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 659, Mon Jun 12 07:37:22 2000 UTC revision 1539, Tue Jul 13 19:05:30 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: Allen Leung (leunga (at) reservoir (dot) com)
17    Date: 2004/07/13 15:04:00 EST
18    Tag: leunga-110_46_1-ppc-lwarx
19    Description:
20    
21       Added the two instructions LWARX and STWCX to the PowerPC
22    instruction set.
23    
24       A (untested) rewrite of loop-structure.sml.  The old version
25    is completely broken.
26    
27    ----------------------------------------------------------------------
28    Name: Matthias Blume (blume (at) tti - c (dot) org)
29    Date: 2004/07/13 13:50:00 CDT
30    Tag: blume-20040713-nlffi
31    Description:
32    
33    - use paramAlloc to report c-calls with too many arguments
34      (for PPC version where parameter area is pre-allocated)
35    - added ccall_maxargspace to machspec (to implement the above)
36    - made "make" commend in CM's "make" tool configurable
37    - added option (default: on) for passing the name of the SML/NJ's "bin"
38      directory to "make"; the call looks like this:
39    
40         make <options> SMLNJ_BINDIR=<dir> <target>
41    
42      This can be used by the Makefile to, e.g., pick the "right" version
43      of ml-nlffigen.
44    - minor code tweaks
45    
46    ----------------------------------------------------------------------
47    Name: Matthias Blume (blume (at) tti - c (dot) org)
48    Date: 2004/07/12 22:50:00 CDT
49    Tag: blume-110_46_1-macosx-nlffi
50    Description:
51    
52    NLFFI under Mac OS X now working (sort of).  This is largely untested,
53    though.
54    
55    Note:  1. You have to make a new, clean build of the runtime system.
56           2. There are new BOOTFILES, you have to use them!
57              (Doing the bootstrap process yourself would be *very* painful!
58               If you absolutely have to do it, build the system under
59               a different architecture and then cross-compile.)
60    
61    Version bumped to 110.46.1 to account for runtime data format changes.
62    
63    ----------------------------------------------------------------------
64    Name: Matthias Blume (blume (at) tti - c (dot) org)
65    Date: 2004/06/18 14:30:00 CDT
66    Tag: blume-20040618-unix
67    Description:
68    
69    Changed the implementation of structure Unix so that the same stream
70    is returned every time one of the {text,bin}{In,Out}streamOf functions
71    is invoked on the same proc.  This is not what the spec currently
72    says -- although IMO it arguably should.  (See discussion below.)
73    
74    ----------------------------------------------------------------------
75    Name: Matthias Blume (blume (at) tti - c (dot) org)
76    Date: 2004/06/17 18:15:00 CDT
77    Tag: Release_110_46
78    Description:
79    
80    New working version (110.46).  NEW BOOTFILES!
81    
82    ----------------------------------------------------------------------
83    Name: Matthias Blume (blume (at) tti - c (dot) org)
84    Date: 2004/06/17 17:20:00 CDT
85    Tag: blume-20040617-timer-unix
86    Description:
87    
88    Changed the interface of structures Timer and Unix to match the most
89    recent Basis spec.
90    
91    In the case of Unix there still seems to be an open/weird issue:
92    
93       The {text,bin}{In,Out}streamOf functions are supposed to create
94       fresh streams whenever they are called -- as opposed to have them
95       return the same stream every time.  This design is supposed to
96       prevent space leaks caused by proc values hanging on to streams.
97    
98       The reap function, on the other hand, is supposed to close the
99       streams.  This cannot be done without having a handle on the
100       stream in proc after all...
101    
102    I took the liberty to implement the following stopgap solution:
103    
104      The proc value hangs on to the most recently created stream(s).
105      Reap closes those.  If either or both of the two streams hadn't
106      been created at all yet, then reap will close the corresponding
107      file descriptors directly.
108    
109    PS: I don't understand the original space leak argument anymore.  If
110    a proc hangs on to the imperative stream, then I/O operations on those
111    will advance the state of the cached stream and avoid the space leak.
112    
113    ----------------------------------------------------------------------
114    Name: Matthias Blume (blume (at) tti - c (dot) org)
115    Date: 2004/05/28 16:45:00 CDT
116    Tag: blume-20040528-basis
117    Description:
118    
119    Added signature PACK_REAL and exported functor PrimIO.
120    
121    ----------------------------------------------------------------------
122    Name: Matthias Blume (blume (at) tti - c (dot) org)
123    Date: 2004/05/25 16:00:00 CDT
124    Tag: blume-20040525-group-owner
125    Description:
126    
127    CM now ignores (but still accepts) the "owner" information in group
128    descriptions.  The owner of a group is its next enclosing
129    library. Each group must have a unique owner.  (There is a virtual
130    "toplevel" library that owns groups which are not nested within a real
131    library.)  Previously, each group had to explicitly declare its owner,
132    and CM would check that such a declaration is correct.  The new scheme
133    is to have CM check that for each group there is precisely one owning
134    library.
135    
136    The advantage of the new scheme is that the programmer no longer needs
137    to maintain the somewhat annoying owner information.  The downside is
138    that CM cannot enforce the ownership rule across multiple runs of
139    CM.make.  Fortunately, enclosing the same group in two different
140    libraries A and B which are not part of the same program does not
141    cause real problems.
142    
143    ----------------------------------------------------------------------
144    Name: Matthias Blume (blume (at) tti - c (dot) org)
145    Date: 2004/05/20 16:00:00 CDT
146    Tag: blume-20040520-win32
147    Description:
148    
149    Made the win32 version work again.  (Strangely, a misplaced comma had
150    slipped into win32-process.c which prevented the runtime from being
151    compiled correctly.)
152    
153    Also, included a minor addition to ml-build.bat analogous to what was
154    done in blume-20040519-ml-build.
155    
156    ----------------------------------------------------------------------
157    Name: Matthias Blume (blume (at) tti - c (dot) org)
158    Date: 2004/05/19 22:10:00 CDT
159    Tag: blume-20040519-ml-build
160    Description:
161    
162    Arranged for ml-build to clean up after itself a little bit better.
163    The script generates a temporary SML source file and compiles it using
164    CM, so CM generates metadata (GUID, SKEL, objectfile) for it.  It now
165    gets rid of those at the end, so they don't accumulate under .cm.
166    
167    This required a minor change to install.sh because the name of the
168    metadata directory (default: .cm) is actually configurable at
169    installation time.
170    
171    ----------------------------------------------------------------------
172    Name: Matthias Blume (blume (at) tti - c (dot) org)
173    Date: 2004/05/18 15:50:00 CDT
174    Tag: blume-20040518-mkreader
175    Description:
176    
177    Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
178    implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
179    
180    ----------------------------------------------------------------------
181    Name: Matthias Blume (blume (at) tti - c (dot) org)
182    Date: 2004/05/11 14:35:00 CDT
183    Tag: blume-20040511-win32sock
184    Description:
185    
186    Added previously missing support for many socket-related functions
187    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
188    for the voluminous patch!
