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 595, Wed Apr 5 03:30:17 2000 UTC revision 1500, Fri May 28 21:44:42 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: 2004/05/28 16:45:00 CDT
18    Tag: blume-20040528-basis
19    Description:
20    
21    Added signature PACK_REAL and exported functor PrimIO.
22    
23    ----------------------------------------------------------------------
24    Name: Matthias Blume (blume (at) tti - c (dot) org)
25    Date: 2004/05/25 16:00:00 CDT
26    Tag: blume-20040525-group-owner
27    Description:
28    
29    CM now ignores (but still accepts) the "owner" information in group
30    descriptions.  The owner of a group is its next enclosing
31    library. Each group must have a unique owner.  (There is a virtual
32    "toplevel" library that own groups which are not nested within a real
33    library.)  Previously, each group had to explicitly declare its owner,
34    and CM would check that such a declaration is correct.  The new scheme
35    is to have CM check that for each group there is precisely one owning
36    library.
37    
38    The advantage of the new scheme is that the programmer no longer needs
39    to maintain the somewhat annoying owner information.  The downside is
40    that CM cannot enforce the ownership rule across multiple runs of
41    CM.make.  Fortunately, enclosing the same group in two different
42    libraries A and B which are not part of the same program does not
43    cause real problems.
44    
45    ----------------------------------------------------------------------
46    Name: Matthias Blume (blume (at) tti - c (dot) org)
47    Date: 2004/05/20 16:00:00 CDT
48    Tag: blume-20040520-win32
49    Description:
50    
51    Made the win32 version work again.  (Strangely, a misplaced comma had
52    slipped into win32-process.c which prevented the runtime from being
53    compiled correctly.)
54    
55    Also, included a minor addition to ml-build.bat analogous to what was
56    done in blume-20040519-ml-build.
57    
58    ----------------------------------------------------------------------
59    Name: Matthias Blume (blume (at) tti - c (dot) org)
60    Date: 2004/05/19 22:10:00 CDT
61    Tag: blume-20040519-ml-build
62    Description:
63    
64    Arranged for ml-build to clean up after itself a little bit better.
65    The script generates a temporary SML source file and compiles it using
66    CM, so CM generates metadata (GUID, SKEL, objectfile) for it.  It now
67    gets rid of those at the end, so they don't accumulate under .cm.
68    
69    This required a minor change to install.sh because the name of the
70    metadata directory (default: .cm) is actually configurable at
71    installation time.
72    
73    ----------------------------------------------------------------------
74    Name: Matthias Blume (blume (at) tti - c (dot) org)
75    Date: 2004/05/18 15:50:00 CDT
76    Tag: blume-20040518-mkreader
77    Description:
78    
79    Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
80    implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
81    
82    ----------------------------------------------------------------------
83    Name: Matthias Blume (blume (at) tti - c (dot) org)
84    Date: 2004/05/11 14:35:00 CDT
85    Tag: blume-20040511-win32sock
86    Description:
87    
88    Added previously missing support for many socket-related functions
89    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
90    for the voluminous patch!
91    
92    (I have not tested this patch under win32 yet.)
93    
94    Here is David's e-mail:
95    
96    Hi,
97    
98    Attached to this email you find a diff against sml/nj 110.45
99    that will enable socket support under Windows.
100    
101    To apply the patch (using unix or cygwin)
102    1) gunzip runtime.diff.gz
103    2) "cd" into "src/runtime" in the source tree of a fresh
104       110.45 installation.
105    3) patch -p 1 < [your/path/to]runtime.diff
106    
107    The code compiles fine but has NOT yet been extensively tested.
108    I only ran a few tests for basic socket client functionality
109    (which worked fine).  Especially the functions that use ioctl
110    are not tested at all and might not work (see below).
111    
112    I implemented this since we want to move to a newer version of sml/nj
113    but need socket support in order to use it.  This is the first time I
114    even had a look at the sml/nj source,  so please review my changes
115    before making this part of the distribution!  Here are a few issues
116    that I think might be better for someone to solve who is more
117    familiar with the sml/nj source (and socket programming):
118    
119    - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
120      exception since I could not figure out what the windows equivalent
121      of these functions is
122    
123    - In sockets-osdep.h there are a some #include statements that are
124      only used in a few files that include sockets-osdep.h
125    
126    - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
127      does not process its return value since I don't know how to
128      report an error upwards.
129    
130    - It would probably be good to have a call to WSACleanup() when
131      the library is unloaded (if there is such a possibility).
132      Otherwise I think Windows will take care of this automatically
133      when the process finishes.
134    
135    - I used ioctlsocket() as a replacement for ioctl() but I have
136      no idea if that is actually the proper replacement on Windows.
137    
138    - All these issues are marked in the code by "FIXME" comments.
139    
140    We use sml/nj extensively in our products and are quite happy
141    with it.  I hope this contribution will help you.
142    
143    Keep up the good work!
144    
145    David
146    
147    ----------------------------------------------------------------------
148    Name: Matthias Blume (blume (at) tti - c (dot) org)
149    Date: 2004/05/11 14:20:00 CDT
150    Tag: blume-20040511-installml
151    Description:
152    
153    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
154    the patch.)
155    
156    ----------------------------------------------------------------------
157    Name: Matthias Blume (blume (at) tti - c (dot) org)
158    Date: 2004/05/11 14:05:00 CDT
159    Tag: blume-20040511-nlffi-netbsd
160    Description:
161    
162    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
163    the patch.)
164    
165    ----------------------------------------------------------------------
166    Name: Matthias Blume (blume (at) tti - c (dot) org)
167    Date: 2004/05/11 12:05:00 CDT
168    Tag: blume-20040511-exports
169    Description:
170    
171    As per request by Adam Chlipala <adam@hcoop.net>, extended various
172    export lists in compiler-related .cm-files.
173    
174    ----------------------------------------------------------------------
175    Name: Matthias Blume (blume (at) tti - c (dot) org)
176    Date: 2004/05/11 11:35:00 CDT
177    Tag: blume-20040511-allsource
178    Description:
179    
180    The installer now honors the "src-smlnj" target again, although its meaning
181    has changed from "all sources required for the compiler" to "all sources
182    the installer knows about".  In other words, if you enable "src-smlnj"
183    in the "targets" file, then the installer will pull in sources for
184    everything.  (Notice that this refers to source code only.  Compiled
185    code is still only installed for modules that were requested explicitly
186    or which are required for other modules that were requested explicitly.)
187    
188    ----------------------------------------------------------------------
189    Name: Matthias Blume (blume (at) tti - c (dot) org)
190    Date: 2004/04/23 17:40:00 CDT
191    Tag: blume-20040423-ieee-scan
192    Description:
193    
194    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
195    in the exponent calculation we get INF or ZERO (depending on the mantissa
196    and the sign of the exponent).
197    
198    ----------------------------------------------------------------------
199    Name: Matthias Blume (blume (at) tti - c (dot) org)
200    Date: 2004/04/23 10:40:00 CDT
201    Tag: blume-20040423-ml-build
202    Description:
203    
204    The ml-build script now terminates with a non-0 status when something
205    goes wrong.
206    
207    ----------------------------------------------------------------------
208    Name: Matthias Blume (blume (at) tti - c (dot) org)
209    Date: 2004/04/22 16:35:00 CDT
210    Tag: blume-20040422-Option
211    Description:
212    
213    Made exception Option to be the same as exception Option.Option
214    (as it should be).
215    
216    ----------------------------------------------------------------------
217    Name: Allen Leung (leunga (at) reservoir (dot) com)
218    Date: 2004/03/19 14:40:00 EST
219    Tag: leunga-20040319-cygwin-nlffi
220    Description:
221    
222    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
223    of SML/NJ.  The problem is that
224    
225       lib = dlopen(NULL, ...)
226       f   = dlsym(lib, "malloc");
227    
228    does not work on Windows unless we explicitly export symbols
229    such as 'malloc' during linking.  We fixed this by explicitly
230    exporting the required symbols with the magic gcc incantation:
231    
232        -Wl,--export-all cygwin.def
233    
234    where cygwin.def is a file containing all the symbols that
235    we wish to export.
236    
237    I suspect this is a Windows problem and we'll have to
238    do the same (somehow with windows compilers) when
239    we build the native win32 version with the system
240    calls LoadLibrary/GetProcAddress.
241    
242    ----------------------------------------------------------------------
243    Name: Matthias Blume (blume (at) tti - c (dot) org)
244    Date: 2004/03/04 16:35:00 CST
245    Tag: blume-20040304-intinf-fmt
246    Description:
247    
248    Fixed problem with IntInf.fmt (sign would show up on the right instead
249    of on the left for BIN, OCT, and HEX).
250    
251    ----------------------------------------------------------------------
252    Name: Matthias Blume (blume (at) tti - c (dot) org)
253    Date: 2004/03/04 11:25:00 CST
254    Tag: blume-20040304-symlinks
255    Description:
256    
257    Fixed problem with installer script (unix only) where bin/ml-yacc and
258    friends pointed (via symlinks) to absolute locations instead of just
259    .run-sml.  This was reported by Vesa A Norrman.
260    
261    ----------------------------------------------------------------------
262    Name: Matthias Blume (blume (at) tti - c (dot) org)
263    Date: 2004/02/13 14:50:00 CST
264    Tag: Release_110_45
265    Description:
266    
267    New working version (110.45). New bootfiles.
268    
269    ----------------------------------------------------------------------
270    Name: Matthias Blume (blume (at) tti - c (dot) org)
271    Date: 2004/01/26 15:15:15 CST
272    Tag: blume-20040126-toplevel
273    Description:
274    
275    Improved handling of exceptions at the interactive toplevel.
276    
277    ----------------------------------------------------------------------
278    Name: Matthias Blume (blume (at) tti - c (dot) org)
279    Date: 2004/01/26 11:25:00
280    Tag: blume-20040126-app
281    Description:
282    
283    Type of top-level "app" corrected.
284    Added code for setting vp_limitPtrMask to Win32-specific runtime.
285    
286    ----------------------------------------------------------------------
287    Name: Matthias Blume (blume (at) tti - c (dot) org)
288    Date: 2003/11/18 17:10 CST
289    Tag: blume-20031118-basis-fiddle
290    Description:
291    
292    - changed Timer interface to what might become the spec
293    - POSIX_FLAGS -> BIT_FLAGS according to spec
294    - some other minor discrepancies wrt. spec eliminated
295    
296    ----------------------------------------------------------------------
297    Name: Matthias Blume (blume (at) tti - c (dot) org)
298    Date: 2003/11/06 12:00:00 CST
299    Tag: Release_110_44
300    Description:
301    
302    New working version (110.44). New bootfiles.
303    
304    ----------------------------------------------------------------------
305    Name: Matthias Blume (blume (at) tti - c (dot) org)
306    Date: 2003/11/04 11:50:00 CST
307    Tag: blume-20031104-move-libraries
308    Description:
309    
310    Eliminated the "dont_move_libraries" directive in config/targets.
311    (The mechanism was broken and could not be fixed easily.  Moreover,
312    there does not seem to be any reason not to move all libraries into
313    lib during installation.  I originally implemented this directive as a
314    backward-compatibility feature when I first introduced the new CM.
315    Now that things have been stable for a long time and going back to the
316    old CM is not an option, there is no reason to keep it around.)
317    
318    ----------------------------------------------------------------------
319    Name: Matthias Blume (blume (at) tti - c (dot) org)
320    Date: 2003/11/03 16:00:00 CST
321    Tag: blume-20031103-installdir
322    Description:
323    
324    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
325    Richards for pointing out the problem and providing the solution.)
326    
327    ----------------------------------------------------------------------
328    Name: Matthias Blume (blume (at) tti - c (dot) org)
329    Date: 2003/10/01 17:05:00 CDT
330    Tag: blume-20031001-lal-mlrisc
331    Description:
332    
333    MLRISC bug fix from Lal.
334    
335    ----------------------------------------------------------------------
336    Name: Matthias Blume (blume (at) tti - c (dot) org)
337    Date: 2003/09/30 16:10:00 CDT
338    Tag: blume-20030930-primio-bat
339    Description:
340    
341    1. Added openVector, nullRd, and nullWr to PRIM_IO.
342    2. Improved .bat files (for Win32 port) to make things work under Win95.
343       (thanks to Aaron S. Hawley for this one)
344    
345    ----------------------------------------------------------------------
346    Name: Matthias Blume (blume (at) tti - c (dot) org)
347    Date: 2003/09/26 16:05:00 CDT
348    Tag: blume-20030926-wrappriv
349    Description:
350    
351    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
352    
353    ----------------------------------------------------------------------
354    Name: Matthias Blume (blume (at) tti - c (dot) org)
355    Date: 2003/09/26 15:00:00 CDT
356    Tag: blume-20030926-110_43_3
357    Description:
358    
359    - additional cleanup
360    - version number bump, NEW BOOTFILES
361    
362    ----------------------------------------------------------------------
363    Name: Matthias Blume (blume (at) tti - c (dot) org)
364    Date: 2003/09/26 12:00:00 CDT
365    Tag: blume-20030926-ppautoload
366    Description:
367    
368    I modified the read-eval-print loop so that the autoloader gets
369    invoked whenever the prettyprinter tries to look up a symbol that
370    is not currently defined in the toplevel environment but which
371    appears in CM's autoload registry.  As a result, we see far fewer of
372    those ?.Foo.Bar.xxx names in the prettyprinter's output.
373    
374    In addition to this I tried to clean up some pieces of the Basis
375    implementation (e.g., Socket, Word8Array) in order to prevent other
376    instances of these ?.Foo.Bar.xxx names from being printed.
377    
378    The mechanism that picks names for types still needs some work, though.
379    (Right now it seems that if there is a type A.t which is defined to
380    be B.u, but B is unavailable at toplevel, then A.t gets printed as
381    "?.B.u" although the perhaps more sensible solution would be to use
382    "A.t" in this case.  In other words, the prettyprinter should follow
383    a chain of DEFtycs not farther than there are corresponding toplevel
384    names in the current environment.)
385    
386    ----------------------------------------------------------------------
387    Name: Matthias Blume (blume (at) tti - c (dot) org)
388    Date: 2003/09/24 16:31:00 CDT
389    Tag: blume-20030924-installer
390    Description:
391    
392    Another installer tweak:  All the ML code for the installer is now
393    compiled during CMB.make and put into a little library called
394    $smlnj/installer.cm.  The installation then simply invokes
395    
396       sml -m $smlnj/installer.cm
397    
398    and everything happens automagically.
399    
400    Win32:  ML code senses value of environment variable SMLNJ_HOME.
401    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
402            and BINDIR.
403    
404    The new scheme guarantees that the ML code responsible for the installation
405    is in sync with the APIs of the main system.  Also, the installer is
406    somewhat faster because the installer script is precompiled.
407    
408    ----------------------------------------------------------------------
409    Name: Matthias Blume (blume (at) tti - c (dot) org)
410    Date: 2003/09/24 15:35:00 CDT
411    Tag: blume-20030924-synsock
412    Description:
413    
414    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
415    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
416    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
417    implementations of non-blocking operations from CML's Socket
418    structure.
419    
420    ----------------------------------------------------------------------
421    Name: Matthias Blume (blume (at) tti - c (dot) org)
422    Date: 2003/09/24 15:10:05 CDT
423    Tag: blume-20030924-sockets
424    Description:
425    
426    1. Fixed SOCKET API and implementation to match Basis spec.
427       This required changing the internal representation of sockets to one
428       that remembers (for each socket file descriptor) whether it is currently
429       blocking or non-blocking.  This state is maintained lazily (i.e., a system
430       call is made only if the state actually needs to change).
431    
432    2. OS-specific details of sockets were moved into separate files, thus
433       making it possible to unify the bulk of the socket implementations
434       between Unix and Win32.
435    
436    3. CML's socket API changed accordingly.
437       (Note that we need to remove non-blocking functions from this API
438       since they are redundant in the case of CML!)
439    
440    4. CML's socket implementation now makes use of non-blocking functions
441       provided by Basis, thus removing all OS-dependent code from this part
442       of CML.
443    
444    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
445    
446    ----------------------------------------------------------------------
447    Name: Matthias Blume (blume (at) tti - c (dot) org)
448    Date: 2003/09/22 12:10:00 CDT
449    Tag: blume-20030922-110_43_2
450    Description:
451    
452    Made a new interim version and bootfiles for developer's bootstrapping
453    convenience.
