val r0 : cell           (* register zero *)
3413                    val y : cell            (* Y register *)
3414                    val psr : cell          (* processor status register *)
3415                    ...
3416    
3417            The structure returned by applying the Cells functor is opened
3418            in this interface.
3419    
3420    The main implication of all this is that the datatypes for cells is
3421    split between CellsBasis and CELLS -- a fairly simple change for user
3422    code.
3423    
3424    In the old scheme the CELLS interface had a definitional binding of
3425    the form:
3426    
3427            signature CELLS = sig
3428    
3429               structure CellsBasis = CellsBasis
3430    
3431               ...
3432    
3433            end
3434    
3435    With all the sharing constraints that goes on in MLRISC, this old
3436    design  quickly leads to errors such as:
3437    
3438            "structure definition spec inside of sharing ... "
3439    
3440    
3441    and appears to require an unacceptable amount of sharing and where
3442    constraint hackery.
3443    
3444    I think this error message (the interaction of definitional specs and
3445    sharing) requires more explanation on our web page.
3446    
3447    ----------------------------------------------------------------------
3448    Name: Matthias Blume
3449    Date: 2001/07/19 15:00:00 EDT
3450    Tag: blume-20010719-libreorg
3451    Description:
3452    
3453    This update puts together a fairly extensive but straightforward change
3454    to the way the libraries that implement the interactive system are
3455    organized:
3456    
3457       The biggest change is the elimination of structure Compiler.  As a
3458       replacement for this structure, there is now a CM library
3459       (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
3460       that exports all the substructures of the original structure Compiler
3461       directly.  So instead of saying Compiler.Foo.bar one now simply
3462       says Foo.bar.  (The CM libraries actually export a collection of
3463       structures that is richer than the collection of substructures of
3464       structure Compiler.)
3465    
3466       To make the transition smooth, there is a separate library called
3467       $smlnj/compiler/compiler.cm which puts together and exports the
3468       original structure Compiler (or at least something very close to it).
3469    
3470       There are five members of the original structure Compiler
3471       that are not exported directly but which instead became members
3472       of a new structure Backend (described by signature BACKEND).  These are:
3473       structure Profile (: PROFILE), structure Compile (: COMPILE), structure
3474       Interact (: INTERACT), structure Machine (: MACHINE), and val
3475       architecture (: string).
3476    
3477       Structure Compiler.Version has become structure CompilerVersion.
3478    
3479       Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
3480       by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
3481       or x86, respectively.
3482       Each of these exports the same frontend structures that
3483       $smlnj/compiler.cm exports.  But they do not have a structure Backend
3484       and instead export some structure <Arch>Backend where <Arch> is Alpha32,
3485       Hppa, PPC, Sparc, or X86, respectively.
3486    
3487       Library $smlnj/compiler/all.cm exports the union of the exports of
3488       $smlnj/compiler/<arch>.cm
3489    
3490       There are no structures <Arch>Compiler anymore, use
3491       $smlnj/compiler/<arch>.cm instead.
3492    
3493       Library host-compiler-0.cm is gone.  Instead, the internal library
3494       that instantiates CM is now called cm0.cm.  Selection of the host
3495       compiler (backend) is no longer done here but. (Responsibility for it
3496       now lies with $smlnj/compiler/current.cm.  This seems to be more
3497       logical.)
3498    
3499       Many individual files have been moved or renamed.  Some files have
3500       been split into multiple files, and some "dead" files have been deleted.
3501    
3502    Aside from these changes to library organization, there are also changes
3503    to the way the code itself is organized:
3504    
3505       Structure Binfile has been re-implemented in such a way that it no
3506       longer needs any knowledge of the compiler.  It exclusively deals
3507       with the details of binfile layout.  It no longer invokes the
3508       compiler (for the purpose of creating new prospective binfile
3509       content), and it no longer has any knowledge of how to interpret
3510       pickles.
3511    
3512       Structure Compile (: COMPILE) has been stripped down to the bare
3513       essentials of compilation.  It no longer deals with linking/execution.