189    
190    (I have not tested this patch under win32 yet.)
191    
192    Here is David's e-mail:
193    
194    Hi,
195    
196    Attached to this email you find a diff against sml/nj 110.45
197    that will enable socket support under Windows.
198    
199    To apply the patch (using unix or cygwin)
200    1) gunzip runtime.diff.gz
201    2) "cd" into "src/runtime" in the source tree of a fresh
202       110.45 installation.
203    3) patch -p 1 < [your/path/to]runtime.diff
204    
205    The code compiles fine but has NOT yet been extensively tested.
206    I only ran a few tests for basic socket client functionality
207    (which worked fine).  Especially the functions that use ioctl
208    are not tested at all and might not work (see below).
209    
210    I implemented this since we want to move to a newer version of sml/nj
211    but need socket support in order to use it.  This is the first time I
212    even had a look at the sml/nj source,  so please review my changes
213    before making this part of the distribution!  Here are a few issues
214    that I think might be better for someone to solve who is more
215    familiar with the sml/nj source (and socket programming):
216    
217    - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
218      exception since I could not figure out what the windows equivalent
219      of these functions is
220    
221    - In sockets-osdep.h there are a some #include statements that are
222      only used in a few files that include sockets-osdep.h
223    
224    - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
225      does not process its return value since I don't know how to
226      report an error upwards.
227    
228    - It would probably be good to have a call to WSACleanup() when
229      the library is unloaded (if there is such a possibility).
230      Otherwise I think Windows will take care of this automatically
231      when the process finishes.
232    
233    - I used ioctlsocket() as a replacement for ioctl() but I have
234      no idea if that is actually the proper replacement on Windows.
235    
236    - All these issues are marked in the code by "FIXME" comments.
237    
238    We use sml/nj extensively in our products and are quite happy
239    with it.  I hope this contribution will help you.
240    
241    Keep up the good work!
242    
243    David
244    
245    ----------------------------------------------------------------------
246    Name: Matthias Blume (blume (at) tti - c (dot) org)
247    Date: 2004/05/11 14:20:00 CDT
248    Tag: blume-20040511-installml
249    Description:
250    
251    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
252    the patch.)
253    
254    ----------------------------------------------------------------------
255    Name: Matthias Blume (blume (at) tti - c (dot) org)
256    Date: 2004/05/11 14:05:00 CDT
257    Tag: blume-20040511-nlffi-netbsd
258    Description:
259    
260    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
261    the patch.)
262    
263    ----------------------------------------------------------------------
264    Name: Matthias Blume (blume (at) tti - c (dot) org)
265    Date: 2004/05/11 12:05:00 CDT
266    Tag: blume-20040511-exports
267    Description:
268    
269    As per request by Adam Chlipala <adam@hcoop.net>, extended various
270    export lists in compiler-related .cm-files.
271    
272    ----------------------------------------------------------------------
273    Name: Matthias Blume (blume (at) tti - c (dot) org)
274    Date: 2004/05/11 11:35:00 CDT
275    Tag: blume-20040511-allsource
276    Description:
277    
278    The installer now honors the "src-smlnj" target again, although its meaning
279    has changed from "all sources required for the compiler" to "all sources
280    the installer knows about".  In other words, if you enable "src-smlnj"
281    in the "targets" file, then the installer will pull in sources for
282    everything.  (Notice that this refers to source code only.  Compiled
283    code is still only installed for modules that were requested explicitly
284    or which are required for other modules that were requested explicitly.)
285    
286    ----------------------------------------------------------------------
287    Name: Matthias Blume (blume (at) tti - c (dot) org)
288    Date: 2004/04/23 17:40:00 CDT
289    Tag: blume-20040423-ieee-scan
290    Description:
291    
292    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
293    in the exponent calculation we get INF or ZERO (depending on the mantissa
294    and the sign of the exponent).
295    
296    ----------------------------------------------------------------------
297    Name: Matthias Blume (blume (at) tti - c (dot) org)
298    Date: 2004/04/23 10:40:00 CDT
299    Tag: blume-20040423-ml-build
300    Description:
301    
302    The ml-build script now terminates with a non-0 status when something
303    goes wrong.
304    
305    ----------------------------------------------------------------------
306    Name: Matthias Blume (blume (at) tti - c (dot) org)
307    Date: 2004/04/22 16:35:00 CDT
308    Tag: blume-20040422-Option
309    Description:
310    
311    Made exception Option to be the same as exception Option.Option
312    (as it should be).
313    
314    ----------------------------------------------------------------------
315    Name: Allen Leung (leunga (at) reservoir (dot) com)
316    Date: 2004/03/19 14:40:00 EST
317    Tag: leunga-20040319-cygwin-nlffi
318    Description:
319    
320    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
321    of SML/NJ.  The problem is that
322    
323       lib = dlopen(NULL, ...)
324       f   = dlsym(lib, "malloc");
325    
326    does not work on Windows unless we explicitly export symbols
327    such as 'malloc' during linking.  We fixed this by explicitly
328    exporting the required symbols with the magic gcc incantation:
329    
330        -Wl,--export-all cygwin.def
331    
332    where cygwin.def is a file containing all the symbols that
333    we wish to export.
334    
335    I suspect this is a Windows problem and we'll have to
336    do the same (somehow with windows compilers) when
337    we build the native win32 version with the system
338    calls LoadLibrary/GetProcAddress.
339    
340    ----------------------------------------------------------------------
341    Name: Matthias Blume (blume (at) tti - c (dot) org)
342    Date: 2004/03/04 16:35:00 CST
343    Tag: blume-20040304-intinf-fmt
344    Description:
345    
346    Fixed problem with IntInf.fmt (sign would show up on the right instead
347    of on the left for BIN, OCT, and HEX).
348    
349    ----------------------------------------------------------------------
350    Name: Matthias Blume (blume (at) tti - c (dot) org)
351    Date: 2004/03/04 11:25:00 CST
352    Tag: blume-20040304-symlinks
353    Description:
354    
355    Fixed problem with installer script (unix only) where bin/ml-yacc and
356    friends pointed (via symlinks) to absolute locations instead of just
357    .run-sml.  This was reported by Vesa A Norrman.