454    
455       110.43.2 -- NEW BOOTFILES
456    
457    ----------------------------------------------------------------------
458    Name: Matthias Blume (blume (at) tti - c (dot) org)
459    Date: 2003/09/19 15:55:00 CDT
460    Tag: blume-20030919-cmdir
461    Description:
462    
463    1. new-install.sh -> install.sh
464    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
465    3. tweaked installer so that another name instead of .cm can be chosen
466       at install time (by setting the CM_DIR_ARC environment variable
467       during installation); once installation is complete, the name is
468       fixed
469    
470    ----------------------------------------------------------------------
471    Name: Matthias Blume (blume (at) tti - c (dot) org)
472    Date: 2003/09/18 16:00:00 CDT
473    Tag: blume-20030918-110_43_1
474    Description:
475    
476    Made a new interim version and bootfiles for developer's bootstrapping
477    convenience.
478    
479       110.43.1 -- NEW BOOTFILES
480    
481    ----------------------------------------------------------------------
482    Name: Matthias Blume (blume (at) tti - c (dot) org)
483    Date: 2003/09/18 15:20:00 CDT
484    Tag: blume-20030918-misc
485    Description:
486    
487    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
488       this seems to be controversial at the moment) and used those in
489       Posix.ProcEnv.times.
490    
491    2. Added Time.{from,to}Nanoseconds to Time.
492    
493    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
494       For example, fromLargeInt never needs to look at more than 3 "big
495       digits" to get its 53 bits of precision.
496    
497    ----------------------------------------------------------------------
498    Name: Matthias Blume (blume (at) tti - c (dot) org)
499    Date: 2003/09/17 16:30:00 CDT
500    Tag: blume-20030917-real32-slices
501    Description:
502    
503    Added an entry to the primitive environment
504    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
505    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
506    
507    Removed some of the "magic" constants in real64.sml and replaced them
508    with code that generates these values from their corresponding
509    integer counterparts.
510    
511    Made all(?) the slice-related changes to the Basis and made everything
512    compile again...
513    
514    ----------------------------------------------------------------------
515    Name: Matthias Blume (blume (at) tti - c (dot) org)
516    Date: 2003/09/15 17:45:00 CDT
517    Tag: blume-20030915-rbase
518    Description:
519    
520    Fixed bug in Real.fromLargeInt.
521    
522    ----------------------------------------------------------------------
523    Name: Matthias Blume (blume (at) tti - c (dot) org)
524    Date: 2003/09/13 18:11:00 CDT
525    Tag: blume-20030913-libinstall
526    Description:
527    
528    Minor bugfix in config/libinstall (set anchor with path to
529    standalone tool after installing it, otherwise libraries that
530    need ml-lex or ml-yacc won't compile the first time the installer
531    runs).
532    
533    ----------------------------------------------------------------------
534    Name: Matthias Blume (blume (at) tti - c (dot) org)
535    Date: 2003/09/12 11:45:00 CDT
536    Tag: blume-20030912-various
537    Description:
538    
539     - fixed bug in Real.toLargeInt
540     - fixed bug in Posix.ProcEnv.times
541     - changed inputLine functions to return an option
542     - minor installer improvements / bugfixes
543     - changed default @SMLalloc parameter for x86/celeron to 64k
544    
545    ----------------------------------------------------------------------
546    Name: Matthias Blume (blume (at) tti - c (dot) org)
547    Date: 2003/09/09 22:00:00 CDT
548    Tag: Release_110_43
549    Description:
550    
551        New working release 110.43.  New bootfiles.
552    
553    ----------------------------------------------------------------------
554    Name: Matthias Blume (blume (at) tti - c (dot) org)
555    Date: 2003/09/09 19:20:00 CDT
556    Tag: blume-20030909-installer
557    Description:
558    
559    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
560    Modified config/install.bat to take advantage of it.  Also modified
561    config/install.sh (and called it config/new-install.sh) to take advantage
562    of it on Unix systems.  (The SML code is (supposed to be) platform-
563    independent.)
564    
565    The installer can now install everything under Win32
566    as well as under *nix as long as it compiles.
567    
568    Other changes:
569    
570      - made CML compile again under Win32
571      - made eXene compile under Win32 (by providing a fake structure UnixSock
572        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
573      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
574        same as type int;  under Win32 it isn't
575      - fixed some slice-related problems in the win32-specific parts of CML
576      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
577        (under Win32, the volume name is case-insensitive, and the
578        OS.Path code compares volume names for equality)
579    
580    ----------------------------------------------------------------------
581    Name: Matthias Blume (blume (at) tti - c (dot) org)
582    Date: 2003/09/08 11:55:00 CDT
583    Tag: blume-20030908-fullpath
584    Description:
585    
586    Made Win32 version of OS.FileSys.fullPath return current directory
587    when given an empty string.  This is what the spec says, and incidentally,
588    CM depends on it.  (CM otherwise goes into an infinite loop in certain
589    cases when presented with the name of a non-existing .cm file.)
590    
591    ----------------------------------------------------------------------
592    Name: Matthias Blume (blume (at) tti - c (dot) org)
593    Date: 2003/09/04 16:30:00 CDT
594    Tag: blume-20030905-slices-etc
595    Description:
596    
597      1. Changed interface to vectors and arrays in Basis to match
598         (draft) Basis spec.
599      2. Added signatures and implementations of slices according to
600         Basis spec.
601      3. Edited source code throughout the system to make it compile again
602         under 1. and 2.  (In some cases code had to be added to have it
603         match the new signatures.)
604      4. MLRISC should be backward-compatible: the copies of the originals
605         of files that needed to change under 3. were retained, the .cm files
606         check the compiler version number and use old versions when
607         appropriate.
608      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
609         dirstream -> string option (in accordance with Basis spec).
610      6. When generating code that counts lines, ml-lex used function
611         CharVector.foldli, taking advantage of its old interface.
612         This has been replaced with the corresponding code from
613         CharVectorSlice.  (html-lex must be re-lexed!)
614      7. BitArray in smlnj-lib/Util has been extended/modified to match the
615         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
616      8. Removed temporary additions (fromInternal, toInternal) from the
617         (now obsolete) IntInf in smlnj-lib/Util.
618      9. Cleaned up structure Byte.
619     10. Added localOffset, scan, and fromString to Date (according to spec).
620         Cleaned/corrected implementation of Date.
621         (Still need to check for correctness; implement better canonicalizeDate.)
622     11. Added "scan" to signature IEEE_REAL.
623     12. Some improvements to IntInf [in particular: efficiency-hack for
624         mod and rem when second operand is 2 (for parity checks).]
625     13. Changed representation of type Time.time, using a single IntInf.int
626         value counting microseconds.  This considerably simplified the
627         implementation of structure Time.  We now support negative time
628         values; scan and fromString handle signs.
629     14. Functor PrimIO now takes two additional arguments (VectorSlice and
630         ArraySlice).
631    
632    ----------------------------------------------------------------------
633    Name: Matthias Blume (blume (at) tti - c (dot) org)
634    Date: 2003/08/28 17:00:00 CDT
635    Tag: blume-20030828-intinf
636    Description:
637    
638    This is a major update which comes with a version number bump
639    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
640    and an implementation of IntInf in the Basis.
641    
642    There are a fairly large number of related changes and updates throughout
643    the system:
644    
645    Basis:
646            - Implemented IntInf.
647            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
648            - Added some missing Real64 operations, most notably Real.toLargeInt.
649            - Added FixedInt as a synonym for Int32.
650    
651    compiler:
652            * Added support for a built-in intinf type.
653              - literals
654              - pattern matching
655              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
656              - overloading on literals and operations
657    
658            This required adding a primitive type intinf, some additional
659            primops, and implementations for several non-trivial intinf
660            operations in Core.  (The intinf type is completely abstract
661            to the compiler; all operations get delegated back to the Core.)
662    
663            * Intinf equality is handled by polyequal.  However, the compiler
664            does not print its usual warning in this case (since polyequal
665            is the right thing to do there).
666    
667            * Improved the organization of structure InlineT.
668    
669            * A word about conversion primops:
670            If conversions involving intinf do not cancel out during
671            CPS contract, then the compiler must insert calls to Core functions.
672            Since all core access must be resolved already during the FLINT
673            translate phase, it would be too late a the time of CPS contract
674            to add new Core calls.  For this reason, conversion primops
675            for intinf carry two arguments:  1. the numeric argument that
676            they are supposed to convert, and 2. the Core function that
677            can help with this conversion if necessary.  If CPS contract
678            eliminates a primop, then the associated Core function becomes
679            dead and goes away.  Intinf conversion primops that do not get
680            eliminated by CPS contract get rewritten into calls of their
681            core functions by a separate, new phase.
682    
683    interactive system:
684            - Control.Print.intinfDepth controls max length of intinf constants
685            being printed. (Analogous to Control.Print.stringDepth.)
686            - Cleanup in printutil and pputil: got rid of unused stuff and
687            duplicates; replaced some of the code with code that makes better
688            use of library functionality.
689    
690    CM:
691            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
692            are no longer silent.
693    
694    CKIT:
695            Fixed mismatched uses of Int32 and LargeInt.  I always decided
696            in favor of LargeInt -- which is now the same as IntInf.
697            CKIT-knowledgable people should check whether this is what's
698            intended and otherwise change things back to using Int32 or
699            FixedInt.
700    
701    Throughout the code:
702            Started using IntInf.int literals and built-in operations
703            (e.g., comparison with 0) where this seems appropriate.
704    
705    
706    ----------------------------------------------------------------------
707    Name: Dave MacQueen (dbm@cs.uchicago.edu)
708    Date: 2003/08/13 11:36:00 CDT
709    Tag: dbm-20030813-mcz-merge1
710    Description:
711    
712    Merging changes from the mcz-branch development branch into trunk.
713    These changes involve replacement of the emulated old prettyprinter
714    interface with direct use of the SML/NJ Lib PP library, and fixing
715    of a couple of bugs (895, 1186) relating to error messages.  A new
716    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
717    has been added.
718    
719    ----------------------------------------------------------------------
720    Name: Matthias Blume (blume (at) tti - c (dot) org)
721    Date: 2003/08/11 15:45:00 CDT
722    Tag: blume-20030811-windows
723    Description:
724    
725    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
726    
727        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
728    
729    This patch restores SML/NJ's ability to run under win32.  There are a
730    number of changes, including fixes for several bugs that had gone
731    unnoticed until now:
732    
733      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
734        trying to run the win32 version from within cygwin.)
735    
736      - There are a number of simple .bat scripts that substitute for their
737        corresponding Unix shell-scripts.  (See below.)
738    
739      - The internals of ml-build have been modified slightly.  The main
740        difference is that instead of calling ".link-sml" (or link-sml.bat)
741        using OS.Process.system, the ML process delegates this task back
742        to the script.  Otherwise problems arise in mixed environments such
743        as Cygwin where scripts look and work like Unix scripts, but
744        where OS.Process.system cannot run them.
745    
746      - In CM, the srcpath pickler used native pathname syntax -- which
747        is incorrect in the case of cross-compilation.  The new pickle format
748        is independent of platform-specific naming conventions.
749    
750      - Path configuration files (such as lib/pathconfig) can now choose
751        between native and standard syntax.  Placing a line of the form
752    
753            standard!
754    
755        into the file causes all subsequent paths to be interpreted using
756        CM standard pathname syntax (= Unix conventions); a line
757    
758            native!
759    
760        switches back to native style.  This was needed so that
761        path config files can be written portably, see src/system/pathconfig.
762    
763      - Runtime system:
764    
765          - win32-filesys.c:  get_file_time and set_file_time now
766            access modification time, not creation time.
767    
768          - I/O code made aware of new array representation.
769    
770          - Bug fixes in X86.prim.masm.
771    
772       - src/system/makeml made aware of win32. (For use under cygwin
773            and other Unix-environments for windows.)
774    
775       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
776         which caused BinIO.inputAll to fail consistently.
777    
778    .bat scripts:
779    
780       Windows .bat scripts assume that SMLNJ_HOME is defined.
781    
782       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
783         applications (sml, ml-yacc, ml-lex).
784       - ml-build.bat: analogous to ml-build.
785       - config\install.bat:  Analogous to config/install.sh.  This requires
786         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
787         (nmake etc. must be on the path, and vcvars32 must have been run.)
788         Moreover, sources for ml-lex and ml-yacc need to exist under src,
789         and the bootfile hierarchy must have been unpacked under
790         sml.boot.x86-win32.
791         The script is very primitive and does a poor job at error checking.
792         It only installs the base system, ml-lex, and ml-yacc.  No other
793         libraries are being installed (i.e., you get only those that
794         are part of the compiler.)
795       - link-sml.bat:  analogous to .link-sml, but not currently used
796    
797    Unrelated bug fixes:
798    
799      - ml-nlffigen now exports structures ST_* corresponding to incomplete
800        types.
801      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
802        otherwise.)
803    
804    ----------------------------------------------------------------------
805    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
806    Date: 2003/06/17
807    Tag: macqueen-20030617-bug895
808    Description:
809    
810    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
811    Tag will be used for new development branch (mcz-branch) for
812    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
813    
814    ----------------------------------------------------------------------
815    Name: Matthias Blume (blume (at) tti - c (dot) org)
816    Date: 2003/05/27 16:55:00 CDT
817    Tag: blume-20030527-polyeq
818    Description:
819    
820    Tried to eliminated most cases of polymorphic equality.
821    
822    ----------------------------------------------------------------------
823    Name: Matthias Blume (blume (at) tti - c (dot) org)
824    Date: 2003/05/21 17:45:00 CDT
825    Tag: blume-20030517-complete
826    Description:
827    
828    Two changes:
829    
830      1. Added a flag for controlling whether non-exhaustive bindings will
831         be treated as errors (default is false).
832      2. Cleaned up the *entire* source tree so that CMB.make goes through
833         without a single non-exhaustive match- or bind warning.
834    
835    ----------------------------------------------------------------------
836    Name: Matthias Blume (blume (at) tti - c (dot) org)
837    Date: 2003/05/17 10:20:00 CDT
838    Tag: blume-20030517-absyn
839    Description:
840    
841    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
842    
843       This mainly affects the quality of error messages.  However, some
844       of the code is now more straightforward than before. (Treatment of
845       the above four constructs in translate.sml is much simpler than
846       the "macro-expansion" that was going on before.  Plus, the mach-
847       compiler no longer gets invoked just to be able to compile an
848       if-expression.)
849    
850    2. The ErrorMsg.Error exception is now caught and absorbed by the
851       interactive loop.
852    
853    ----------------------------------------------------------------------
854    Name: Allen Leung
855    Date: 2003/05/16 13:05:00 CDT
856    Tag: leunga-20030516-cygwin-runtime
857    Description:
858    
859        Ported the runtime system to cygwin, which uses the unix
860    x86-unix bin files.  Missing/buggy features:
861    
862       o getnetbyname, getnetbyaddr: these functions seem to be missing in
863         the Cygwin library.
864       o Ctrl-C handling may be flaky.
865       o Windows system calls and Windows I/O are not supported.
866    
867       A new set of binfiles is located at:
868    
869         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
870    
871       This is only needed for bootstrapping the cygwin version of smlnj.
872       Other x86 versions can use the existing binfiles.
873    
874    ----------------------------------------------------------------------
875    Name: Matthias Blume
876    Date: 2003/04/08 15:42:00 CDT
877    Tag: blume-20030408-listpair
878    Description:
879    
880    1. Added a target 'mlrisc' to installer.
881    
882    2. Added missing elements to structure ListPair.
883    
884    ----------------------------------------------------------------------
885    Name: Allen Leung
886    Date: 2003/01/07 10:40:00 EST
887    Tag: leunga-20030107-int-rem
888    Description:
889    
890    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
891    The arguments to the SUBL instruction were swapped.
892    
893    ----------------------------------------------------------------------
894    Name: Matthias Blume
895    Date: 2002/12/12 16:25:00 EST
896    Tag: blume-20021212-risc-ra
897    Description:
898    
899    Fixed a serious bug in the rewrite code for FP spilling/reloading that
900    sent the RA into an infinite loop when floating point registers get
901    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
902    110.37 and 110.38.)