3514       The interface has been cleaned up considerably.
3515    
3516       Utility routines for dealing with linking and execution have been
3517       moved into their own substructures.
3518    
3519       (The ultimate goal of these changes is to provide a light-weight
3520       binfile loader/linker (at least for, e.g., stable libraries) that
3521       does not require CM or the compiler to be present.)
3522    
3523    CM documentation has been updated to reflect the changes to library
3524    organization.
3525    
3526    ----------------------------------------------------------------------
3527    Name: Matthias Blume
3528    Date: 2001/07/10 17:30:00 EDT
3529    Tag: Release_110_34
3530    Description:
3531    
3532    Minor tweak to 110.34 (re-tagged):
3533    
3534      - README.html file added to CVS repository
3535      - runtime compiles properly under FreeBSD 3.X and 4.X
3536    
3537    ----------------------------------------------------------------------
3538    Name: Matthias Blume
3539    Date: 2001/07/10 17:30:00 EDT
3540    Tag: Release_110_34
3541    Description:
3542    
3543    New version number (110.34). New bootfiles.
3544    
3545    ----------------------------------------------------------------------
3546    Name: Matthias Blume
3547    Date: 2001/07/09 16:00:00 EDT
3548    Tag: blume-20010709-more-varargs
3549    Description:
3550    
3551    I changed the handling of varargs in ml-nlffigen again:
3552    The ellipsis ... will now simply be ignored (with an accompanying warning).
3553    
3554    The immediate effect is that you can actually call a varargs function
3555    from ML -- but you can't actually supply any arguments beyond the ones
3556    specified explicitly.  (For example, you can call printf with its format
3557    string, but you cannot pass additional arguments.)
3558    
3559    This behavior is only marginally more useful than the one before, but
3560    it has the advantage that a function or, more importantly, a function
3561    type never gets dropped on the floor, thus avoiding follow-up problems with
3562    other types that refer to the offending one.
3563    
3564    ----------------------------------------------------------------------
3565    Name: Matthias Blume
3566    Date: 2001/07/09 11:25:00 EDT
3567    Tag: blume-20010709-varargs
3568    Description:
3569    
3570    1. ckit-lib.cm now exports structure Error
3571    2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
3572       with a warning accompanied by a source location.  Moreover, it
3573       merely skips the offending function or type and proceeds with the
3574       rest of its work.u  As a result, one can safely feed C code containing
3575       "..." to ml-nlffigen.
3576    3. There are some internal improvements to CM, providing slightly
3577       more general string substitutions in the tools subsystem.
3578    
3579    ----------------------------------------------------------------------
3580    Name: Matthias Blume
3581    Date: 2001/06/27 15:10:00 EDT
3582    Tag: blume-20010627-concur
3583    Description:
3584    
3585    Fixed a small bug in CM's handling of parallel compilation.
3586    (You could observe the bug by Control-C-interrupting an ordinary
3587    CMB.make or CM.stabilize and then attaching some compile servers.
3588    The result was that all of a sudden the previously interrupted
3589    compilation would continue on its own.  This was because of
3590    an over-optimization: CM did not bother to clean out certain queues
3591    when no servers were attached "anyway", resulting in the contents
3592    of these queues to grab control when new servers did get attached.)
3593    
3594    There is also another minor update to the CM manual.
3595    
3596    ----------------------------------------------------------------------
3597    Name: Matthias Blume
3598    Date: 2001/06/26 16:15:00 EDT
3599    Tag: blume-20010626-cmdoc
3600    Description:
3601    
3602    Minor typo fixed in CM manual (syntax diagram for libraries).
3603    
3604    ----------------------------------------------------------------------
3605    Name: Matthias Blume
3606    Date: 2001/06/25 22:55:00 EDT
3607    Tag: blume-20010625-x86pc
3608    Description:
3609    
3610    Fixed a nasty bug in the X86 assembly code that caused signal
3611    handlers to fail (crash) randomly.
3612    
3613    ----------------------------------------------------------------------