358    
359    ----------------------------------------------------------------------
360    Name: Matthias Blume (blume (at) tti - c (dot) org)
361    Date: 2004/02/13 14:50:00 CST
362    Tag: Release_110_45
363    Description:
364    
365    New working version (110.45). New bootfiles.
366    
367    ----------------------------------------------------------------------
368    Name: Matthias Blume (blume (at) tti - c (dot) org)
369    Date: 2004/01/26 15:15:15 CST
370    Tag: blume-20040126-toplevel
371    Description:
372    
373    Improved handling of exceptions at the interactive toplevel.
374    
375    ----------------------------------------------------------------------
376    Name: Matthias Blume (blume (at) tti - c (dot) org)
377    Date: 2004/01/26 11:25:00
378    Tag: blume-20040126-app
379    Description:
380    
381    Type of top-level "app" corrected.
382    Added code for setting vp_limitPtrMask to Win32-specific runtime.
383    
384    ----------------------------------------------------------------------
385    Name: Matthias Blume (blume (at) tti - c (dot) org)
386    Date: 2003/11/18 17:10 CST
387    Tag: blume-20031118-basis-fiddle
388    Description:
389    
390    - changed Timer interface to what might become the spec
391    - POSIX_FLAGS -> BIT_FLAGS according to spec
392    - some other minor discrepancies wrt. spec eliminated
393    
394    ----------------------------------------------------------------------
395    Name: Matthias Blume (blume (at) tti - c (dot) org)
396    Date: 2003/11/06 12:00:00 CST
397    Tag: Release_110_44
398    Description:
399    
400    New working version (110.44). New bootfiles.
401    
402    ----------------------------------------------------------------------
403    Name: Matthias Blume (blume (at) tti - c (dot) org)
404    Date: 2003/11/04 11:50:00 CST
405    Tag: blume-20031104-move-libraries
406    Description:
407    
408    Eliminated the "dont_move_libraries" directive in config/targets.
409    (The mechanism was broken and could not be fixed easily.  Moreover,
410    there does not seem to be any reason not to move all libraries into
411    lib during installation.  I originally implemented this directive as a
412    backward-compatibility feature when I first introduced the new CM.
413    Now that things have been stable for a long time and going back to the
414    old CM is not an option, there is no reason to keep it around.)
415    
416    ----------------------------------------------------------------------
417    Name: Matthias Blume (blume (at) tti - c (dot) org)
418    Date: 2003/11/03 16:00:00 CST
419    Tag: blume-20031103-installdir
420    Description:
421    
422    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
423    Richards for pointing out the problem and providing the solution.)
424    
425    ----------------------------------------------------------------------
426    Name: Matthias Blume (blume (at) tti - c (dot) org)
427    Date: 2003/10/01 17:05:00 CDT
428    Tag: blume-20031001-lal-mlrisc
429    Description:
430    
431    MLRISC bug fix from Lal.
432    
433    ----------------------------------------------------------------------
434    Name: Matthias Blume (blume (at) tti - c (dot) org)
435    Date: 2003/09/30 16:10:00 CDT
436    Tag: blume-20030930-primio-bat
437    Description:
438    
439    1. Added openVector, nullRd, and nullWr to PRIM_IO.
440    2. Improved .bat files (for Win32 port) to make things work under Win95.
441       (thanks to Aaron S. Hawley for this one)
442    
443    ----------------------------------------------------------------------
444    Name: Matthias Blume (blume (at) tti - c (dot) org)
445    Date: 2003/09/26 16:05:00 CDT
446    Tag: blume-20030926-wrappriv
447    Description:
448    
449    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
450    
451    ----------------------------------------------------------------------
452    Name: Matthias Blume (blume (at) tti - c (dot) org)
453    Date: 2003/09/26 15:00:00 CDT
454    Tag: blume-20030926-110_43_3
455    Description:
456    
457    - additional cleanup
458    - version number bump, NEW BOOTFILES
459    
460    ----------------------------------------------------------------------
461    Name: Matthias Blume (blume (at) tti - c (dot) org)
462    Date: 2003/09/26 12:00:00 CDT
463    Tag: blume-20030926-ppautoload
464    Description:
465    
466    I modified the read-eval-print loop so that the autoloader gets
467    invoked whenever the prettyprinter tries to look up a symbol that
468    is not currently defined in the toplevel environment but which
469    appears in CM's autoload registry.  As a result, we see far fewer of
470    those ?.Foo.Bar.xxx names in the prettyprinter's output.
471    
472    In addition to this I tried to clean up some pieces of the Basis
473    implementation (e.g., Socket, Word8Array) in order to prevent other
474    instances of these ?.Foo.Bar.xxx names from being printed.
475    
476    The mechanism that picks names for types still needs some work, though.
477    (Right now it seems that if there is a type A.t which is defined to
478    be B.u, but B is unavailable at toplevel, then A.t gets printed as
479    "?.B.u" although the perhaps more sensible solution would be to use
480    "A.t" in this case.  In other words, the prettyprinter should follow
481    a chain of DEFtycs not farther than there are corresponding toplevel
482    names in the current environment.)
483    
484    ----------------------------------------------------------------------
485    Name: Matthias Blume (blume (at) tti - c (dot) org)
486    Date: 2003/09/24 16:31:00 CDT
487    Tag: blume-20030924-installer
488    Description:
489    
490    Another installer tweak:  All the ML code for the installer is now
491    compiled during CMB.make and put into a little library called
492    $smlnj/installer.cm.  The installation then simply invokes
493    
494       sml -m $smlnj/installer.cm
495    
496    and everything happens automagically.
497    
498    Win32:  ML code senses value of environment variable SMLNJ_HOME.
499    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
500            and BINDIR.
501    
502    The new scheme guarantees that the ML code responsible for the installation
503    is in sync with the APIs of the main system.  Also, the installer is
504    somewhat faster because the installer script is precompiled.
505    
506    ----------------------------------------------------------------------
507    Name: Matthias Blume (blume (at) tti - c (dot) org)
508    Date: 2003/09/24 15:35:00 CDT
509    Tag: blume-20030924-synsock
510    Description:
511    
512    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
513    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
514    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
515    implementations of non-blocking operations from CML's Socket
516    structure.
517    
518    ----------------------------------------------------------------------
519    Name: Matthias Blume (blume (at) tti - c (dot) org)
520    Date: 2003/09/24 15:10:05 CDT
521    Tag: blume-20030924-sockets
522    Description:
523    
524    1. Fixed SOCKET API and implementation to match Basis spec.
525       This required changing the internal representation of sockets to one
526       that remembers (for each socket file descriptor) whether it is currently
527       blocking or non-blocking.  This state is maintained lazily (i.e., a system
528       call is made only if the state actually needs to change).
529    
530    2. OS-specific details of sockets were moved into separate files, thus
531       making it possible to unify the bulk of the socket implementations
532       between Unix and Win32.
533    
534    3. CML's socket API changed accordingly.
535       (Note that we need to remove non-blocking functions from this API
536       since they are redundant in the case of CML!)
537    
538    4. CML's socket implementation now makes use of non-blocking functions
539       provided by Basis, thus removing all OS-dependent code from this part
540       of CML.
541    
542    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
543    
544    ----------------------------------------------------------------------
545    Name: Matthias Blume (blume (at) tti - c (dot) org)
546    Date: 2003/09/22 12:10:00 CDT
547    Tag: blume-20030922-110_43_2
548    Description:
549    
550    Made a new interim version and bootfiles for developer's bootstrapping
551    convenience.