903    There was another set of potential problems related to the handling of
904    MLRISC annotations (but those did not yet cause real problems, apparently).
905    
906    ----------------------------------------------------------------------
907    Name: Matthias Blume
908    Date: 2002/12/06 22:40:00 EST
909    Tag: blume-20021206-cm-fileid
910    Description:
911    
912    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
913    This fixes the problem of CM getting confused by files that suddenly
914    change their identity (e.g., by getting unlinked and recreated by some
915    text editor such as vi).  There might be a better/cheaper/cleaner way
916    of doing this, but for now this will have to do.
917    
918    ----------------------------------------------------------------------
919    Name: Matthias Blume
920    Date: 2002/10/28 09:50:00 EST
921    Tag: blume-20021028-typecheck
922    Description:
923    
924    Exported structure Typecheck from $smlnj/viscomp/core.cm.
925    
926    ----------------------------------------------------------------------
927    Name: Matthias Blume
928    Date: 2002/10/17 09:10:00 EDT
929    Tag: Release_110_42
930    Description:
931    
932    In good old tradition, there has been a slight hiccup so that we have
933    to patch 110.42 after the fact.   The old release tag has been replaced
934    (see below).
935    
936    The change solves a problem with two competing approaches the
937    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
938    each other's way.
939    
940    This change only affects the runtime system code and the installer script.
941    (No new bootfiles.)
942    
943    ----------------------------------------------------------------------
944    Name: Matthias Blume
945    Date: 2002/10/16 12:00:00 EDT
946    Tag: Release_110_42_removed
947    Description:
948    
949    New working release.  New bootfiles.
950    
951    ----------------------------------------------------------------------
952    Name: Matthias Blume
953    Date: 2002/10/10 13:10:00 EDT
954    Tag: blume-20021010-ppc-divs
955    Description:
956    
957    The mltree operator DIVS must be implemented with an overflow check on
958    the PPC because the hardware indicates divide-by-zero using "overflow" as
959    well.
960    
961    ----------------------------------------------------------------------
962    Name: Matthias Blume
963    Date: 2002/07/23 11:45:00
964    Tag: blume-20020723-smlnj-home
965    Description:
966    
967    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
968    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
969    CM_PATHCONFIG is also set), the path configuration file is assumed
970    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
971    the entire tree to some other place and everything will "just work".
972    
973    (Companion commands such as ml-build and ml-makedepend also sense this
974    variable.)
975    
976    ----------------------------------------------------------------------
977    Name: Matthias Blume
978    Date: 2002/07/12 21:19:00 EDT
979    Tag: blume-20020712-liveness
980    Description:
981    
982    Exported two useful "step" functions from liveness module (MLRISC).
983    
984    ----------------------------------------------------------------------
985    Name: Matthias Blume
986    Date: 2002/07/05 16:00 EDT
987    Tag: Release_110_41
988    Description:
989    
990    New working release.  New bootfiles.
991    
992    ----------------------------------------------------------------------
993    Name: Matthias Blume
994    Date: 2002/07/05 10:25:00 EDT
995    Tag: blume-20020705-btimp
996    Description:
997    
998    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
999    other clients can set up backtracing support.
1000    
1001    ----------------------------------------------------------------------
1002    Name: Matthias Blume
1003    Date: 2002/06/25 17:23:00 EDT
1004    Tag: blume-20020625-fpmax
1005    Description:
1006    
1007    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
1008    case.  (The sense of the isNaN test was reversed -- which made min and
1009    max always return their first argument.)
1010    
1011    ----------------------------------------------------------------------
1012    Name: Matthias Blume
1013    Date: 2002/06/11
1014    Tag: blume-20020611-unixpath
1015    Description:
1016    
1017    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
1018    
1019    ----------------------------------------------------------------------
1020    Name: Matthias Blume
1021    Date: 2002/06/10 16:35:00 EDT
1022    Tag: blume-20020610-ieeereal
1023    Description:
1024    
1025    I back-ported my implementation of IEEEReal.fromString from the
1026    idlbasis-devel branch so that we can test it.
1027    
1028    Another small change is that ppDec tries to give more information
1029    than just "<sig>" in the case of functors.  However, this code is
1030    broken in some mysterious way if the functor's body's signature
1031    has not been declared by ascription but gets inferred from the
1032    implementation.  This needs fixing...
1033    
1034    ----------------------------------------------------------------------
1035    Name: Matthias Blume
1036    Date: 2002/05/31
1037    Tag: blume-20020531-btrace-mode
1038    Description:
1039    
1040    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
1041    the wayside when I switched over to using Controls everywhere.)
1042    
1043    ----------------------------------------------------------------------
1044    Name: Lal George
1045    Date: 2002/05/23 12:21:40 EDT
1046    Tag: george-20020523-visual-labels
1047    Description:
1048    
1049    Labels are now displayed in the graphical output to make
1050    the fall-through and target blocks obvious.
1051    
1052    ----------------------------------------------------------------------
1053    Name: Matthias Blume
1054    Date: 2002/05/22 11:03:00 EDT
1055    Tag: blume-20020522-shrink
1056    Description:
1057    
1058    John tweaked yesterday's fix for 1131 to handle an out-of-memory
1059    situation that comes up when allocating huge arrays.
1060    
1061    ----------------------------------------------------------------------
1062    Name: Matthias Blume
1063    Date: 2002/05/21 16:00:00 EDT
1064    Tag: Release_110_40
1065    Description:
1066    
1067    New working release (110.40).  New bootfiles.
1068    
1069    [Also: John Reppy fixed GC bug 1131.]
1070    
1071    ----------------------------------------------------------------------
1072    Name: Matthias Blume
1073    Date: 2002/05/21 12:35:00 EDT
1074    Tag: blume-20020521-cmdoc
1075    Description:
1076    
1077    CM documentation update.
1078    
1079    ----------------------------------------------------------------------
1080    Name: Matthias Blume
1081    Date: 2002/05/21 10:55:00 EDT
1082    Tag: blume-20020521-misc
1083    Description:
1084    
1085    - John tweaked runtime to be silent on heap export (except when
1086      GC messages are on).
1087    - I added a few more things (cross-compiling versions of CMB) to
1088      config/preloads (as suggestions).
1089    
1090    ----------------------------------------------------------------------
1091    Name: Matthias Blume
1092    Date: 2002/05/20 22:25:00 EDT
1093    Tag: blume-20020520-controls
1094    Description:
1095    
1096    - Added ControlUtil structure to control-lib.cm.
1097    - Use it throughout.
1098    - Used Controls facility to define MLRISC controls (as opposed to
1099      registering MLRISC control ref cells with Controls after the fact)
1100    - Fixed messed-up controls priorities.
1101    
1102    * Removed again all the stuff from config/preloads that one wouldn't
1103      be able to preload at the time the initial heap image is built.
1104      (Many libraries, e.g., CML, do not exist yet at this time.  The
1105       only libraries that can be preloaded via config/preloads are those
1106       that come bundled with the bootfiles.)
1107    
1108    ----------------------------------------------------------------------
1109    Name: Matthias Blume
1110    Date: 2002/05/20 10:59:00 EDT
1111    Tag: blume-20020520-preloads
1112    Description:
1113    
1114    Added a lot of commented-out suggestions for things to be included
1115    in config/preloads.
1116    
1117    ----------------------------------------------------------------------
1118    Name: Allen Leung
1119    Date: 2002/05/18 14:20:00 EDT
1120    Tag: leunga-20020518-mdl
1121    Description:
1122    
1123    o Made the mdl tool stuff compile and run again.
1124    o I've disabled all the stuff that depends on RTL specifications; they
1125      are all badly broken anyway.
1126    
1127    ----------------------------------------------------------------------
1128    Name: Matthias Blume
1129    Date: 2002/05/17 16:49:00 EDT
1130    Tag: blume-20020517-controls
1131    Description:
1132    
1133    0. John Reppy made several modifications to the SML/NJ library.
1134       In particular, there is a shiny new controls-lib.cm.
1135    
1136    1. Pushed new controls interface through compiler so that everything
1137       compiles again.
1138    
1139    2. Added FormatComb and FORMAT_COMB to the CML version of the
1140       SML/NJ library (so that CML compiles again).
1141    
1142    3. Modified init scripts because XXX_DEFAULT environment variables
1143       are no longer with us.  (Boot-time initialization is now done
1144       using the same environment variables that are also used for
1145       startup-time initialization of controls.)
1146    
1147    ----------------------------------------------------------------------
1148    Name: Lal George
1149    Date: 2002/05/15 09:20:10 EDT
1150    Tag: george-20020515-pseudo-op-decls
1151    Description:
1152    
1153            All pseudo-ops emitted before the first segment declaration
1154    such as TEXT, DATA, and BSS directives are assumed to be global
1155    declarations and are emitted first in the assembly file. This is
1156    useful in a number of situations where one has pseudo-ops that are not
1157    specific to any segment, and also works around the constraint that one
1158    cannot have client pseudo-ops in the TEXT segment.
1159    
1160            Because no segment is associated with these declarations it is
1161    an error to allocate any space or objects before the first segment
1162    directive and an exception will be raised. However, we cannot make
1163    this check for client pseudo-ops.
1164    
1165            These top level declarations are a field in the CFG graph_info.
1166    In theory you can continue to add to this field after the CFG has been
1167    built -- provided you know what you are doing;-)
1168    
1169    ----------------------------------------------------------------------
1170    Name: Matthias Blume
1171    Date: 2002/05/13 16:40:00 EDT
1172    Tag: blume-20020513-pp-etc
1173    Description:
1174    
1175    A few minor bugfixes:
1176    
1177      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1178        (Bogus printouts for redefined bindings still occur.  Compiler
1179        bug should no longer occur now.  We need to redo the prettyprinter
1180        from scratch.)
1181    
1182      - CM pathname printer now also adds escape sequences for ( and )
1183    
1184      - commend and docu fixes for ml-nlffi
1185    
1186    ----------------------------------------------------------------------
1187    Name: Matthias Blume
1188    Date: 2002/05/10 16:40:00 EDT
1189    Tag: blume-20020510-erg-textio
1190    Description:
1191    
1192    Applied the following bugfix provided by Emden Gansner:
1193    
1194        Output is corrupted when outputSubstr is used rather than output.
1195    
1196        The problem occurs when a substring
1197    
1198            ss = (s, dataStart, dataLen)
1199    
1200        where dataStart > 0, fills a stream buffer with avail bytes left.
1201        avail bytes of s, starting at index dataStart, are copied into the
1202        buffer, the buffer is flushed, and then the remaining dataLen-avail
1203        bytes of ss are copied into the beginning of the buffer. Instead of
1204        starting this copy at index dataStart+avail in s, the current code
1205        starts the copy at index avail.
1206    
1207        Fix:
1208        In text-io-fn.sml, change line 695 from
1209             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1210        to
1211             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1212    
1213    ----------------------------------------------------------------------
1214    Name: Matthias Blume
1215    Date: 2002/04/12 13:55:00 EDT
1216    Tag: blume-20020412-assyntax
1217    Description:
1218    
1219    1. Grabbed newer assyntax.h from the XFree86 project.
1220    2. Fiddled with how to compile X86.prim.asm without warnings.
1221    3. (Very) Minor cleanup in CM.
1222    
1223    ----------------------------------------------------------------------
1224    Name: Matthias Blume
1225    Date: 2002/04/01 (no joke!) 17:07:00 EST
1226    Tag: blume-20020401-x86div
1227    Description:
1228    
1229    Added full support for div/mod/rem/quot on the x86, using the machine
1230    instruction's two results (without clumsily recomputing the remainder)
1231    directly where appropriate.
1232    
1233    Some more extensive power-of-two support was added to the x86 instruction
1234    selector (avoiding expensive divs, mods, and muls where they can be
1235    replaced with cheaper shifts and masks).  However, this sort of thing
1236    ought to be done earlier, e.g., within the CPS optimizer so that
1237    all architectures benefit from it.
1238    
1239    The compiler compiles to a fixed point, but changes might be somewhat
1240    fragile nevertheless.  Please, report any strange things that you might
1241    see wrt. div/mod/quot/rem...
1242    
1243    ----------------------------------------------------------------------
1244    Name: Matthias Blume
1245    Date: 2002/03/29 17:22:00
1246    Tag: blume-20020329-div
1247    Description:
1248    
1249    Fixed my broken div/mod logic.  Unfortunately, this means that the
1250    inline code for div/mod now has one more comparison than before.
1251    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1252    The problem was with quotient = 0, because that alone does not tell
1253    us which way the rounding went.  One then has to look at whether
1254    remainder and divisor have the same sign...  :(
1255    
1256    Anyway, I replaced the bootfiles with fresh ones...
1257    
1258    ----------------------------------------------------------------------
1259    Name: Matthias Blume
1260    Date: 2002/03/29 14:10:00 EST
1261    Tag: blume-20020329-inlprims
1262    Description:
1263    
1264    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1265    
1266    Primops have changed. This means that the bin/boot-file formats have
1267    changed as well.
1268    
1269    To make sure that there is no confusion, I made a new version.
1270    
1271    
1272    CHANGES:
1273    
1274    * removed REMT from mltree (remainder should never overflow).
1275    
1276    * added primops to deal with divisions of all flavors to the frontend
1277    
1278    * handled these primops all the way through so they map to their respective
1279      MLRISC support
1280    
1281    * used these primops in the implementation of Int, Int32, Word, Word32
1282    
1283    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1284    
1285    * parameterized INLMIN, INLMAX, and INLABS by a numkind
1286    
1287    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1288      floating point
1289    
1290    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1291      and Real (but Real.abs maps to a separate floating-point-only primop)
1292    
1293    
1294    TODO items:
1295    
1296    * Hacked Alpha32 instruction selection, disabling the selection of REMx
1297      instructions because the machine instruction encoder cannot handle
1298      them.  (Hppa, PPC, and Sparc instruction selection did not handle
1299      REM in the first place, and REM is supported by the x86 machine coder.)
1300    
1301    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1302      selection phase.  (The two can be streamlined because the hardware
1303      delivers both quotient and remainder at the same time anyway.)
1304    
1305    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1306      (Currently the behavior is inconsistent both across architectures and
1307      wrt. the draft Basis spec.)
1308    
1309    * Word8 should eventually be handled natively, too.
1310    
1311    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1312      as if there currently is no execution path that could trigger it in
1313      SML/NJ.  (The assumptions underlying functions arith and promotable do not
1314      hold for things like multiplication and division.)
1315    
1316    ----------------------------------------------------------------------
1317    Name: Matthias Blume
1318    Date: 2002/03/27 16:27:00 EST
1319    Tag: blume-20020327-mlrisc-divisions
1320    Description:
1321    
1322    Added support for all four division operations (ML's div, mod, quot,
1323    and rem) to MLRISC.  In the course of doing so, I also rationalized
1324    the naming (no more annoying switch-around of DIV and QUOT), by
1325    parameterizing the operation by div_rounding_mode (which can be either
1326    DIV_TO_ZERO or DIV_TO_NEGINF).
1327    
1328    The generic MLTreeGen functor takes care of compiling all four
1329    operations down to only round-to-zero div.
1330    
1331    Missing pieces:
1332    
1333      * Doing something smarter than relying on MLTreeGen on architectures
1334        like, e.g., the x86 where hardware division delivers both quotient and
1335        remainder at the same time.  With this, the implementation of the
1336        round-to-neginf operations could be further streamlined.
1337    
1338      * Remove inlining support for div/mod/rem from the frontend and replace it
1339        with primops that get carried through to the backend.  Do this for all
1340        int and word types.
1341    
1342    ----------------------------------------------------------------------
1343    Name: Matthias Blume
1344    Date: 2002/03/25 17:25:00 EST
1345    Tag: blume-20020325-divmod
1346    Description:
1347    
1348    I improved (hopefully without breaking them) the implementation of Int.div,
1349    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1350    advantage of the following observations:
1351    
1352      Let  q = x quot y      r = x rem y
1353           d = x div  y      m = x mod y
1354    
1355    where "quot" is the round-to-zero version of integer division that
1356    hardware usually provides.  Then we have:
1357    
1358         r = x - q * y        where neither the * nor the - will overflow
1359         d = if q >= 0 orelse x = q * y then q else q - 1
1360                              where neither the * nor the - will overflow
1361         m = if q >= 0 orelse r = 0 then r else r + y
1362                              where the + will not overflow
1363    
1364    This results in substantial simplification of the generated code.
1365    The following table shows the number of CFG nodes and edges generated
1366    for
1367            fun f (x, y) = x OPER y
1368            (* with OPER \in div, mod, quot, rem *)
1369    
1370    
1371        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1372        --------------------------------------------------------
1373         div |         24 |         39 |         12 |         16
1374         mod |         41 |         71 |         12 |         16
1375        quot |          8 |         10 |          8 |         10
1376         rem |         10 |         14 |          8 |         10
1377    
1378    
1379    ----------------------------------------------------------------------
1380    Name: Matthias Blume
1381    Date: 2002/03/25 22:06:00 EST
1382    Tag: blume-20020325-cprotobug
1383    Description:
1384    
1385    Fixed a bug in cproto (c prototype decoder).
1386    
1387    ----------------------------------------------------------------------
1388    Name: Matthias Blume
1389    Date: 2002/03/25 16:00:00 EST
1390    Tag: blume-20020325-raw-primops
1391    Description:
1392    
1393    I did some cleanup to Allen's new primop code and
1394    replaced yesterday's bootfiles with new ones.
1395    (But they are stored in the same place.)
1396    
1397    ----------------------------------------------------------------------
1398    Name: Matthias Blume
1399    Date: 2002/03/24 22:40:00 EST
1400    Tag: blume-20020324-bootfiles
1401    Description:
1402    
1403    Made the bootfiles that Allen asked for.
1404    
1405    ----------------------------------------------------------------------
1406    Name: Allen Leung
1407    Date: 2002/03/23 15:50:00 EST
1408    Tag: leunga-20020323-flint-cps-rcc-primops
1409    Description:
1410    
1411      1. Changes to FLINT primops:
1412    
1413        (* make a call to a C-function;
1414         * The primop carries C function prototype information and specifies
1415         * which of its (ML-) arguments are floating point. C prototype
1416         * information is for use by the backend, ML information is for
1417         * use by the CPS converter. *)
1418      | RAW_CCALL of { c_proto: CTypes.c_proto,
1419                       ml_args: ccall_type list,
1420                       ml_res_opt: ccall_type option,
1421                       reentrant : bool
1422                     } option
1423       (* Allocate uninitialized storage on the heap.
1424        * The record is meant to hold short-lived C objects, i.e., they
1425        * are not ML pointers.  With the tag, the representation is
1426        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1427        *)
1428      | RAW_RECORD of {tag:bool,sz:int}
1429      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1430    
1431      2.  These CPS primops are now overloaded:
1432    
1433           rawload of {kind:numkind}
1434           rawstore of {kind:numkind}
1435    
1436          The one argument form is:
1437    
1438             rawload {kind} address
1439    
1440          The two argument form is:
1441    
1442             rawload {kind} [ml object, byte-offset]
1443    
1444      3. RAW_CCALL/RCC now takes two extra arguments:
1445    
1446         a. The first is whether the C call is reentrant, i.e., whether
1447            ML state should be saved and restored.
1448         b. The second argument is a string argument specifying the name of
1449            library and the C function.
1450    
1451         These things are currently not handled in the code generator, yet.
1452    
1453      4. In CProto,
1454    
1455         An encoding type of "bool" means "ml object" and is mapped into
1456         C prototype of PTR.  Note that "bool" is different than "string",
1457         even though "string" is also mapped into PTR, because "bool"
1458         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1459    
1460      5. Pickler/unpicker
1461    
1462         Changed to handle RAW_RECORD and newest RAW_CCALL
1463    
1464      6. MLRiscGen,
1465    
1466         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1467         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1468            in the file CodeGen/cpscompile/cps-c-calls.sml
1469    
1470      7. Added the conditional move operator
1471    
1472             condmove of branch
1473    
1474         to cps.  Generation of this is still buggy so it is currently
1475         disabled.
1476    
1477    ----------------------------------------------------------------------
1478    Name: Lal George
1479    Date: 2002/03/22 14:18:25 EST
1480    Tag: george-20020322-cps-branch-prob
1481    Description:
1482    
1483    Implemented the Ball-Larus branch prediction-heuristics, and
1484    incorporated graphical viewers for control flow graphs.
1485    
1486    Ball-Larus Heuristics:
1487    ---------------------
1488    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1489    
1490    By design it uses the Dempster-Shafer theory for combining
1491    probabilities.  For example, in the function:
1492    
1493        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1494    
1495    the ball-larus heuristics predicts that the n=0 is unlikely
1496    (OH-heuristic), and the 'then' branch is unlikely because of the
1497    RH-heuristic -- giving the 'then' branch an even lower combined
1498    probability using the Dempster-Shafer theory.
1499    
1500    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1501    probability of the 'then' branch because of the loop in the else
1502    branch.
1503    
1504    
1505    Graphical Viewing:
1506    ------------------
1507    I merely plugged in Allen's graphical viewers into the compiler. The
1508    additional code is not much. At the top level, saying:
1509    
1510            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1511    
1512    will display the graphical view of the control flow graph just before
1513    back-patching.  daVinci must be in your path for this to work. If
1514    daVinci is not available, then the default viewer can be changed
1515    using:
1516    
1517            Control.MLRISC.getString "viewer"
1518    
1519    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1520    course, these viewers must be in your path.
1521    
1522    The above will display the compilation unit at the level of clusters,
1523    many of which are small, boring, and un-interesting. Also setting:
1524    
1525            Control.MLRISC.getInt "cfg-graphical-view_size"
1526    
1527    will display clusters that are larger than the value set by the above.
1528    
1529    
1530    ----------------------------------------------------------------------
1531    Name: Matthias Blume
1532    Date: 2002/03/21 22:20:00 EST
1533    Tag: blume-20020321-kmp-bugfix
1534    Description:
1535    
1536    Changed the interface to the KMP routine in PreString and fixed
1537    a minor bug in one place where it was used.
1538    
1539    ----------------------------------------------------------------------
1540    Name: Allen Leung
1541    Date: 2002/03/21 20:30:00 EST
1542    Tag: leunga-20020321-cfg
1543    Description:
1544    
1545      Fixed a potential problem in cfg edge splitting.
1546    
1547    ----------------------------------------------------------------------
1548    Name: Allen Leung
1549    Date: 2002/03/21 17:15:00 EST
1550    Tag: leunga-20020321-x86-fp-cfg
1551    Description:
1552    
1553      1. Recoded the buggy parts of x86-fp.
1554    
1555         a. All the block reordering code has been removed.
1556            We now depend on the block placement phases to do this work.
1557    
1558         b. Critical edge splitting code has been simplified and moved into the
1559            CFG modules, as where they belong.
1560    
1561         Both of these were quite buggy and complex.  The code is now much, much
1562         simpler.
1563    
1564      2. X86 backend.
1565    
1566         a. Added instructions for 64-bit support.  Instruction selection for
1567            64-bit has not been committed, however, since that
1568            requires changes to MLTREE which haven't been approved by
1569            Lal and John.
1570    
1571         b. Added support for FUCOMI and FUCOMIP when generating code for
1572            PentiumPro and above.  We only generate these instructions in
1573            the fast-fp mode.
1574    
1575         c. Added cases for JP and JNP in X86FreqProps.
1576    
1577      3. CFG
1578    
1579         CFG now has a bunch of methods for edge splitting and merging.
1580    
1581      4. Machine description.
1582    
1583         John's simplification of MLTREE_BASIS.fcond broke a few machine
1584         description things:
1585    
1586         rtl-build.{sig,sml} and hppa.mdl fixed.
1587    
1588         NOTE: the machine description stuff in the repository is still broken.
1589               Again, I can't put my fixes in because that involves
1590               changes to MLTREE.
1591    
1592    ----------------------------------------------------------------------
1593    Name: Matthias Blume
1594    Date: 2002/03/20 15:55:00 EST
1595    Tag: blume-20020320-kmp
1596    Description:
1597    
1598    Implemented Knuth-Morris-Pratt string matching in PreString and used
1599    it for String.isSubstring, Substring.isSubstring, and
1600    Substring.position.
1601    
1602    (Might need some stress-testing.  Simple examples worked fine.)
1603    
1604    ----------------------------------------------------------------------
1605    Name: Matthias Blume
1606    Date: 2002/03/19 16:37:00 EST
1607    Tag: blume-20020319-witnesses
1608    Description:
1609    
1610    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1611    
1612    This implements a generic mechanism for changing constness qualifiers
1613    anywhere within big C types without resorting to outright "casts".
1614    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1615    modify the constness at the outermost level.)
1616    The implementation of "convert" is based on the idea of "witness"
1617    values -- values that are not used by the operation but whose types
1618    "testify" to their applicability.  On the implementation side, "convert"
1619    is simply a projection (returning its second curried argument).  With
1620    cross-module inlining, it should not result in any machine code being
1621    generated.
1622    
1623    ----------------------------------------------------------------------
1624    Name: Matthias Blume
1625    Date: 2002/03/15 16:40:00 EST
1626    Tag: blume-20020315-basis
1627    Description:
1628    
1629    Provided (preliminary?) implementations for
1630    
1631      {String,Substring}.{concatWith,isSuffix,isSubstring}
1632    
1633    and
1634    
1635      Substring.full
1636    
1637    Those are in the Basis spec but they were missing in SML/NJ.
1638    
1639    ----------------------------------------------------------------------
1640    Name: Matthias Blume
1641    Date: 2002/03/14 21:30:00 EST
1642    Tag: blume-20020314-controls
1643    Description:
1644    
1645    Controls:
1646    ---------
1647    
1648    1. Factored out the recently-added Controls : CONTROLS stuff and put
1649       it into its own library $/controls-lib.cm.  The source tree for
1650       this is under src/smlnj-lib/Controls.
1651    
1652    2. Changed the names of types and functions in this interface, so they
1653       make a bit more "sense":
1654    
1655          module -> registry
1656          'a registry -> 'a group
1657    
1658    3. The interface now deals in ref cells only.  The getter/setter interface
1659       is (mostly) gone.
1660    
1661    4. Added a function that lets one register an already-existing ref cell.
1662    
1663    5. Made the corresponding modifications to the rest of the code so that
1664       everything compiles again.
1665    
1666    6. Changed the implementation of Controls.MLRISC back to something closer
1667       to the original.  In particular, this module (and therefore MLRISC)
1668       does not depend on Controls.  There now is some link-time code in
1669       int-sys.sml that registers the MLRISC controls with the Controls
1670       module.
1671    
1672    CM:
1673    ---
1674    
1675      * One can now specify the lambda-split aggressiveness in init.cmi.
1676    
1677    ----------------------------------------------------------------------
1678    Name: Allen Leung
1679    Date: 2002/03/13 17:30:00 EST
1680    Tag: leunga-20020313-x86-fp-unary
1681    Description:
1682    
1683    Bug fix for:
1684    
1685    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1686    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1687    > - fun f(x,(y,z)) = Real.~ y;
1688    > [autoloading]
1689    > [autoloading done]
1690    >       fchsl   (%eax), 184(%esp)
1691    > Error: MLRisc bug: X86MCEmitter.emitInstr
1692    >
1693    > uncaught exception Error
1694    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1695    
1696    The problem was that the code generator did not generate any fp registers
1697    in this case, and the ra didn't know that it needed to run the X86FP phase to
1698    translate the pseudo fp instruction.   This only happened with unary fp
1699    operators in certain situations.
1700    
1701    ----------------------------------------------------------------------
1702    Name: Matthias Blume
1703    Date: 2002/03/13 14:00:00 EST
1704    Tag: blume-20020313-overload-etc
1705    Description:
1706    
1707    1. Added _overload as a synonym for overload for backward compatibility.
1708       (Control.overloadKW must be true for either version to be accepted.)
1709    
1710    2. Fixed bug in install script that caused more things to be installed
1711       than what was requested in config/targets.
1712    
1713    3. Made CM aware of the (_)overload construct so that autoloading
1714       works.
1715    
1716    ----------------------------------------------------------------------
1717    Name: Matthias Blume
1718    Date: 2002/03/12 22:03:00 EST
1719    Tag: blume-20020312-url
1720    Description:
1721    
1722    Forgot to update BOOT and srcarchiveurl.
1723    
1724    ----------------------------------------------------------------------
1725    Name: Matthias Blume
1726    Date: 2002/03/12 17:30:00 EST
1727    Tag: blume-20020312-version110392
1728    Description:
1729    
1730    Yet another version number bump (because of small changes to the
1731    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1732    
1733    Changes:
1734    
1735      The new pid generation scheme described a few weeks ago was overly
1736      complicated.  I implemented a new mechanism that is simpler and
1737      provides a bit more "stability":  Once CM has seen a compilation
1738      unit, it keeps its identity constant (as long as you do not delete
1739      those crucial CM/GUID/* files).  This means that when you change
1740      an interface, compile, then go back to the old interface, and
1741      compile again, you arrive at the original pid.
1742    
1743      There now also is a mechanism that instructs CM to use the plain
1744      environment hash as a module's pid (effectively making its GUID
1745      the empty string).  For this, "noguid" must be specified as an
1746      option to the .sml file in question within its .cm file.
1747      This is most useful for code that is being generated by tools such
1748      as ml-nlffigen (because during development programmers tend to
1749      erase the tool's entire output directory tree including CM's cached
1750      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1751      revert to the old, broken behavior of SML/NJ, but in specific cases
1752      where there is no danger of interface confusion, its use is ok
1753      (I think).
1754    
1755      ml-nlffigen by default generates "noguid" annotations.  They can be
1756      turned off by specifying -guid in its command line.
1757    
1758    ----------------------------------------------------------------------
1759    Name: Lal George
1760    Date: 2002/03/12 12 14:42:36 EST
1761    Tag: george-20020312-frequency-computation
1762    Description:
1763    
1764    Integrated jump chaining and static block frequency into the
1765    compiler. More details and numbers later.
1766    
1767    ----------------------------------------------------------------------
1768    Name: Lal George
1769    Date: 2002/03/11 11 22:38:53 EST
1770    Tag: george-20020311-jump-chain-elim
1771    Description:
1772    
1773    Tested the jump chain elimination on all architectures (except the
1774    hppa).  This is on by default right now and is profitable for the
1775    alpha and x86, however, it may not be profitable for the sparc and ppc
1776    when compiling the compiler.
1777    
1778    The gc test will typically jump to a label at the end of the cluster,
1779    where there is another jump to an external cluster containing the actual
1780    code to invoke gc. This is to allow factoring of common gc invocation
1781    sequences. That is to say, we generate:
1782    
1783            f:
1784               testgc
1785               ja   L1      % jump if above to L1
1786    
1787            L1:
1788               jmp L2
1789    
1790    
1791    After jump chain elimination the 'ja L1' instructions is converted to
1792    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1793    up being implemented in their long form (if L2 is far away) using:
1794    
1795            jbe     L3      % jump if below or equal to L3
1796            jmp     L2
1797         L3:
1798            ...
1799    
1800    
1801    For large compilation units L2  may be far away.
1802    
1803    
1804    ----------------------------------------------------------------------
1805    Name: Matthias Blume
1806    Date: 2002/03/11 13:30:00 EST
1807    Tag: blume-20020311-mltreeeval
1808    Description:
1809    
1810    A functor parameter was missing.
1811    
1812    ----------------------------------------------------------------------
1813    Name: Allen Leung
1814    Date: 2002/03/11 10:30:00 EST
1815    Tag: leunga-20020311-runtime-string0
1816    Description:
1817    
1818       The representation of the empty string now points to a
1819    legal null terminated C string instead of unit.  It is now possible
1820    to convert an ML string into C string with InlineT.CharVector.getData.
1821    This compiles into one single machine instruction.
1822    
1823    ----------------------------------------------------------------------
1824    Name: Allen Leung
1825    Date: 2002/03/10 23:55:00 EST
1826    Tag: leunga-20020310-x86-call
1827    Description:
1828    
1829       Added machine generation for CALL instruction (relative displacement mode)
1830    
1831    ----------------------------------------------------------------------
1832    Name: Matthias Blume
1833    Date: 2002/03/08 16:05:00
1834    Tag: blume-20020308-entrypoints
1835    Description:
1836    
1837    Version number bumped to 110.39.1.  NEW BOOTFILES!
1838    
1839    Entrypoints: non-zero offset into a code object where execution should begin.
1840    
1841    - Added the notion of an entrypoint to CodeObj.