552    
553       110.43.2 -- NEW BOOTFILES
554    
555    ----------------------------------------------------------------------
556    Name: Matthias Blume (blume (at) tti - c (dot) org)
557    Date: 2003/09/19 15:55:00 CDT
558    Tag: blume-20030919-cmdir
559    Description:
560    
561    1. new-install.sh -> install.sh
562    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
563    3. tweaked installer so that another name instead of .cm can be chosen
564       at install time (by setting the CM_DIR_ARC environment variable
565       during installation); once installation is complete, the name is
566       fixed
567    
568    ----------------------------------------------------------------------
569    Name: Matthias Blume (blume (at) tti - c (dot) org)
570    Date: 2003/09/18 16:00:00 CDT
571    Tag: blume-20030918-110_43_1
572    Description:
573    
574    Made a new interim version and bootfiles for developer's bootstrapping
575    convenience.
576    
577       110.43.1 -- NEW BOOTFILES
578    
579    ----------------------------------------------------------------------
580    Name: Matthias Blume (blume (at) tti - c (dot) org)
581    Date: 2003/09/18 15:20:00 CDT
582    Tag: blume-20030918-misc
583    Description:
584    
585    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
586       this seems to be controversial at the moment) and used those in
587       Posix.ProcEnv.times.
588    
589    2. Added Time.{from,to}Nanoseconds to Time.
590    
591    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
592       For example, fromLargeInt never needs to look at more than 3 "big
593       digits" to get its 53 bits of precision.
594    
595    ----------------------------------------------------------------------
596    Name: Matthias Blume (blume (at) tti - c (dot) org)
597    Date: 2003/09/17 16:30:00 CDT
598    Tag: blume-20030917-real32-slices
599    Description:
600    
601    Added an entry to the primitive environment
602    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
603    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
604    
605    Removed some of the "magic" constants in real64.sml and replaced them
606    with code that generates these values from their corresponding
607    integer counterparts.
608    
609    Made all(?) the slice-related changes to the Basis and made everything
610    compile again...
611    
612    ----------------------------------------------------------------------
613    Name: Matthias Blume (blume (at) tti - c (dot) org)
614    Date: 2003/09/15 17:45:00 CDT
615    Tag: blume-20030915-rbase
616    Description:
617    
618    Fixed bug in Real.fromLargeInt.
619    
620    ----------------------------------------------------------------------
621    Name: Matthias Blume (blume (at) tti - c (dot) org)
622    Date: 2003/09/13 18:11:00 CDT
623    Tag: blume-20030913-libinstall
624    Description:
625    
626    Minor bugfix in config/libinstall (set anchor with path to
627    standalone tool after installing it, otherwise libraries that
628    need ml-lex or ml-yacc won't compile the first time the installer
629    runs).
630    
631    ----------------------------------------------------------------------
632    Name: Matthias Blume (blume (at) tti - c (dot) org)
633    Date: 2003/09/12 11:45:00 CDT
634    Tag: blume-20030912-various
635    Description:
636    
637     - fixed bug in Real.toLargeInt
638     - fixed bug in Posix.ProcEnv.times
639     - changed inputLine functions to return an option
640     - minor installer improvements / bugfixes
641     - changed default @SMLalloc parameter for x86/celeron to 64k
642    
643    ----------------------------------------------------------------------
644    Name: Matthias Blume (blume (at) tti - c (dot) org)
645    Date: 2003/09/09 22:00:00 CDT
646    Tag: Release_110_43
647    Description:
648    
649        New working release 110.43.  New bootfiles.
650    
651    ----------------------------------------------------------------------
652    Name: Matthias Blume (blume (at) tti - c (dot) org)
653    Date: 2003/09/09 19:20:00 CDT
654    Tag: blume-20030909-installer
655    Description:
656    
657    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
658    Modified config/install.bat to take advantage of it.  Also modified
659    config/install.sh (and called it config/new-install.sh) to take advantage
660    of it on Unix systems.  (The SML code is (supposed to be) platform-
661    independent.)
662    
663    The installer can now install everything under Win32
664    as well as under *nix as long as it compiles.
665    
666    Other changes:
667    
668      - made CML compile again under Win32
669      - made eXene compile under Win32 (by providing a fake structure UnixSock
670        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
671      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
672        same as type int;  under Win32 it isn't
673      - fixed some slice-related problems in the win32-specific parts of CML
674      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
675        (under Win32, the volume name is case-insensitive, and the
676        OS.Path code compares volume names for equality)
677    
678    ----------------------------------------------------------------------
679    Name: Matthias Blume (blume (at) tti - c (dot) org)
680    Date: 2003/09/08 11:55:00 CDT
681    Tag: blume-20030908-fullpath
682    Description:
683    
684    Made Win32 version of OS.FileSys.fullPath return current directory
685    when given an empty string.  This is what the spec says, and incidentally,
686    CM depends on it.  (CM otherwise goes into an infinite loop in certain
687    cases when presented with the name of a non-existing .cm file.)
688    
689    ----------------------------------------------------------------------
690    Name: Matthias Blume (blume (at) tti - c (dot) org)
691    Date: 2003/09/04 16:30:00 CDT
692    Tag: blume-20030905-slices-etc
693    Description:
694    
695      1. Changed interface to vectors and arrays in Basis to match
696         (draft) Basis spec.
697      2. Added signatures and implementations of slices according to
698         Basis spec.
699      3. Edited source code throughout the system to make it compile again
700         under 1. and 2.  (In some cases code had to be added to have it
701         match the new signatures.)
702      4. MLRISC should be backward-compatible: the copies of the originals
703         of files that needed to change under 3. were retained, the .cm files
704         check the compiler version number and use old versions when
705         appropriate.
706      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
707         dirstream -> string option (in accordance with Basis spec).
708      6. When generating code that counts lines, ml-lex used function
709         CharVector.foldli, taking advantage of its old interface.
710         This has been replaced with the corresponding code from
711         CharVectorSlice.  (html-lex must be re-lexed!)
712      7. BitArray in smlnj-lib/Util has been extended/modified to match the
713         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
714      8. Removed temporary additions (fromInternal, toInternal) from the
715         (now obsolete) IntInf in smlnj-lib/Util.
716      9. Cleaned up structure Byte.
717     10. Added localOffset, scan, and fromString to Date (according to spec).
718         Cleaned/corrected implementation of Date.
719         (Still need to check for correctness; implement better canonicalizeDate.)
720     11. Added "scan" to signature IEEE_REAL.
721     12. Some improvements to IntInf [in particular: efficiency-hack for
722         mod and rem when second operand is 2 (for parity checks).]
723     13. Changed representation of type Time.time, using a single IntInf.int
724         value counting microseconds.  This considerably simplified the
725         implementation of structure Time.  We now support negative time
726         values; scan and fromString handle signs.