1842    - Added reading/writing of entrypoint info to Binfile.
1843    - Made runtime system bootloader aware of entrypoints.
1844    - Use the address of the label of the first function given to mlriscGen
1845      as the entrypoint.  This address is currently always 0, but it will
1846      not be 0 once we turn on block placement.
1847    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
1848      with entry points) from mlriscGen.
1849    
1850    ----------------------------------------------------------------------
1851    Name: Allen Leung
1852    Date: 2002/03/07 20:45:00 EST
1853    Tag: leunga-20020307-x86-cmov
1854    Description:
1855    
1856       Bug fixes for CMOVcc on x86.
1857    
1858       1. Added machine code generation for CMOVcc
1859       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1860       3. CMOVcc cannot have an immediate operand as argument.
1861    
1862    ----------------------------------------------------------------------
1863    Name: Matthias Blume
1864    Date: 2002/03/07 16:15:00 EST
1865    Tag: blume-20020307-controls
1866    Description:
1867    
1868    This is a very large but mostly boring patch which makes (almost)
1869    every tuneable compiler knob (i.e., pretty much everything under
1870    Control.* plus a few other things) configurable via both the command
1871    line and environment variables in the style CM did its configuration
1872    until now.
1873    
1874    Try starting sml with '-h' (or, if you are brave, '-H')
1875    
1876    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1877    implements the underlying generic mechanism.
1878    
1879    The interface to some of the existing such facilities has changed somewhat.
1880    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1881    (The getFoo interface is still there for backward-compatibility, but its
1882    use is deprecated.)
1883    
1884    The ml-build script passes -Cxxx=yyy command-line arguments through so
1885    that one can now twiddle the compiler settings when using this "batch"
1886    compiler.
1887    
1888    TODO items:
1889    
1890    We should go through and throw out all controls that are no longer
1891    connected to anything.  Moreover, we should go through and provide
1892    meaningful (and correct!) documentation strings for those controls
1893    that still are connected.
1894    
1895    Currently, multiple calls to Controls.new are accepted (only the first
1896    has any effect).  Eventually we should make sure that every control
1897    is being made (via Controls.new) exactly once.  Future access can then
1898    be done using Controls.acc.
1899    
1900    Finally, it would probably be a good idea to use the getter-setter
1901    interface to controls rather than ref cells.  For the time being, both
1902    styles are provided by the Controls module, but getter-setter pairs are
1903    better if thread-safety is of any concern because they can be wrapped.
1904    
1905    *****************************************
1906    
1907    One bug fix: The function blockPlacement in three of the MLRISC
1908    backpatch files used to be hard-wired to one of two possibilities at
1909    link time (according to the value of the placementFlag).  But (I
1910    think) it should rather sense the flag every time.
1911    
1912    *****************************************
1913    
1914    Other assorted changes (by other people who did not supply a HISTORY entry):
1915    
1916    1. the cross-module inliner now works much better (Monnier)
1917    2. representation of weights, frequencies, and probabilities in MLRISC
1918       changed in preparation of using those for weighted block placement
1919       (Reppy, George)
1920    
1921    ----------------------------------------------------------------------
1922    Name: Lal George
1923    Date: 2002/03/07 14:44:24 EST 2002
1924    Tag: george-20020307-weighted-block-placement
1925    
1926    Tested the weighted block placement optimization on all architectures
1927    (except the hppa) using AMPL to generate the block and edge frequencies.
1928    Changes were required in the machine properties to correctly
1929    categorize trap instructions. There is an MLRISC flag
1930    "weighted-block-placement" that can be used to enable weighted block
1931    placement, but this will be ineffective without block/edge
1932    frequencies (coming soon).
1933    
1934    
1935    ----------------------------------------------------------------------
1936    Name: Lal George
1937    Date: 2002/03/05 17:24:48 EST
1938    Tag: george-20020305-linkage-cluster
1939    
1940    In order to support the block placement optimization, a new cluster
1941    is generated as the very first cluster (called the linkage cluster).
1942    It contains a single jump to the 'real' entry point for the compilation
1943    unit. Block placement has no effect on the linkage cluster itself, but
1944    all the other clusters  have full freedom in the manner in which they
1945    reorder blocks or functions.
1946    
1947    On the x86 the typical linkage code that is generated is:
1948       ----------------------
1949            .align 2
1950       L0:
1951            addl    $L1-L0, 72(%esp)
1952            jmp     L1
1953    
1954    
1955            .align  2
1956       L1:
1957       ----------------------
1958    
1959    72(%esp) is the memory location for the stdlink register. This
1960    must contain the address of the CPS function being called. In the
1961    above example, it contains the address of  L0; before
1962    calling L1 (the real entry point for the compilation unit), it
1963    must contain the address for L1, and hence
1964    
1965            addl $L1-L0, 72(%esp)
1966    
1967    I have tested this on all architectures except the hppa.The increase
1968    in code size is of course negligible
1969    
1970    ----------------------------------------------------------------------
1971    Name: Allen Leung
1972    Date: 2002/03/03 13:20:00 EST
1973    Tag: leunga-20020303-mlrisc-tools
1974    
1975      Added #[ ... ] expressions to mlrisc tools
1976    
1977    ----------------------------------------------------------------------
1978    Name: Matthias Blume
1979    Date: 2002/02/27 12:29:00 EST
1980    Tag: blume-20020227-cdebug
1981    Description:
1982    
1983    - made types in structure C and C_Debug to be equal
1984    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1985    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1986    
1987    ----------------------------------------------------------------------
1988    Name: Matthias Blume
1989    Date: 2002/02/26 12:00:00 EST
1990    Tag: blume-20020226-ffi
1991    Description:
1992    
1993    1. Fixed a minor bug in CM's "noweb" tool:
1994       If numbering is turned off, then truly don't number (i.e., do not
1995       supply the -L option to noweb).  The previous behavior was to supply
1996       -L'' -- which caused noweb to use the "default" line numbering scheme.
1997       Thanks to Chris Richards for pointing this out (and supplying the fix).
1998    
1999    2. Once again, I reworked some aspects of the FFI:
2000    
2001       A. The incomplete/complete type business:
2002    
2003       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
2004         gone!
2005       - ML types representing an incomplete type are now *equal* to
2006         ML types representing their corresponding complete types (just like
2007         in C).  This is still safe because ml-nlffigen will not generate
2008         RTTI for incomplete types, nor will it generate functions that
2009         require access to such RTTI.   But when ML code generated from both
2010         incomplete and complete versions of the C type meet, the ML types
2011         are trivially interoperable.
2012    
2013         NOTE:  These changes restore the full generality of the translation
2014         (which was previously lost when I eliminated functorization)!
2015    
2016       B. Enum types:
2017    
2018       - Structure C now has a type constructor "enum" that is similar to
2019         how the "su" constructor works.  However, "enum" is not a phantom
2020         type because each "T enum" has values (and is isomorphic to
2021         MLRep.Signed.int).
2022       - There are generic access operations for enum objects (using
2023         MLRep.Signed.int).
2024       - ml-nlffigen will generate a structure E_foo for each "enum foo".
2025         * The structure contains the definition of type "mlrep" (the ML-side
2026         representation type of the enum).  Normally, mlrep is the same
2027         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
2028         then mlrep will be defined as a datatype -- thus facilitating
2029         pattern matching on mlrep values.
2030         ("-ec" will be suppressed if there are duplicate values in an
2031          enumeration.)
2032         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
2033         will be generated for each C enum constant xxx.
2034         * Conversion functions m2i and i2m convert between mlrep and
2035         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
2036         * Coversion functions c and ml convert between mlrep and "tag enum".
2037         * Access functions (get/set) fetch and store mlrep values.
2038       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
2039         enumerations are merged into one single enumeration represented by
2040         structure E_'.
2041    
2042    ----------------------------------------------------------------------
2043    Name: Allen Leung
2044    Date: 2002/02/25 04:45:00 EST
2045    Tag: leunga-20020225-cps-spill
2046    
2047    This is a new implementation of the CPS spill phase.
2048    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
2049    In case of problems, replace it with the old file spill.sml
2050    
2051    The current compiler runs into some serious performance problems when
2052    constructing a large record.  This can happen when we try to compile a
2053    structure with many items.  Even a very simple structure like the following
2054    makes the compiler slow down.
2055    
2056        structure Foo = struct
2057           val x_1 = 0w1 : Word32.int
2058           val x_2 = 0w2 : Word32.int
2059           val x_3 = 0w3 : Word32.int
2060           ...
2061           val x_N = 0wN : Word32.int
2062        end
2063    
2064    The following table shows the compile time, from N=1000 to N=4000,
2065    with the old compiler:
2066    
2067    N
2068    1000   CPS 100 spill                           0.04u  0.00s  0.00g
2069           MLRISC ra                               0.06u  0.00s  0.05g
2070              (spills = 0 reloads = 0)
2071           TOTAL                                   0.63u  0.07s  0.21g
2072    
2073    1100   CPS 100 spill                           8.25u  0.32s  0.64g
2074           MLRISC ra                               5.68u  0.59s  3.93g
2075              (spills = 0 reloads = 0)
2076           TOTAL                                   14.71u  0.99s  4.81g
2077    
2078    1500   CPS 100 spill                           58.55u  2.34s  1.74g
2079           MLRISC ra                               5.54u  0.65s  3.91g
2080              (spills = 543 reloads = 1082)
2081           TOTAL                                   65.40u  3.13s  6.00g
2082    
2083    2000   CPS 100 spill                           126.69u  4.84s  3.08g
2084           MLRISC ra                               0.80u  0.10s  0.55g
2085              (spills = 42 reloads = 84)
2086           TOTAL                                   129.42u  5.10s  4.13g
2087    
2088    3000   CPS 100 spill                           675.59u  19.03s  11.64g
2089           MLRISC ra                               2.69u  0.27s  1.38g
2090              (spills = 62 reloads = 124)
2091           TOTAL                                   682.48u  19.61s  13.99g
2092    
2093    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
2094           MLRISC ra                               4.96u  0.27s  2.72g
2095              (spills = 85 reloads = 170)
2096           TOTAL                                   2375.26u  57.21s  48.00g
2097    
2098    As you can see the old cps spill module suffers from some serious
2099    performance problem.  But since I cannot decipher the old code fully,
2100    instead of patching the problems up, I'm reimplementing it
2101    with a different algorithm.  The new code is more modular,
2102    smaller when compiled, and substantially faster
2103    (O(n log n) time and O(n) space).  Timing of the new spill module:
2104    
2105    4000  CPS 100 spill                           0.02u  0.00s  0.00g
2106          MLRISC ra                               0.25u  0.02s  0.15g
2107             (spills=1 reloads=3)
2108          TOTAL                                   7.74u  0.34s  1.62g
2109    
2110    Implementation details:
2111    
2112    As far as I can tell, the purpose of the CPS spill module is to make sure the
2113    number of live variables at any program point (the bandwidth)
2114    does not exceed a certain limit, which is determined by the
2115    size of the spill area.
2116    
2117    When the bandwidth is too large, we decrease the register pressure by
2118    packing live variables into spill records.  How we achieve this is
2119    completely different than what we did in the old code.
2120    
2121    First, there is something about the MLRiscGen code generator
2122    that we should be aware of:
2123    
2124    o MLRiscGen performs code motion!
2125    
2126       In particular, it will move floating point computations and
2127       address computations involving only the heap pointer to
2128       their use sites (if there is only a single use).
2129       What this means is that if we have a CPS record construction
2130       statement
2131    
2132           RECORD(k,vl,w,e)
2133    
2134       we should never count the new record address w as live if w
2135       has only one use (which is often the case).
2136    
2137       We should do something similar to floating point, but the transformation
2138       there is much more complex, so I won't deal with that.
2139    
2140    Secondly, there are now two new cps primops at our disposal:
2141    
2142     1. rawrecord of record_kind option
2143        This pure operator allocates some uninitialized storage from the heap.
2144        There are two forms:
2145    
2146         rawrecord NONE [INT n]  allocates a tagless record of length n
2147         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2148                                     and initializes the tag.
2149    
2150     2. rawupdate of cty
2151          rawupdate cty (v,i,x)
2152          Assigns to x to the ith component of record v.
2153          The storelist is not updated.
2154    
2155    We use these new primops for both spilling and increment record construction.
2156    
2157     1. Spilling.
2158    
2159        This is implemented with a linear scan algorithm (but generalized
2160        to trees).  The algorithm will create a single spill record at the
2161        beginning of the cps function and use rawupdate to spill to it,
2162        and SELECT or SELp to reload from it.  So both spills and reloads
2163        are fine-grain operations.  In contrast, in the old algorithm
2164        "spills" have to be bundled together in records.
2165    
2166        Ideally, we should sink the spill record construction to where
2167        it is needed.  We can even split the spill record into multiple ones
2168        at the places where they are needed.  But CPS is not a good
2169        representation for global code motion, so I'll keep it simple and
2170        am not attempting this.
2171    
2172     2. Incremental record construction (aka record splitting).
2173    
2174        Long records with many component values which are simulatenously live
2175        (recall that single use record addresses are not considered to
2176         be live) are constructed with rawrecord and rawupdate.
2177        We allocate space on the heap with rawrecord first, then gradually
2178        fill it in with rawupdate.  This is the technique suggested to me
2179        by Matthias.
2180    
2181        Some restrictions on when this is applicable:
2182        1. It is not a VECTOR record.  The code generator currently does not handle
2183           this case. VECTOR record uses double indirection like arrays.
2184        2. All the record component values are defined in the same "basic block"
2185           as the record constructor.  This is to prevent speculative
2186           record construction.
2187    
2188    ----------------------------------------------------------------------
2189    Name: Allen Leung
2190    Date: 2002/02/22 01:02:00 EST
2191    Tag: leunga-20020222-mlrisc-tools
2192    
2193    Minor bug fixes in the parser and rewriter
2194    
2195    ----------------------------------------------------------------------
2196    Name: Allen Leung
2197    Date: 2002/02/21 20:20:00 EST
2198    Tag: leunga-20020221-peephole
2199    
2200    Regenerated the peephole files.  Some contained typos in the specification
2201    and some didn't compile because of pretty printing bugs in the old version
2202    of 'nowhere'.
2203    
2204    ----------------------------------------------------------------------
2205    Name: Allen Leung
2206    Date: 2002/02/19 20:20:00 EST
2207    Tag: leunga-20020219-mlrisc-tools
2208    Description:
2209    
2210       Minor bug fixes to the mlrisc-tools library:
2211    
2212       1.  Fixed up parsing colon suffixed keywords
2213       2.  Added the ability to shut the error messages up
2214       3.  Reimplemented the pretty printer and fixed up/improved
2215           the pretty printing of handle and -> types.
2216       4.  Fixed up generation of literal symbols in the nowhere tool.
2217       5.  Added some SML keywords to to sml.sty
2218    
2219    ----------------------------------------------------------------------
2220    Name: Matthias Blume
2221    Date: 2002/02/19 16:20:00 EST
2222    Tag: blume-20020219-cmffi
2223    Description:
2224    
2225    A wild mix of changes, some minor, some major:
2226    
2227    * All C FFI-related libraries are now anchored under $c:
2228        $/c.cm      --> $c/c.cm
2229        $/c-int.cm  --> $c/internals/c-int.cm
2230        $/memory.cm --> $c/memory/memory.cm
2231    
2232    * "make" tool (in CM) now treats its argument pathname slightly
2233      differently:
2234        1. If the native expansion is an absolute name, then before invoking
2235           the "make" command on it, CM will apply OS.Path.mkRelative
2236           (with relativeTo = OS.FileSys.getDir()) to it.
2237        2. The argument will be passed through to subsequent phases of CM
2238           processing without "going native".  In particular, if the argument
2239           was an anchored path, then "make" will not lose track of that anchor.
2240    
2241    * Compiler backends now "know" their respective C calling conventions
2242      instead of having to be told about it by ml-nlffigen.  This relieves
2243      ml-nlffigen from one of its burdens.
2244    
2245    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
2246    
2247    * Export C_DEBUG and C_Debug from $c/c.cm.
2248    
2249    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
2250      subtyping relationship between incomplete pointers and their complete
2251      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
2252      with the convention of instantiating 'o with ('t, 'c) obj whenever
2253      the pointer target type is complete.  In the incomplete case, 'o
2254      will be instantiated with some "'c iobj" -- a type obtained by
2255      using one of the functors PointerToIncompleteType or PointerToCompleteType.
2256    
2257      Operations that work on both incomplete and complete pointer types are
2258      typed as taking an 'o ptr while operations that require the target to
2259      be known are typed as taking some ('t, 'c) obj ptr.
2260    
2261      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
2262      where void is an eqtype without any values.  This makes it possible
2263      to work on voidptr values using functions meant to operate on light
2264      incomplete pointers.
2265    
2266    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
2267      been vastly simplified.
2268    
2269    ----------------------------------------------------------------------
2270    Name: Matthias Blume
2271    Date: 2002/02/19 10:48:00 EST
2272    Tag: blume-20020219-pqfix
2273    Description:
2274    
2275    Applied Chris Okasaki's bug fix for priority queues.
2276    
2277    ----------------------------------------------------------------------
2278    Name: Matthias Blume
2279    Date: 2002/02/15 17:05:00
2280    Tag: Release_110_39
2281    Description:
2282    
2283    Last-minute retagging is becoming a tradition... :-(
2284    
2285    This is the working release 110.39.
2286    
2287    ----------------------------------------------------------------------
2288    Name: Matthias Blume
2289    Date: 2002/02/15 16:00:00 EST
2290    Tag: Release_110_39-orig
2291    Description:
2292    
2293    Working release 110.39.  New bootfiles.
2294    
2295    (Update: There was a small bug in the installer so it wouldn't work
2296    with all shells.  So I retagged. -Matthias)
2297    
2298    ----------------------------------------------------------------------
2299    Name: Matthias Blume
2300    Date: 2002/02/15 14:17:00 EST
2301    Tag: blume-20020215-showbindings
2302    Description:
2303    
2304    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
2305    the latter can be useful for exploring what bindings are available at
2306    the interactive prompt.  (The first function returns only the list
2307    of symbols that are really bound, the second prints those but also the
2308    ones that CM's autoloading mechanism knows about.)
2309    
2310    ----------------------------------------------------------------------
2311    Name: Matthias Blume
2312    Date: 2002/02/15 12:08:00 EST
2313    Tag: blume-20020215-iptrs
2314    Description:
2315    
2316    Two improvements to ml-nlffigen:
2317    
2318      1. Write files only if they do not exist or if their current contents
2319         do not coincide with what's being written.  (That is, avoid messing
2320         with the time stamps unless absolutely necessary.)
2321    
2322      2. Implement a "repository" mechanism for generated files related
2323         to "incomplete pointer types".   See the README file for details.
2324    
2325    ----------------------------------------------------------------------
2326    Name: Matthias Blume
2327    Date: 2002/02/14 11:50:00 EST
2328    Tag: blume-20020214-quote
2329    Description:
2330    
2331    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
2332    because of the new and improved tag generation scheme.  (Thanks to Allen
2333    Leung for pointing it out.)
2334    
2335    ----------------------------------------------------------------------
2336    Name: Lal George
2337    Date: 2002/02/14 09:55:27 EST 2002
2338    Tag: george-20020214-isabelle-bug
2339    Description:
2340    
2341    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2342    of Isabelle on the x86.
2343    
2344    From Allen:
2345    -----------
2346     I've found the problem:
2347    
2348         in ra-core.sml, I use the counter "blocked" to keep track of the
2349         true number of elements in the freeze queue.  When the counter goes
2350         to zero, I skip examining the queue.  But I've messed up the
2351         bookkeeping in combine():
2352    
2353             else ();
2354             case !ucol of
2355               PSEUDO => (if !cntv > 0 then
2356                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2357                                        ^^^^^^^^^^^^^^^^^^^^^^^
2358                      moveu := mergeMoveList(!movev, !moveu)
2359                     )
2360                  else ();
2361    
2362         combine() is called to coalesce two nodes u and v.
2363         I think I was thinking that if the move counts of u and v are both
2364         greater than zero then after they are coalesced then one node is
2365         removed from the freeze queue.  Apparently I was thinking that
2366         both u and v are of low degree, but that's clearly not necessarily true.
2367    
2368    
2369    02/12/2002:
2370        Here's the patch.  HOL now compiles.
2371    
2372        I don't know how this impact on performance (compile
2373        time or runtime).  This bug caused the RA (especially on the x86)
2374        to go thru the potential spill phase when there are still nodes on the
2375        freeze queue.
2376    
2377    
2378    
2379    
2380    ----------------------------------------------------------------------
2381    Name: Matthias Blume
2382    Date: 2002/02/13 22:40:00 EST
2383    Tag: blume-20020213-fptr-rtti
2384    Description:
2385    
2386    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2387    updates.
2388    
2389    ----------------------------------------------------------------------
2390    Name: Matthias Blume
2391    Date: 2002/02/13 16:41:00 EST
2392    Tag: blume-20020213-cmlpq
2393    Description:
2394    
2395    Added new priority queue export symbols (which have just been added to
2396    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2397    not compile and the installer would choke.)
2398    
2399    ----------------------------------------------------------------------
2400    Name: Matthias Blume
2401    Date: 2002/02/13 16:15:00 EST
2402    Tag: blume-20020213-various
2403    Description:
2404    
2405    1. More tweaks to ml-nlffigen:
2406    
2407       - better internal datastructures (resulting in slight speedup)
2408       - "-match" option requires exact match
2409       - "localized" gensym counters (untagged structs/unions nested within
2410         other structs/unions or within typedefs get a fresh counter; their
2411         tag will be prefixed by a concatenation of their parents' tags)
2412       - bug fixes (related to calculation of transitive closure of types
2413         to be included in the output)
2414    
2415    2. Minor Basis updates:
2416    
2417       - added implementations for List.collate and Option.app
2418    
2419    ----------------------------------------------------------------------
2420    Name: Matthias Blume
2421    Date: 2002/02/11 15:55:00 EST
2422    Tag: blume-20020211-gensym
2423    Description:
2424    
2425    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2426    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2427    names (ML structure names that correspond to unnamed C structs, unions,
2428    and enums), so that separate runs of ml-nlffigen do not clash.
2429    
2430    ----------------------------------------------------------------------
2431    Name: Matthias Blume
2432    Date: 2002/02/11 12:05:00 EST
2433    Tag: blume-20020211-gensml
2434    Description:
2435    
2436    A quick fix for a problem with GenSML (in the pgraph-util library):
2437    Make generation of toplevel "local" optional.  (Strictly speaking,
2438    signature definitions within "local" are not legal SML.)
2439    
2440    Other than that: updates to INSTALL and cm/TODO.
2441    
2442    ----------------------------------------------------------------------
2443    Name: Matthias Blume
2444    Date: 2002/02/08 15:00:00 EST
2445    Tag: blume-20020208-uniquepid
2446    Description:
2447    
2448    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2449    
2450    1. The installer (config/install.sh) has gotten smarter:
2451    
2452         - Configuration options are a bit easier to specify now
2453           (in config/targets).
2454         - Bug in recognizing .tar.bz2 files fixed.
2455         - Installer automatically resolves dependencies between
2456           configuration options (e.g., if you ask for eXene, you will
2457           also get cml -- regardless whether you asked for it or not).
2458         - Installer can run in "quieter mode" by setting the environment
2459           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2460           "completely silent", though.
2461         - Build HashCons library as part of smlnj-lib.
2462    
2463    2. A new scheme for assigning persistent identifiers to compilation
2464       units (and, by extension, to types etc.) has been put into place.
2465       This fixes a long-standing bug where types and even dynamic values
2466       can get internally confused, thereby compromising type safety
2467       (abstraction) and dynamic correctness.  See
2468    
2469         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2470    
2471       for an example of how things could go wrong until now.
2472    
2473       The downside of the new scheme is that pids are not quite as
2474       persistent as they used to be: CM will generate a fresh pid
2475       for every compilation unit that it thinks it sees for the first
2476       time.  That means that if you compile starting from a clean, fresh
2477       source tree at two different times, you end up with different
2478       binaries.
2479    
2480       Cutoff recompilation, however, has not been compromised because
2481       CM keeps pid information in special caches between runs.
2482    
2483    ----------------------------------------------------------------------
2484    Name: Lal George
2485    Date: 2002/02/07 15:34:13 EST 2002
2486    Tag: <none>
2487    Description:
2488    
2489    Compilers that generate assembly code may produce  global labels
2490    whose value is resolved at link time. The various peephole optimization
2491    modules did not take this in account.
2492    
2493    TODO. The Labels.addrOf function should really return an option
2494    type so that clients are forced to deal with this issue, rather
2495    than an exception being raised.
2496    
2497    ----------------------------------------------------------------------
2498    Name: Lal George
2499    Date: 2002/02/06 13:55:02 EST
2500    Tag: george-20020206-ra-breakup
2501    Description:
2502    
2503    1. A bug fix from Allen.
2504    
2505        A typo causes extra fstp %st(0)'s to be generated at compensation
2506        edges, which might cause stack underflow traps at runtime.  This
2507        occurs in fft where there are extraneous fstps right before the 'into'
2508        trap instruction (in this case they are harmless since none of the
2509        integers overflow.)
2510    
2511    2. Pulled out various utility modules that were embedded in the modules
2512       of the register allocator. I need these modules for other purposes, but
2513       they are not complete enough to put into a library (just yet).
2514    ----------------------------------------------------------------------
2515    Name: Matthias Blume
2516    Date: 2002/01/31 16:05:00 EST
2517    Tag: blume-20020131-sparc-ccalls
2518    Description:
2519    
2520    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2521       of extra stack space by mistake.  Fixed.
2522    
2523    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2524    
2525    ----------------------------------------------------------------------
2526    Name: Allen Leung
2527    Date: 2002/01/30
2528    Tag: leunga-20020130-nowhere-bug-fix
2529    Description:
2530    
2531       MLRISC bug fixes:
2532       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2533       2. MachineInt.fromString was negating its value.
2534    
2535    ----------------------------------------------------------------------
2536    Name: Matthias Blume
2537    Date: 2002/01/29
2538    Tag: blume-20020129-INSTALL
2539    Description:
2540    
2541    - Added somewhat detailed installation instructions (file INSTALL).
2542    - Fixed curl-detection bug in config/install.sh.
2543    - It is now possible to select the URL getter using the URLGETTER
2544      environment variable:
2545    
2546          not set / "unknown"      --> automatic detection (script tries wget,
2547                                       curl, and lynx)
2548          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2549                                       how to properly invoke them)
2550          other                    --> use $URLGETTER directly, it must take
2551                                       precisely two command-line arguments
2552                                       (source URL and destination file name)
2553    
2554    ----------------------------------------------------------------------
2555    Name: Matthias Blume
2556    Date: 2002/01/28
2557    Tag: blume-20020128-sparc-ccalls
2558    Description:
2559    
2560    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2561    - Make use of the allocParam argument in sparc-c-calls.
2562    
2563    ----------------------------------------------------------------------
2564    Name: Matthias Blume
2565    Date: 2002/01/28
2566    Tag: blume-20020128-allocParam
2567    Description:
2568    
2569    John Reppy:  Changes c-calls API to accept client-callback for
2570    allocating extra stack space.
2571    me: Corresponding changes to mlriscGen (using a dummy argument that
2572        does not change the current behavior).
2573    
2574    ----------------------------------------------------------------------
2575    Name: Matthias Blume
2576    Date: 2002/01/28 12:00:00
2577    Tag: Release_110_38
2578    Description:
2579    
2580    This time for real!!!
2581    
2582    ----------------------------------------------------------------------
2583    Name: Matthias Blume
2584    Date: 2002/01/28 10:56:00 EST
2585    Tag: blume-20020128-retraction
2586    Description:
2587    
2588    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
2589       with blume-Release_110_38-retracted.)
2590    
2591    1. Fixed a problem with incorrect rounding modes in real64.sml.
2592       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2593    
2594    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2595       and enums fixed.  The naming of corresponding ML identifiers should
2596       now be consistent again.
2597    
2598    ----------------------------------------------------------------------
2599    Name: Allen Leung
2600    Date: 2002/01/27
2601    Tag: leunga-20020127-nowhere
2602    Description:
2603    
2604       Added a target called nowhere in the configuration scripts.
2605       Enabling this will build the MLRISC 'nowhere' tool (for translating
2606       programs with where-clauses into legal SML code) during installation.
2607    
2608    ----------------------------------------------------------------------
2609    Name: Matthias Blume
2610    Date: 2002/01/25 21:27:00 EST
2611    Tag: blume-Release_110_38-retracted
2612    Description:
2613    
2614    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
2615    
2616    README will be added later.
2617    
2618    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
2619    (Release_110_38) removed.  Reason: Last-minute bug fixes.
2620    
2621    ----------------------------------------------------------------------
2622    Name: Matthias Blume
2623    Date: 2002/01/25
2624    Tag: blume-20020125-ffi
2625    Description:
2626    
2627    A large number of tweaks and improvements to ml-nlffi-lib and
2628    ml-nlffigen:
2629    
2630       - ML represenation types have been streamlined
2631       - getter and setter functions work with concrete values, not abstract
2632         ones where possible
2633       - ml-nlffigen command line more flexible (see README file there)
2634       - some bugs have been fixed (hopefully)
2635    
2636    ----------------------------------------------------------------------
2637    Name: Lal George
2638    Date: 2002/01/24
2639    Tag: george-20020124-risc-ra-interface
2640    Description:
2641    
2642       There is a dramatic simplification in the interface to the
2643       register allocator for RISC architectures as a result of making
2644       parallel copy instructions explicit.
2645    
2646    ----------------------------------------------------------------------
2647    Name: Matthias Blume
2648    Date: 2002/01/22
2649    Tag: blume-20020122-x86-ccalls
2650    Description:
2651    
2652    Bug fix for c-calls on x86 (having to do with how char- and
2653    short-arguments are being handled).
2654    
2655    ----------------------------------------------------------------------
2656    Name: Matthias Blume
2657    Date: 2002/01/21
2658    Tag: blume-20020121-ff
2659    Description:
2660    
2661    Another day of fiddling with the FFI...
2662    
2663    1. Bug fix/workaround:  CKIT does not complain about negative array
2664       dimensions, so ml-nlffigen has to guard itself against this possibility.
2665       (Otherwise a negative dimension would send it into an infinite loop.)
2666    
2667    2. Some of the abstract types (light objects, light pointers, most "base"
2668       types) in structure C are now eqtypes.
2669    
2670    3. Added constructors and test functions for NULL function pointers.
2671    
2672    ----------------------------------------------------------------------
2673    Name: Matthias Blume
2674    Date: 2002/01/18
2675    Tag: blume-20020118-ready-for-new-release
2676    Description:
2677    
2678    Made config/srcarchiveurl point to a new place.  (Will provide boot
2679    files shortly.)
2680    
2681    Maybe we christen this to be 110.38?
2682    
2683    ----------------------------------------------------------------------
2684    Name: Matthias Blume
2685    Date: 2002/01/18
2686    Tag: blume-20020118-more-ffifiddle
2687    Description:
2688    
2689    Today's FFI fiddling:
2690    
2691      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
2692        C.Set.* that use concrete (MLRep.*) arguments and results instead
2693        of abstract ones.
2694    
2695      - Provided word-style bit operations etc. for "int" representation
2696        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
2697        and Long.
2698    
2699    ----------------------------------------------------------------------
2700    Name: Matthias Blume
2701    Date: 2002/01/18
2702    Tag: blume-20020118-use-x86-fp
2703    Description:
2704    
2705    Now that x86-fast-fp seems to be working, I turned it back on again
2706    by default.  (Seems to work fine now, even with the FFI.)
2707    
2708    Other than that, I added some documentation about the FFI to
2709    src/ml-nlffigen/README and updated the FFI test examples in
2710    src/ml-nlffi-lib/Tests/*.
2711    
2712    ----------------------------------------------------------------------
2713    Name: Allen Leung
2714    Date: 2002/01/17
2715    Tag: leunga-20020117-x86-fast-fp-call
2716    Description:
2717    
2718       1. Fixed a problem with handling return fp values when x86's fast fp
2719          mode is turned on.
2720    
2721       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
2722          of %f32.
2723    
2724       3. Added a constructor INT32lit to the ast of MLRISC tools.