727     14. Functor PrimIO now takes two additional arguments (VectorSlice and
728         ArraySlice).
729    
730    ----------------------------------------------------------------------
731    Name: Matthias Blume (blume (at) tti - c (dot) org)
732    Date: 2003/08/28 17:00:00 CDT
733    Tag: blume-20030828-intinf
734    Description:
735    
736    This is a major update which comes with a version number bump
737    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
738    and an implementation of IntInf in the Basis.
739    
740    There are a fairly large number of related changes and updates throughout
741    the system:
742    
743    Basis:
744            - Implemented IntInf.
745            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
746            - Added some missing Real64 operations, most notably Real.toLargeInt.
747            - Added FixedInt as a synonym for Int32.
748    
749    compiler:
750            * Added support for a built-in intinf type.
751              - literals
752              - pattern matching
753              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
754              - overloading on literals and operations
755    
756            This required adding a primitive type intinf, some additional
757            primops, and implementations for several non-trivial intinf
758            operations in Core.  (The intinf type is completely abstract
759            to the compiler; all operations get delegated back to the Core.)
760    
761            * Intinf equality is handled by polyequal.  However, the compiler
762            does not print its usual warning in this case (since polyequal
763            is the right thing to do there).
764    
765            * Improved the organization of structure InlineT.
766    
767            * A word about conversion primops:
768            If conversions involving intinf do not cancel out during
769            CPS contract, then the compiler must insert calls to Core functions.
770            Since all core access must be resolved already during the FLINT
771            translate phase, it would be too late a the time of CPS contract
772            to add new Core calls.  For this reason, conversion primops
773            for intinf carry two arguments:  1. the numeric argument that
774            they are supposed to convert, and 2. the Core function that
775            can help with this conversion if necessary.  If CPS contract
776            eliminates a primop, then the associated Core function becomes
777            dead and goes away.  Intinf conversion primops that do not get
778            eliminated by CPS contract get rewritten into calls of their
779            core functions by a separate, new phase.
780    
781    interactive system:
782            - Control.Print.intinfDepth controls max length of intinf constants
783            being printed. (Analogous to Control.Print.stringDepth.)
784            - Cleanup in printutil and pputil: got rid of unused stuff and
785            duplicates; replaced some of the code with code that makes better
786            use of library functionality.
787    
788    CM:
789            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
790            are no longer silent.
791    
792    CKIT:
793            Fixed mismatched uses of Int32 and LargeInt.  I always decided
794            in favor of LargeInt -- which is now the same as IntInf.
795            CKIT-knowledgable people should check whether this is what's
796            intended and otherwise change things back to using Int32 or
797            FixedInt.
798    
799    Throughout the code:
800            Started using IntInf.int literals and built-in operations
801            (e.g., comparison with 0) where this seems appropriate.
802    
803    
804    ----------------------------------------------------------------------
805    Name: Dave MacQueen (dbm@cs.uchicago.edu)
806    Date: 2003/08/13 11:36:00 CDT
807    Tag: dbm-20030813-mcz-merge1
808    Description:
809    
810    Merging changes from the mcz-branch development branch into trunk.
811    These changes involve replacement of the emulated old prettyprinter
812    interface with direct use of the SML/NJ Lib PP library, and fixing
813    of a couple of bugs (895, 1186) relating to error messages.  A new
814    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
815    has been added.
816    
817    ----------------------------------------------------------------------
818    Name: Matthias Blume (blume (at) tti - c (dot) org)
819    Date: 2003/08/11 15:45:00 CDT
820    Tag: blume-20030811-windows
821    Description:
822    
823    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
824    
825        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
826    
827    This patch restores SML/NJ's ability to run under win32.  There are a
828    number of changes, including fixes for several bugs that had gone
829    unnoticed until now:
830    
831      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
832        trying to run the win32 version from within cygwin.)
833    
834      - There are a number of simple .bat scripts that substitute for their
835        corresponding Unix shell-scripts.  (See below.)
836    
837      - The internals of ml-build have been modified slightly.  The main
838        difference is that instead of calling ".link-sml" (or link-sml.bat)
839        using OS.Process.system, the ML process delegates this task back
840        to the script.  Otherwise problems arise in mixed environments such
841        as Cygwin where scripts look and work like Unix scripts, but
842        where OS.Process.system cannot run them.
843    
844      - In CM, the srcpath pickler used native pathname syntax -- which
845        is incorrect in the case of cross-compilation.  The new pickle format
846        is independent of platform-specific naming conventions.
847    
848      - Path configuration files (such as lib/pathconfig) can now choose
849        between native and standard syntax.  Placing a line of the form
850    
851            standard!
852    
853        into the file causes all subsequent paths to be interpreted using
854        CM standard pathname syntax (= Unix conventions); a line
855    
856            native!
857    
858        switches back to native style.  This was needed so that
859        path config files can be written portably, see src/system/pathconfig.
860    
861      - Runtime system:
862    
863          - win32-filesys.c:  get_file_time and set_file_time now
864            access modification time, not creation time.
865    
866          - I/O code made aware of new array representation.
867    
868          - Bug fixes in X86.prim.masm.
869    
870       - src/system/makeml made aware of win32. (For use under cygwin
871            and other Unix-environments for windows.)
872    
873       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
874         which caused BinIO.inputAll to fail consistently.
875    
876    .bat scripts:
877    
878       Windows .bat scripts assume that SMLNJ_HOME is defined.
879    
880       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
881         applications (sml, ml-yacc, ml-lex).
882       - ml-build.bat: analogous to ml-build.
883       - config\install.bat:  Analogous to config/install.sh.  This requires
884         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
885         (nmake etc. must be on the path, and vcvars32 must have been run.)
886         Moreover, sources for ml-lex and ml-yacc need to exist under src,
887         and the bootfile hierarchy must have been unpacked under
888         sml.boot.x86-win32.
889         The script is very primitive and does a poor job at error checking.
890         It only installs the base system, ml-lex, and ml-yacc.  No other
891         libraries are being installed (i.e., you get only those that
892         are part of the compiler.)
893       - link-sml.bat:  analogous to .link-sml, but not currently used
894    
895    Unrelated bug fixes:
896    
897      - ml-nlffigen now exports structures ST_* corresponding to incomplete
898        types.
899      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
900        otherwise.)
901    
902    ----------------------------------------------------------------------
903    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
904    Date: 2003/06/17
905    Tag: macqueen-20030617-bug895
906    Description:
907    
908    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
909    Tag will be used for new development branch (mcz-branch) for
910    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
911    
912    ----------------------------------------------------------------------
913    Name: Matthias Blume (blume (at) tti - c (dot) org)
914    Date: 2003/05/27 16:55:00 CDT
915    Tag: blume-20030527-polyeq
916    Description:
917    
918    Tried to eliminated most cases of polymorphic equality.
919    
920    ----------------------------------------------------------------------
921    Name: Matthias Blume (blume (at) tti - c (dot) org)
922    Date: 2003/05/21 17:45:00 CDT
923    Tag: blume-20030517-complete
924    Description:
925    
926    Two changes:
927    
928      1. Added a flag for controlling whether non-exhaustive bindings will
929         be treated as errors (default is false).