2725    
2726    ----------------------------------------------------------------------
2727    Name: Matthias Blume
2728    Date: 2002/01/16
2729    Tag: blume-20020116-ffifiddle
2730    Description:
2731    
2732    More fiddling with the FFI interface:
2733    
2734     - Make constness 'c instead of rw wherever possible.  This eliminates
2735       the need for certain explicit coercions.  (However, due to ML's
2736       value polymorphism, there will still be many cases where explicit
2737       coercions are necessary.  Phantom types are not the whole answer
2738       to modeling a subtyping relationship in ML.)
2739    
2740     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
2741    
2742     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
2743       clumsy workaround for varargs problem.)
2744    
2745    ----------------------------------------------------------------------
2746    Name: Lal George
2747    Date: 2002/01/15
2748    Tag: <none>
2749    Description:
2750    
2751    1. Since COPY instructions are no longer native to the architecture,
2752       a generic functor can be used to implement the expandCopies function.
2753    
2754    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
2755       TEXT segment.
2756    
2757    ----------------------------------------------------------------------
2758    Name: Matthias Blume
2759    Date: 2002/01/15
2760    Tag: blume-20020115-ffiupdates
2761    Description:
2762    
2763    1. Fix for bug resulting in single-precision float values being returned
2764       incorrectly from FFI calls.
2765    
2766    2. Small modifications to C FFI API:
2767    
2768        - memory-allocation routines return straight objects (no options)
2769          and raise an exception in out-of-memory situations
2770        - unsafe extensions to cast between function pointers and pointers
2771          from/to ints
2772        - added structure C_Debug as an alternative to structure C where
2773          pointer-dereferencing (|*| and |*!) always check for null-pointers
2774        - added open_lib' to DynLinkage;  open_lib' works like open_lib
2775          but also takes a (possibly empty) list of existing library handles
2776          that the current library depends on
2777    
2778    ----------------------------------------------------------------------
2779    Name: Matthias Blume
2780    Date: 2002/01/10
2781    Tag: blume-20020110-newffigen
2782    Description:
2783    
2784    1. Updates to portable graph code.
2785    
2786    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
2787       more scalable now so that even huge interfaces such as the one
2788       for GTK compile in finite time and space. :-)
2789       See src/ml-nlffigen/README for details on what's new.
2790    
2791    ----------------------------------------------------------------------
2792    Name: Lal George
2793    Date: 2001/01/09 14:31:35 EST 2002
2794    Tag: george-20011206-rm-native-copy
2795    Description:
2796    
2797            Removed the native COPY and FCOPY instructions
2798            from all the architectures and replaced it with the
2799            explicit COPY instruction from the previous commit.
2800    
2801            It is now possible to simplify many of the optimizations
2802            modules that manipulate copies. This has not been
2803            done in this change.
2804    
2805    ----------------------------------------------------------------------
2806    Name: Lal George
2807    Date: 2001/12/06 16:50:13 EST 2001
2808    Tag: george-20011206-mlrisc-instruction
2809    Description:
2810    
2811    Changed the representation of instructions from being fully abstract
2812    to being partially concrete. That is to say:
2813    
2814      from
2815            type instruction
2816    
2817      to
2818            type instr                              (* machine instruction *)
2819    
2820            datatype instruction =
2821                LIVE of {regs: C.cellset, spilled: C.cellset}
2822              | KILL of {regs: C.cellset, spilled: C.cellset}
2823              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
2824              | ANNOTATION of {i: instruction, a: Annotations.annotation}
2825              | INSTR of instr
2826    
2827    This makes the handling of certain special instructions that appear on
2828    all architectures easier and uniform.
2829    
2830    LIVE and KILL say that a list of registers are live or killed at the
2831    program point where they appear. No spill code is generated when an
2832    element of the 'regs' field is spilled, but the register is moved to
2833    the 'spilled' (which is present, more for debugging than anything else).
2834    
2835    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
2836    We used to generate:
2837    
2838            DEFFREG f1
2839            f1 := f2 + f3
2840            trapb
2841    
2842    but now generate:
2843    
2844            f1 := f2 + f3
2845            trapb
2846            LIVE {regs=[f1,f2,f3], spilled=[]}
2847    
2848    Furthermore, the DEFFREG (hack) required that all floating point instruction
2849    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
2850    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
2851    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
2852    this architecture).
2853    
2854    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
2855    all the architectures. This will result in further simplification of the
2856    register allocator that must be aware of them for coalescing purposes, and
2857    will also simplify certain aspects of the machine description that provides
2858    callbacks related to parallel copies.
2859    
2860    ANNOTATION should be obvious, and now INSTR represents the honest to God
2861    machine instruction set!
2862    
2863    The <arch>/instructions/<arch>Instr.sml files define certain utility
2864    functions for making porting easier -- essentially converting upper case
2865    to lower case. All machine instructions (of type instr) are in upper case,
2866    and the lower case form generates an MLRISC instruction. For example on
2867    the alpha we have:
2868    
2869      datatype instr =
2870         LDA of {r:cell, b:cell, d:operand}
2871       | ...
2872    
2873      val lda : {r:cell, b:cell, d:operand} -> instruction
2874        ...
2875    
2876    where lda is just (INSTR o LDA), etc.
2877    
2878    ----------------------------------------------------------------------
2879    Name: Matthias Blume
2880    Date: 2001/11/22 21:40:00 EST
2881    Tag: Release_110_37
2882    Description:
2883    
2884    Release 110.37.  This time for real.
2885    
2886    ----------------------------------------------------------------------
2887    Name: Matthias Blume
2888    Date: 2001/11/21 16:35:00 EST
2889    Tag: blume-20011121-foot-in-mouth
2890    Description:
2891    
2892    Removed the "Release_110_37" tag because of a serious bug.
2893    This will be re-tagged once the bug is fixed.
2894    
2895    ----------------------------------------------------------------------
2896    Name: Matthias Blume
2897    Date: 2001/11/21 16:14:00 EST
2898    Tag: blume-20011121-forgottenfile
2899    Description:
2900    
2901    Forgot to add a file.  (Just a .tex-file -- part of
2902    the CM manual source.)
2903    
2904    ----------------------------------------------------------------------
2905    Name: Matthias Blume
2906    Date: 2001/11/21 16:10:00 EST
2907    Tag: blume-20011121-invalid_110_37
2908    Description:
2909    
2910    Note: I removed the original tag "Release_110_37" from this commit
2911          because we found a serious bug in all non-x86 backends.
2912          - Matthias
2913    
2914    1. Modifications to the SML/NJ code generator and to the runtime system
2915       so that code object name strings are directly inserted into code
2916       objects at code generation time.  The only business the runtime system
2917       has with this is now to read the name strings on occasions.
2918       (The encoding of the name string has also changed somewhat.)
2919    
2920    2. CM now implements a simple "set calculus" for specifying export lists.
2921       In particular, it is now possible to refer to the export lists of
2922       other libraries/groups/sources and form unions as well as differences.
2923       See the latest CM manual for details.
2924    
2925    3. An separate notion of "proxy" libraries has again be eliminated from
2926       CM's model.  (Proxy libraries are now simply a special case of using
2927       the export list calculus.)
2928    
2929    4. Some of the existing libraries now take advantage of the new set
2930       calculus.
2931       (Notice that not all libraries have been converted because some
2932       of the existing .cm-files are supposed to be backward compatible
2933       with 110.0.x.)
2934    
2935    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
2936       "General.exnMessage"!  The former relies on a certain hook to be
2937       initialized, and that often does not happen in the stand-alone case.)
2938    
2939    ----------------------------------------------------------------------
2940    Name: Lal George
2941    Date: 2001/11/21  13:56:18 EST
2942    Tag: george-2001121-pseudo-ops
2943    Description:
2944    
2945      Implemented a complete redesign of MLRISC pseudo-ops. Now there
2946      ought to never be any question of incompatabilities with
2947      pseudo-op syntax expected by host assemblers.
2948    
2949      For now, only modules supporting GAS syntax are implemented
2950      but more should follow, such as MASM, and vendor assembler
2951      syntax, e.g. IBM as, Sun as, etc.
2952    
2953    ----------------------------------------------------------------------
2954    Name: Matthias Blume
2955    Date: 2001/11/14 11:52:00 EST
2956    Tag: blume-20011114-srcname
2957    Description:
2958    
2959    1. Routed the name of the current source file to mlriscgen where it
2960       should be directly emitted into the code object.  (This last part
2961       is yet to be done.)
2962    
2963    2. Some cleanup of the pgraph code to make it match the proposal that
2964       I put out the other day.  (The proposal notwithstanding, things are
2965       still in flux here.)
2966    
2967    ----------------------------------------------------------------------
2968    Name: Lal George
2969    Date: 2001/11/14 09:44:04 EST
2970    Tag:
2971    Description:
2972    
2973      Fix for a backpatching bug reported by Allen.
2974    
2975      Because the boundary between short and long span-dependent
2976      instructions is +/- 128, there are an astounding number of
2977      span-dependent instructions whose size is over estimated.
2978    
2979      Allen came up with the idea of letting the size of span
2980      dependent instructions be non-monotonic, for a maxIter
2981      number of times, after which the size must be monotonically
2982      increasing.
2983    
2984      This table shows the number of span-dependent instructions
2985      whose size was over-estimated as a function of maxIter, for the
2986      file Parse/parse/ml.grm.sml:
2987    
2988         maxIter            # of instructions:
2989            10                      687
2990            20                      438
2991            30                      198
2992            40                        0
2993    
2994      In compiling the compiler, there is no significant difference in
2995      compilation speed between maxIter=10 and maxIter=40. Actually,
2996      my measurements showed that maxIter=40 was a tad faster than
2997      maxIter=10! Also 96% of the  files in the compiler reach a fix
2998      point within 13 iterations, so fixing maxIter at 40, while high,
2999      is okay.
3000    
3001    ----------------------------------------------------------------------
3002    Name: Matthias Blume
3003    Date: 2001/10/31 15:25:00 EST
3004    Tag: blume-20011031-pgraph
3005    Description:
3006    
3007    CKIT:
3008    * Changed the "Function" constructor of type Ast.ctype to carry optional
3009      argument identifiers.
3010    * Changed the return type of TypeUtil.getFunction accordingly.
3011    * Type equality ignores the argument names.
3012    * TypeUtil.composite tries to preserve argument names but gives up quickly
3013      if there is a mismatch.
3014    
3015    installation script:
3016    * attempts to use "curl" if available (unless "wget" is available as well)
3017    
3018    CM:
3019    * has an experimental implementation of "portable graphs" which I will
3020      soon propose as an implementation-independent library format
3021    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
3022    
3023    NLFFI-LIB:
3024    * some cleanup (all cosmetic)
3025    
3026    NLFFIGEN:
3027    * temporarily disabled the mechanism that suppresses ML output for
3028      C definitions whose identifiers start with an underscore character
3029    * generate val bindings for enum constants
3030    * user can request that only one style (light or heavy) is being used;
3031      default is to use both (command-line arguments: -heavy and -light)
3032    * fixed bug in handling of function types involving incomplete pointers
3033    * generate ML entry points that take record arguments (i.e., using
3034      named arguments) for C functions that have a prototype with named
3035      arguments
3036      (see changes to CKIT)
3037    
3038    ----------------------------------------------------------------------
3039    Name: Allen Leung
3040    Date: 2001/10/27 20:34:00 EDT
3041    Tag: leunga-20011027-x86-fast-fp-call
3042    Description:
3043    
3044       Fixed the bug described in blume-20010920-slowfp.
3045    
3046       The fix involves
3047          1. generating FCOPYs in FSTP in ia32-svid
3048          2. marking a CALL with the appropriate annotation
3049    
3050    ----------------------------------------------------------------------
3051    Name: Matthias Blume
3052    Date: 2001/10/16 11:32:00 EDT
3053    Tag: blume-20011016-netbsd
3054    Description:
3055    
3056    Underscore patch from Chris Richards (fixing problem with compiling
3057    runtime system under recent NetBSD).
3058    
3059    ----------------------------------------------------------------------
3060    Name: Allen Leung
3061    Date: 2001/10/12 17:18:32 EDT 2001
3062    Tag: leung-20011012-x86-printflowgraph
3063    Description:
3064    
3065    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
3066    
3067    ----------------------------------------------------------------------
3068    Name: Lal George
3069    Date: 2001/10/11 23:51:34 EDT
3070    Tag: george-20011011-too-many-instrs
3071    Description:
3072    
3073    The representation of a program point never expected to see more
3074    than 65536 instructions in a basic block!
3075    
3076    ----------------------------------------------------------------------
3077    Name: Lal George
3078    Date: 2001/10/09 09:41:37 EDT
3079    Tag: george-20011008-mlrisc-labels
3080    Description:
3081    
3082    Changed the machine description files to support printing of
3083    local and global labels in assembly code, based on host assembler
3084    conventions.
3085    
3086    ----------------------------------------------------------------------
3087    Name: Matthias Blume
3088    Date: 2001/09/25 15:25:00 EDT
3089    Tag: blume-20010925-exninfo
3090    Description:
3091    
3092    I provided a non-hook implementation of exnName (at the toplevel) and
3093    made the "dummy" implementation of exnMessage (at the toplevel) more
3094    useful: if nothing gets "hooked in", then at least you are going to
3095    see the exception name and a message indicating why you don't see more.
3096    
3097    [For the time being, programs that need exnMessage and want to use
3098    ml-build should either use General.exnMessage (strongly recommended) or
3099    refer to structure General at some other point so that CM sees a
3100    static dependency.]
3101    
3102    [Similar remarks go for "print" and "use":  If you want to use their
3103    functionality in stand-alone programs generated by ml-build, then use
3104    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
3105    
3106    ----------------------------------------------------------------------
3107    Name: Matthias Blume
3108    Date: 2001/09/20 17:28:00 EDT
3109    Tag: blume-20010920-slowfp
3110    Description:
3111    
3112    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
3113    
3114    ----------------------------------------------------------------------
3115    Name: Matthias Blume
3116    Date: 2001/09/20 17:20:00 EDT
3117    Tag: blume-20010920-canonicalpaths
3118    Description:
3119    
3120    0. Updated the BOOT file (something that I forgot to do earlier).
3121    
3122    1. Small internal change to CM so that it avoids "/../" in filenames
3123       as much as possible (but only where it is safe).
3124    
3125    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
3126       that arguments that contain delimiters are passed through correctly.
3127       This change also means that all "special" arguments of the form
3128       @SMLxxx... must come first.
3129    
3130    3. Changed install script to put relative anchor names for tool commands
3131       into pathconfig.
3132    
3133    ----------------------------------------------------------------------
3134    Name: Matthias Blume
3135    Date: 2001/09/18 15:35:00 EDT
3136    Tag: blume-20010918-readme11036
3137    Description:
3138    
3139    Added README files.
3140    
3141    ----------------------------------------------------------------------
3142    Name: Matthias Blume
3143    Date: 2001/09/18 11:45:00 EDT
3144    Tag: Release_110_36 (retag)
3145    Description:
3146    
3147    Fixed mistake in config/preloads. Retagged as 110.36.
3148    
3149    ----------------------------------------------------------------------
3150    Name: Matthias Blume
3151    Date: 2001/09/18 09:40:00 EDT
3152    Tag: Release_110_36_orig (tag changed)
3153    Description:
3154    
3155    New version (110.36).  New bootfiles.
3156    
3157    ----------------------------------------------------------------------
3158    Name: Matthias Blume
3159    Date: 2001/09/14 16:15:00 EDT
3160    Tag: blume-20010914-x86fastfp
3161    Description:
3162    
3163    John committed some changes that Allen made, in particular a (hopefully)
3164    correctly working version of the x86-fp module.
3165    
3166    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
3167    flag to "true".  Everything seems to compile to a fixpoint ok, and
3168    "mandelbrot" speeds up by about 15%.
3169    
3170    ----------------------------------------------------------------------
3171    Name: Matthias Blume
3172    Date: 2001/09/13 11:20:00 EDT
3173    Tag: blume-20010913-minimal
3174    Description:
3175    
3176    1. Stefan Monnier's patch to fix a miscompilation problem that
3177       was brought to light by John Reppy's work on Moby.
3178    
3179    2. Implemented a minimal "structure Compiler" that contains just
3180       "version" and "architecture".  The minimal version will be
3181       available when the full version is not.  This is for backward-
3182       compatibility with code that wants to test Compiler.version.