930      2. Cleaned up the *entire* source tree so that CMB.make goes through
931         without a single non-exhaustive match- or bind warning.
932    
933    ----------------------------------------------------------------------
934    Name: Matthias Blume (blume (at) tti - c (dot) org)
935    Date: 2003/05/17 10:20:00 CDT
936    Tag: blume-20030517-absyn
937    Description:
938    
939    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
940    
941       This mainly affects the quality of error messages.  However, some
942       of the code is now more straightforward than before. (Treatment of
943       the above four constructs in translate.sml is much simpler than
944       the "macro-expansion" that was going on before.  Plus, the mach-
945       compiler no longer gets invoked just to be able to compile an
946       if-expression.)
947    
948    2. The ErrorMsg.Error exception is now caught and absorbed by the
949       interactive loop.
950    
951    ----------------------------------------------------------------------
952    Name: Allen Leung
953    Date: 2003/05/16 13:05:00 CDT
954    Tag: leunga-20030516-cygwin-runtime
955    Description:
956    
957        Ported the runtime system to cygwin, which uses the unix
958    x86-unix bin files.  Missing/buggy features:
959    
960       o getnetbyname, getnetbyaddr: these functions seem to be missing in
961         the Cygwin library.
962       o Ctrl-C handling may be flaky.
963       o Windows system calls and Windows I/O are not supported.
964    
965       A new set of binfiles is located at:
966    
967         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
968    
969       This is only needed for bootstrapping the cygwin version of smlnj.
970       Other x86 versions can use the existing binfiles.
971    
972    ----------------------------------------------------------------------
973    Name: Matthias Blume
974    Date: 2003/04/08 15:42:00 CDT
975    Tag: blume-20030408-listpair
976    Description:
977    
978    1. Added a target 'mlrisc' to installer.
979    
980    2. Added missing elements to structure ListPair.
981    
982    ----------------------------------------------------------------------
983    Name: Allen Leung
984    Date: 2003/01/07 10:40:00 EST
985    Tag: leunga-20030107-int-rem
986    Description:
987    
988    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
989    The arguments to the SUBL instruction were swapped.
990    
991    ----------------------------------------------------------------------
992    Name: Matthias Blume
993    Date: 2002/12/12 16:25:00 EST
994    Tag: blume-20021212-risc-ra
995    Description:
996    
997    Fixed a serious bug in the rewrite code for FP spilling/reloading that
998    sent the RA into an infinite loop when floating point registers get
999    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
1000    110.37 and 110.38.)
1001    There was another set of potential problems related to the handling of
1002    MLRISC annotations (but those did not yet cause real problems, apparently).
1003    
1004    ----------------------------------------------------------------------
1005    Name: Matthias Blume
1006    Date: 2002/12/06 22:40:00 EST
1007    Tag: blume-20021206-cm-fileid
1008    Description:
1009    
1010    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
1011    This fixes the problem of CM getting confused by files that suddenly
1012    change their identity (e.g., by getting unlinked and recreated by some
1013    text editor such as vi).  There might be a better/cheaper/cleaner way
1014    of doing this, but for now this will have to do.
1015    
1016    ----------------------------------------------------------------------
1017    Name: Matthias Blume
1018    Date: 2002/10/28 09:50:00 EST
1019    Tag: blume-20021028-typecheck
1020    Description:
1021    
1022    Exported structure Typecheck from $smlnj/viscomp/core.cm.
1023    
1024    ----------------------------------------------------------------------
1025    Name: Matthias Blume
1026    Date: 2002/10/17 09:10:00 EDT
1027    Tag: Release_110_42
1028    Description:
1029    
1030    In good old tradition, there has been a slight hiccup so that we have
1031    to patch 110.42 after the fact.   The old release tag has been replaced
1032    (see below).
1033    
1034    The change solves a problem with two competing approaches the
1035    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
1036    each other's way.
1037    
1038    This change only affects the runtime system code and the installer script.
1039    (No new bootfiles.)
1040    
1041    ----------------------------------------------------------------------
1042    Name: Matthias Blume
1043    Date: 2002/10/16 12:00:00 EDT
1044    Tag: Release_110_42_removed
1045    Description:
1046    
1047    New working release.  New bootfiles.
1048    
1049    ----------------------------------------------------------------------
1050    Name: Matthias Blume
1051    Date: 2002/10/10 13:10:00 EDT
1052    Tag: blume-20021010-ppc-divs
1053    Description:
1054    
1055    The mltree operator DIVS must be implemented with an overflow check on
1056    the PPC because the hardware indicates divide-by-zero using "overflow" as
1057    well.
1058    
1059    ----------------------------------------------------------------------
1060    Name: Matthias Blume
1061    Date: 2002/07/23 11:45:00
1062    Tag: blume-20020723-smlnj-home
1063    Description:
1064    
1065    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
1066    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
1067    CM_PATHCONFIG is also set), the path configuration file is assumed
1068    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
1069    the entire tree to some other place and everything will "just work".
1070    
1071    (Companion commands such as ml-build and ml-makedepend also sense this
1072    variable.)
1073    
1074    ----------------------------------------------------------------------
1075    Name: Matthias Blume
1076    Date: 2002/07/12 21:19:00 EDT
1077    Tag: blume-20020712-liveness
1078    Description:
1079    
1080    Exported two useful "step" functions from liveness module (MLRISC).
1081    
1082    ----------------------------------------------------------------------
1083    Name: Matthias Blume
1084    Date: 2002/07/05 16:00 EDT
1085    Tag: Release_110_41
1086    Description:
1087    
1088    New working release.  New bootfiles.
1089    
1090    ----------------------------------------------------------------------
1091    Name: Matthias Blume
1092    Date: 2002/07/05 10:25:00 EDT
1093    Tag: blume-20020705-btimp
1094    Description:
1095    
1096    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
1097    other clients can set up backtracing support.
1098    
1099    ----------------------------------------------------------------------
1100    Name: Matthias Blume
1101    Date: 2002/06/25 17:23:00 EDT
1102    Tag: blume-20020625-fpmax
1103    Description:
1104    
1105    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
1106    case.  (The sense of the isNaN test was reversed -- which made min and
1107    max always return their first argument.)
1108    
1109    ----------------------------------------------------------------------
1110    Name: Matthias Blume
1111    Date: 2002/06/11
1112    Tag: blume-20020611-unixpath
1113    Description:
1114    
1115    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
1116    
1117    ----------------------------------------------------------------------
1118    Name: Matthias Blume
1119    Date: 2002/06/10 16:35:00 EDT
1120    Tag: blume-20020610-ieeereal
1121    Description:
1122    
1123    I back-ported my implementation of IEEEReal.fromString from the
1124    idlbasis-devel branch so that we can test it.
1125    
1126    Another small change is that ppDec tries to give more information
1127    than just "<sig>" in the case of functors.  However, this code is
1128    broken in some mysterious way if the functor's body's signature
1129    has not been declared by ascription but gets inferred from the
1130    implementation.  This needs fixing...
1131    
1132    ----------------------------------------------------------------------
1133    Name: Matthias Blume
1134    Date: 2002/05/31
1135    Tag: blume-20020531-btrace-mode
1136    Description:
1137    
1138    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
1139    the wayside when I switched over to using Controls everywhere.)
1140    
1141    ----------------------------------------------------------------------
1142    Name: Lal George
1143    Date: 2002/05/23 12:21:40 EDT
1144    Tag: george-20020523-visual-labels
1145    Description:
1146    
1147    Labels are now displayed in the graphical output to make
1148    the fall-through and target blocks obvious.
1149    
1150    ----------------------------------------------------------------------
1151    Name: Matthias Blume
1152    Date: 2002/05/22 11:03:00 EDT
1153    Tag: blume-20020522-shrink
1154    Description:
1155    
1156    John tweaked yesterday's fix for 1131 to handle an out-of-memory
1157    situation that comes up when allocating huge arrays.
1158    
1159    ----------------------------------------------------------------------
1160    Name: Matthias Blume
1161    Date: 2002/05/21 16:00:00 EDT
1162    Tag: Release_110_40
1163    Description:
1164    
1165    New working release (110.40).  New bootfiles.
1166    
1167    [Also: John Reppy fixed GC bug 1131.]
1168    
1169    ----------------------------------------------------------------------
1170    Name: Matthias Blume
1171    Date: 2002/05/21 12:35:00 EDT
1172    Tag: blume-20020521-cmdoc
1173    Description:
1174    
1175    CM documentation update.
1176    
1177    ----------------------------------------------------------------------
1178    Name: Matthias Blume
1179    Date: 2002/05/21 10:55:00 EDT
1180    Tag: blume-20020521-misc
1181    Description:
1182    
1183    - John tweaked runtime to be silent on heap export (except when
1184      GC messages are on).
1185    - I added a few more things (cross-compiling versions of CMB) to
1186      config/preloads (as suggestions).
1187    
1188    ----------------------------------------------------------------------
1189    Name: Matthias Blume
1190    Date: 2002/05/20 22:25:00 EDT
1191    Tag: blume-20020520-controls
1192    Description:
1193    
1194    - Added ControlUtil structure to control-lib.cm.
1195    - Use it throughout.
1196    - Used Controls facility to define MLRISC controls (as opposed to
1197      registering MLRISC control ref cells with Controls after the fact)
1198    - Fixed messed-up controls priorities.
1199    
1200    * Removed again all the stuff from config/preloads that one wouldn't
1201      be able to preload at the time the initial heap image is built.
1202      (Many libraries, e.g., CML, do not exist yet at this time.  The
1203       only libraries that can be preloaded via config/preloads are those
1204       that come bundled with the bootfiles.)
1205    
1206    ----------------------------------------------------------------------
1207    Name: Matthias Blume
1208    Date: 2002/05/20 10:59:00 EDT
1209    Tag: blume-20020520-preloads
1210    Description:
1211    
1212    Added a lot of commented-out suggestions for things to be included
1213    in config/preloads.
1214    
1215    ----------------------------------------------------------------------
1216    Name: Allen Leung
1217    Date: 2002/05/18 14:20:00 EDT
1218    Tag: leunga-20020518-mdl
1219    Description:
1220    
1221    o Made the mdl tool stuff compile and run again.
1222    o I've disabled all the stuff that depends on RTL specifications; they
1223      are all badly broken anyway.
1224    
1225    ----------------------------------------------------------------------
1226    Name: Matthias Blume
1227    Date: 2002/05/17 16:49:00 EDT
1228    Tag: blume-20020517-controls
1229    Description:
1230    
1231    0. John Reppy made several modifications to the SML/NJ library.
1232       In particular, there is a shiny new controls-lib.cm.
1233    
1234    1. Pushed new controls interface through compiler so that everything
1235       compiles again.
1236    
1237    2. Added FormatComb and FORMAT_COMB to the CML version of the
1238       SML/NJ library (so that CML compiles again).
1239    
1240    3. Modified init scripts because XXX_DEFAULT environment variables
1241       are no longer with us.  (Boot-time initialization is now done
1242       using the same environment variables that are also used for
1243       startup-time initialization of controls.)
1244    
1245    ----------------------------------------------------------------------
1246    Name: Lal George
1247    Date: 2002/05/15 09:20:10 EDT
1248    Tag: george-20020515-pseudo-op-decls
1249    Description:
1250    
1251            All pseudo-ops emitted before the first segment declaration
1252    such as TEXT, DATA, and BSS directives are assumed to be global
1253    declarations and are emitted first in the assembly file. This is
1254    useful in a number of situations where one has pseudo-ops that are not
1255    specific to any segment, and also works around the constraint that one
1256    cannot have client pseudo-ops in the TEXT segment.
1257    
1258            Because no segment is associated with these declarations it is
1259    an error to allocate any space or objects before the first segment
1260    directive and an exception will be raised. However, we cannot make
1261    this check for client pseudo-ops.
1262    
1263            These top level declarations are a field in the CFG graph_info.
1264    In theory you can continue to add to this field after the CFG has been
1265    built -- provided you know what you are doing;-)
1266    
1267    ----------------------------------------------------------------------
1268    Name: Matthias Blume
1269    Date: 2002/05/13 16:40:00 EDT
1270    Tag: blume-20020513-pp-etc
1271    Description:
1272    
1273    A few minor bugfixes:
1274    
1275      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1276        (Bogus printouts for redefined bindings still occur.  Compiler
1277        bug should no longer occur now.  We need to redo the prettyprinter
1278        from scratch.)
1279    
1280      - CM pathname printer now also adds escape sequences for ( and )
1281    
1282      - commend and docu fixes for ml-nlffi
1283    
1284    ----------------------------------------------------------------------
1285    Name: Matthias Blume
1286    Date: 2002/05/10 16:40:00 EDT
1287    Tag: blume-20020510-erg-textio
1288    Description:
1289    
1290    Applied the following bugfix provided by Emden Gansner:
1291    
1292        Output is corrupted when outputSubstr is used rather than output.
1293    
1294        The problem occurs when a substring
1295    
1296            ss = (s, dataStart, dataLen)
1297    
1298        where dataStart > 0, fills a stream buffer with avail bytes left.