3183    
3184    ----------------------------------------------------------------------
3185    Name: Matthias Blume
3186    Date: 2001/08/28 14:03:00 EDT
3187    Tag: blume-20010828-ml-lex
3188    Description:
3189    
3190    Fix for bug 1581, received from Neophytos Michael.
3191    
3192    ----------------------------------------------------------------------
3193    Name: Matthias Blume
3194    Date: 2001/08/27 11:20:00 EDT
3195    Tag: blume-20010827-readme11035
3196    Description:
3197    
3198    Fleshed out the README file for 110.35.
3199    
3200    ----------------------------------------------------------------------
3201    Name: Matthias Blume
3202    Date: 2001/08/24 17:10:00 EDT
3203    Tag: Release_110_35
3204    Description:
3205    
3206    New version number (110.35).  New bootfiles.
3207    
3208    ----------------------------------------------------------------------
3209    Name: Lal George
3210    Date: 2001/08/24 13:47:18 EDT 2001
3211    Tag: george-20010824-MLRISC-graphs
3212    Description:
3213    
3214     removed clusters from MLRISC completely and replaced with graphs.
3215    
3216    ----------------------------------------------------------------------
3217    Name: Matthias Blume
3218    Date: 2001/08/23 17:50:00 EDT
3219    Tag: blume-20010823-toplevel
3220    Description:
3221    
3222    - some reorganization of the code that implements various kinds of
3223      environments in the compiler (static, dynamic, symbolic, combined)
3224    - re-implemented the EnvRef module so that evalStream works properly
3225      (if the stream contains references to "use", "CM.make", etc.)
3226    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
3227    
3228    ----------------------------------------------------------------------
3229    Name: Matthias Blume
3230    Date: 2001/08/20 15:50 EDT
3231    Tag: blume20010820-slipup
3232    Description:
3233    
3234    I forgot to commit a few files.  Here they are...
3235    
3236    ----------------------------------------------------------------------
3237    Name: Matthias Blume
3238    Date: 2001/08/20 15:35:00 EDT
3239    Tag: blume-20010820-debugprof
3240    Description:
3241    
3242    !!!! NEW BOOTFILES !!!!
3243    
3244    This is another round of reorganizing the compiler sources.  This
3245    time the main goal was to factor out all the "instrumentation"
3246    passes (for profiling and backtracing) into their own library.
3247    The difficulty was to do it in such a way that it does not depend
3248    on elaborate.cm but only on elabdata.cm.
3249    
3250    Therefore there have been further changes to both elaborate.cm and
3251    elabdata.cm -- more "generic" things have been moved from the former
3252    to the latter.  As a result, I was forced to split the assignment
3253    of numbers indicating "primtyc"s into two portions: SML-generic and
3254    SML/NJ-specific.  Since it would have been awkward to maintain,
3255    I bit the bullet and actually _changed_ the mapping between these
3256    numbers and primtycs.  The bottom line of this is that you need
3257    a new set of bin- and bootfiles.
3258    
3259    I have built new bootfiles for all architectures, so doing a fresh
3260    checkout and config/install.sh should be all you need.
3261    
3262    The newly created library's name is
3263    
3264        $smlnj/viscomp/debugprof.cm
3265    
3266    and its sources live under
3267    
3268        src/compiler/DebugProf
3269    
3270    ----------------------------------------------------------------------
3271    Name: Matthias Blume
3272    Date: 2001/08/15 17:15:00 EDT
3273    Tag: blume-20010815-compreorg
3274    Description:
3275    
3276    This is a first cut at reorganizing the CM libraries that make up the
3277    core of the compiler.  The idea is to separate out pieces that could
3278    be used independently by tools, e.g., the parser, the typechecker, etc.
3279    
3280    The current status is a step in this direction, but it is not quite
3281    satisfactory yet.  Expect more changes in the future.
3282    
3283    Here is the current (new) organization...
3284    
3285        What used to be $smlnj/viscomp/core.cm is now divided into
3286        six CM libraries:
3287    
3288             $smlnj/viscomp/basics.cm
3289                           /parser.cm
3290                           /elabdata.cm
3291                           /elaborate.cm
3292                           /execute.cm
3293                           /core.cm
3294    
3295        The CM files for these libraries live under src/system/smlnj/viscomp.
3296        All these libraries are proxy libraries that contain precisely
3297        one CM library component.  Here are the locations of the components
3298        (all within the src/compiler tree):
3299    
3300             Basics/basics.cm
3301             Parse/parser.cm
3302             ElabData/elabdata.cm
3303             Elaborator/elaborate.cm
3304             Execution/execute.cm
3305             core.cm
3306    
3307         [This organization is the same that has been used already
3308         for a while for the architecture-specific parts of the visible
3309         compiler and for the old version of core.cm.]
3310    
3311         As you will notice, many source files have been moved from their
3312         respective original locations to a new home in one of the above
3313         subtrees.
3314    
3315         The division of labor between the new libraries is the following:
3316    
3317             basics.cm:
3318                - Simple, basic definitions that pertain to many (or all) of
3319                  the other libraries.
3320             parser.cm:
3321                - The SML parser, producing output of type Ast.dec.
3322                - The type family for Ast is also defined and exported here.
3323             elabdata.cm:
3324                - The datatypes that describe input and output of the elaborator.
3325                  This includes types, absyn, and static environments.
3326             elaborator.cm:
3327                - The SML/NJ type checker and elaborator.
3328                  This maps an Ast.dec (with a given static environment) to
3329                  an Absyn.dec (with a new static environment).
3330                - This libraries implements certain modules that used to be
3331                  structures as functors (to remove dependencies on FLINT).
3332             execute.cm:
3333                - Everything having to do with executing binary code objects.
3334                - Dynamic environments.
3335             core.cm:
3336                - SML/NJ-specific instantiations of the elaborator and MLRISC.
3337                - Top-level modules.
3338                - FLINT (this should eventually become its own library)
3339    
3340    Notes:
3341    
3342    I am not 100% happy with the way I separated the elaborator (and its
3343    data structures) from FLINT.  Two instances of the same problem:
3344    
3345        1. Data structures contain certain fields that carry FLINT-specific
3346           information.  I hacked around this using exn and the property list
3347           module from smlnj-lib.  But the fact that there are middle-end
3348           specific fields around at all is a bit annoying.
3349    
3350        2. The elaborator calculates certain FLINT-related information.  I tried
3351           to make this as abstract as I could using functorization, but, again,
3352           the fact that the elaborator has to perform calculations on behalf
3353           of the middle-end at all is not nice.
3354    
3355        3. Having to used exn and property lists is unfortunate because it
3356           weakens type checking.  The other alternative (parameterizing
3357           nearly *everything*) is not appealing, though.
3358    
3359    I removed the "rebinding =" warning hack because due to the new organization
3360    it was awkward to maintain it.  As a result, the compiler now issues some of
3361    these warnings when compiling init.cmi during bootstrap compilation. On
3362    the plus side, you also get a warning when you do, for example:
3363       val op = = Int32.+
3364    which was not the case up to now.
3365    
3366    I placed "assign" and "deref" into the _Core structure so that the
3367    code that deals with the "lazy" keyword can find them there.  This
3368    removes the need for having access to the primitive environment
3369    during elaboration.
3370    
3371    ----------------------------------------------------------------------
3372    Name: Matthias Blume
3373    Date: 2001/08/13
3374    Tag: blume-20010813-closures
3375    Description:
3376    
3377    This fix was sent to us by Zhong Shao.  It is supposed to improve the
3378    performance of certain loops by avoiding needless closure allocation.
3379    
3380    ----------------------------------------------------------------------
3381    Name: Lal George
3382    Date: 2001/07/31 10:03:23 EDT 2001
3383    Tag: george-20010731-x86-fmalloc
3384    Description: Fixed bug in x86 calls
3385    
3386        There was a bug where call instructions would mysteriously
3387        vanish. The call instruction had to be one that returned
3388        a floating point value.
3389    
3390    ----------------------------------------------------------------------
3391    Name: Lal George
3392    Date: 2001/07/19 16:36:29 EDT 2001
3393    Tag: george-20010719-simple-cells
3394    Description:
3395    
3396    I have dramatically simplified the interface for CELLS in MLRISC.
3397    
3398    In summary, the cells interface is broken up into three parts:
3399    
3400      1. CellsBasis : CELLS_BASIS
3401    
3402            CellsBasis is a top level structure and common for all
3403            architectures.  it contains the definitions of basic datatypes
3404            and utility  functions over these types.
3405    
3406      2. functor Cells() : CELLS
3407    
3408            Cells generates an interface for CELLS that incorporates the
3409            specific resources on the target architecture, such as the
3410            presence of special register classes, their number and size,
3411            and various useful substructures.
3412    
3413      3. <ARCH>CELLS
3414    
3415            e.g. SparcCells: SPARCCELLS
3416    
3417            <ARCH>CELLS usually contains additional bindings for special
3418            registers  on the architecture, such as:
3419    
3420                    val r0 : cell           (* register zero *)
3421                    val y : cell            (* Y register *)
3422                    val psr : cell          (* processor status register *)
3423                    ...
3424    
3425            The structure returned by applying the Cells functor is opened
3426            in this interface.
3427    
3428    The main implication of all this is that the datatypes for cells is
3429    split between CellsBasis and CELLS -- a fairly simple change for user
3430    code.
3431    
3432    In the old scheme the CELLS interface had a definitional binding of
3433    the form:
3434    
3435            signature CELLS = sig
3436    
3437               structure CellsBasis = CellsBasis
3438    
3439               ...
3440    
3441            end
3442    
3443    With all the sharing constraints that goes on in MLRISC, this old
3444    design  quickly leads to errors such as:
3445    
3446            "structure definition spec inside of sharing ... "
3447    
3448    
3449    and appears to require an unacceptable amount of sharing and where
3450    constraint hackery.
3451    
3452    I think this error message (the interaction of definitional specs and
3453    sharing) requires more explanation on our web page.
3454    
3455    ----------------------------------------------------------------------
3456    Name: Matthias Blume
3457    Date: 2001/07/19 15:00:00 EDT
3458    Tag: blume-20010719-libreorg
3459    Description:
3460    
3461    This update puts together a fairly extensive but straightforward change
3462    to the way the libraries that implement the interactive system are
3463    organized:
3464    
3465       The biggest change is the elimination of structure Compiler.  As a
3466       replacement for this structure, there is now a CM library
3467       (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
3468       that exports all the substructures of the original structure Compiler
3469       directly.  So instead of saying Compiler.Foo.bar one now simply
3470       says Foo.bar.  (The CM libraries actually export a collection of
3471       structures that is richer than the collection of substructures of
3472       structure Compiler.)
3473    
3474       To make the transition smooth, there is a separate library called
3475       $smlnj/compiler/compiler.cm which puts together and exports the
3476       original structure Compiler (or at least something very close to it).
3477    
3478       There are five members of the original structure Compiler
3479       that are not exported directly but which instead became members
3480       of a new structure Backend (described by signature BACKEND).  These are:
3481       structure Profile (: PROFILE), structure Compile (: COMPILE), structure
3482       Interact (: INTERACT), structure Machine (: MACHINE), and val
3483       architecture (: string).
3484    
3485       Structure Compiler.Version has become structure CompilerVersion.
3486    
3487       Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
3488       by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
3489       or x86, respectively.
3490       Each of these exports the same frontend structures that
3491       $smlnj/compiler.cm exports.  But they do not have a structure Backend
3492       and instead export some structure <Arch>Backend where <Arch> is Alpha32,
3493       Hppa, PPC, Sparc, or X86, respectively.
3494    
3495       Library $smlnj/compiler/all.cm exports the union of the exports of
3496       $smlnj/compiler/<arch>.cm
3497    
3498       There are no structures <Arch>Compiler anymore, use
3499       $smlnj/compiler/<arch>.cm instead.
3500    
3501       Library host-compiler-0.cm is gone.  Instead, the internal library
3502       that instantiates CM is now called cm0.cm.  Selection of the host
3503       compiler (backend) is no longer done here but. (Responsibility for it
3504       now lies with $smlnj/compiler/current.cm.  This seems to be more
3505       logical.)
3506    
3507       Many individual files have been moved or renamed.  Some files have
3508       been split into multiple files, and some "dead" files have been deleted.
3509    
3510    Aside from these changes to library organization, there are also changes
3511    to the way the code itself is organized:
3512    
3513       Structure Binfile has been re-implemented in such a way that it no
3514       longer needs any knowledge of the compiler.  It exclusively deals
3515       with the details of binfile layout.  It no longer invokes the
3516       compiler (for the purpose of creating new prospective binfile
3517       content), and it no longer has any knowledge of how to interpret
3518       pickles.
3519    
3520       Structure Compile (: COMPILE) has been stripped down to the bare
3521       essentials of compilation.  It no longer deals with linking/execution.
3522       The interface has been cleaned up considerably.
3523    
3524       Utility routines for dealing with linking and execution have been
3525       moved into their own substructures.
3526    
3527       (The ultimate goal of these changes is to provide a light-weight
3528       binfile loader/linker (at least for, e.g., stable libraries) that
3529       does not require CM or the compiler to be present.)
3530    
3531    CM documentation has been updated to reflect the changes to library
3532    organization.
3533    
3534    ----------------------------------------------------------------------
3535    Name: Matthias Blume
3536    Date: 2001/07/10 17:30:00 EDT
3537    Tag: Release_110_34
3538    Description:
3539    
3540    Minor tweak to 110.34 (re-tagged):
3541    
3542      - README.html file added to CVS repository
3543      - runtime compiles properly under FreeBSD 3.X and 4.X
3544    
3545    ----------------------------------------------------------------------
3546    Name: Matthias Blume
3547    Date: 2001/07/10 17:30:00 EDT
3548    Tag: Release_110_34
3549    Description:
3550    
3551    New version number (110.34). New bootfiles.
3552    
3553    ----------------------------------------------------------------------
3554    Name: Matthias Blume
3555    Date: 2001/07/09 16:00:00 EDT
3556    Tag: blume-20010709-more-varargs
3557    Description:
3558    
3559    I changed the handling of varargs in ml-nlffigen again:
3560    The ellipsis ... will now simply be ignored (with an accompanying warning).
3561    
3562    The immediate effect is that you can actually call a varargs function
3563    from ML -- but you can't actually supply any arguments beyond the ones
3564    specified explicitly.  (For example, you can call printf with its format
3565    string, but you cannot pass additional arguments.)
3566    
3567    This behavior is only marginally more useful than the one before, but
3568    it has the advantage that a function or, more importantly, a function
3569    type never gets dropped on the floor, thus avoiding follow-up problems with
3570    other types that refer to the offending one.
3571    
3572    ----------------------------------------------------------------------
3573    Name: Matthias Blume
3574    Date: 2001/07/09 11:25:00 EDT
3575    Tag: blume-20010709-varargs
3576    Description:
3577    
3578    1. ckit-lib.cm now exports structure Error
3579    2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
3580       with a warning accompanied by a source location.  Moreover, it
3581       merely skips the offending function or type and proceeds with the
3582       rest of its work.u  As a result, one can safely feed C code containing
3583       "..." to ml-nlffigen.
3584    3. There are some internal improvements to CM, providing slightly
3585       more general string substitutions in the tools subsystem.
3586    
3587    ----------------------------------------------------------------------
3588    Name: Matthias Blume
3589    Date: 2001/06/27 15:10:00 EDT
3590    Tag: blume-20010627-concur
3591    Description:
3592    
3593    Fixed a small bug in CM's handling of parallel compilation.
3594    (You could observe the bug by Control-C-interrupting an ordinary
3595    CMB.make or CM.stabilize and then attaching some compile servers.
3596    The result was that all of a sudden the previously interrupted
3597    compilation would continue on its own.  This was because of
3598    an over-optimization: CM did not bother to clean out certain queues
3599    when no servers were attached "anyway", resulting in the contents
3600    of these queues to grab control when new servers did get attached.)
3601    
3602    There is also another minor update to the CM manual.
3603    
3604    ----------------------------------------------------------------------
3605    Name: Matthias Blume
3606    Date: 2001/06/26 16:15:00 EDT
3607    Tag: blume-20010626-cmdoc
3608    Description:
3609    
3610    Minor typo fixed in CM manual (syntax diagram for libraries).
3611    
3612    ----------------------------------------------------------------------
3613    Name: Matthias Blume