1299        avail bytes of s, starting at index dataStart, are copied into the
1300        buffer, the buffer is flushed, and then the remaining dataLen-avail
1301        bytes of ss are copied into the beginning of the buffer. Instead of
1302        starting this copy at index dataStart+avail in s, the current code
1303        starts the copy at index avail.
1304    
1305        Fix:
1306        In text-io-fn.sml, change line 695 from
1307             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1308        to
1309             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1310    
1311    ----------------------------------------------------------------------
1312    Name: Matthias Blume
1313    Date: 2002/04/12 13:55:00 EDT
1314    Tag: blume-20020412-assyntax
1315    Description:
1316    
1317    1. Grabbed newer assyntax.h from the XFree86 project.
1318    2. Fiddled with how to compile X86.prim.asm without warnings.
1319    3. (Very) Minor cleanup in CM.
1320    
1321    ----------------------------------------------------------------------
1322    Name: Matthias Blume
1323    Date: 2002/04/01 (no joke!) 17:07:00 EST
1324    Tag: blume-20020401-x86div
1325    Description:
1326    
1327    Added full support for div/mod/rem/quot on the x86, using the machine
1328    instruction's two results (without clumsily recomputing the remainder)
1329    directly where appropriate.
1330    
1331    Some more extensive power-of-two support was added to the x86 instruction
1332    selector (avoiding expensive divs, mods, and muls where they can be
1333    replaced with cheaper shifts and masks).  However, this sort of thing
1334    ought to be done earlier, e.g., within the CPS optimizer so that
1335    all architectures benefit from it.
1336    
1337    The compiler compiles to a fixed point, but changes might be somewhat
1338    fragile nevertheless.  Please, report any strange things that you might
1339    see wrt. div/mod/quot/rem...
1340    
1341    ----------------------------------------------------------------------
1342    Name: Matthias Blume
1343    Date: 2002/03/29 17:22:00
1344    Tag: blume-20020329-div
1345    Description:
1346    
1347    Fixed my broken div/mod logic.  Unfortunately, this means that the
1348    inline code for div/mod now has one more comparison than before.
1349    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1350    The problem was with quotient = 0, because that alone does not tell
1351    us which way the rounding went.  One then has to look at whether
1352    remainder and divisor have the same sign...  :(
1353    
1354    Anyway, I replaced the bootfiles with fresh ones...
1355    
1356    ----------------------------------------------------------------------
1357    Name: Matthias Blume
1358    Date: 2002/03/29 14:10:00 EST
1359    Tag: blume-20020329-inlprims
1360    Description:
1361    
1362    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1363    
1364    Primops have changed. This means that the bin/boot-file formats have
1365    changed as well.
1366    
1367    To make sure that there is no confusion, I made a new version.
1368    
1369    
1370    CHANGES:
1371    
1372    * removed REMT from mltree (remainder should never overflow).
1373    
1374    * added primops to deal with divisions of all flavors to the frontend
1375    
1376    * handled these primops all the way through so they map to their respective
1377      MLRISC support
1378    
1379    * used these primops in the implementation of Int, Int32, Word, Word32
1380    
1381    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1382    
1383    * parameterized INLMIN, INLMAX, and INLABS by a numkind
1384    
1385    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1386      floating point
1387    
1388    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1389      and Real (but Real.abs maps to a separate floating-point-only primop)
1390    
1391    
1392    TODO items:
1393    
1394    * Hacked Alpha32 instruction selection, disabling the selection of REMx
1395      instructions because the machine instruction encoder cannot handle
1396      them.  (Hppa, PPC, and Sparc instruction selection did not handle
1397      REM in the first place, and REM is supported by the x86 machine coder.)
1398    
1399    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1400      selection phase.  (The two can be streamlined because the hardware
1401      delivers both quotient and remainder at the same time anyway.)
1402    
1403    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1404      (Currently the behavior is inconsistent both across architectures and
1405      wrt. the draft Basis spec.)
1406    
1407    * Word8 should eventually be handled natively, too.
1408    
1409    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1410      as if there currently is no execution path that could trigger it in
1411      SML/NJ.  (The assumptions underlying functions arith and promotable do not
1412      hold for things like multiplication and division.)
1413    
1414    ----------------------------------------------------------------------
1415    Name: Matthias Blume
1416    Date: 2002/03/27 16:27:00 EST
1417    Tag: blume-20020327-mlrisc-divisions
1418    Description:
1419    
1420    Added support for all four division operations (ML's div, mod, quot,
1421    and rem) to MLRISC.  In the course of doing so, I also rationalized
1422    the naming (no more annoying switch-around of DIV and QUOT), by
1423    parameterizing the operation by div_rounding_mode (which can be either
1424    DIV_TO_ZERO or DIV_TO_NEGINF).
1425    
1426    The generic MLTreeGen functor takes care of compiling all four
1427    operations down to only round-to-zero div.
1428    
1429    Missing pieces:
1430    
1431      * Doing something smarter than relying on MLTreeGen on architectures
1432        like, e.g., the x86 where hardware division delivers both quotient and
1433        remainder at the same time.  With this, the implementation of the
1434        round-to-neginf operations could be further streamlined.
1435    
1436      * Remove inlining support for div/mod/rem from the frontend and replace it
1437        with primops that get carried through to the backend.  Do this for all
1438        int and word types.
1439    
1440    ----------------------------------------------------------------------
1441    Name: Matthias Blume
1442    Date: 2002/03/25 17:25:00 EST
1443    Tag: blume-20020325-divmod
1444    Description:
1445    
1446    I improved (hopefully without breaking them) the implementation of Int.div,
1447    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1448    advantage of the following observations:
1449    
1450      Let  q = x quot y      r = x rem y
1451           d = x div  y      m = x mod y
1452    
1453    where "quot" is the round-to-zero version of integer division that
1454    hardware usually provides.  Then we have:
1455    
1456         r = x - q * y        where neither the * nor the - will overflow
1457         d = if q >= 0 orelse x = q * y then q else q - 1
1458                              where neither the * nor the - will overflow
1459         m = if q >= 0 orelse r = 0 then r else r + y
1460                              where the + will not overflow
1461    
1462    This results in substantial simplification of the generated code.
1463    The following table shows the number of CFG nodes and edges generated
1464    for
1465            fun f (x, y) = x OPER y
1466            (* with OPER \in div, mod, quot, rem *)
1467    
1468    
1469        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1470        --------------------------------------------------------
1471         div |         24 |         39 |         12 |         16
1472         mod |         41 |         71 |         12 |         16
1473        quot |          8 |         10 |          8 |         10
1474         rem |         10 |         14 |          8 |         10
1475    
1476    
1477    ----------------------------------------------------------------------
1478    Name: Matthias Blume
1479    Date: 2002/03/25 22:06:00 EST
1480    Tag: blume-20020325-cprotobug
1481    Description:
1482    
1483    Fixed a bug in cproto (c prototype decoder).
1484    
1485    ----------------------------------------------------------------------
1486    Name: Matthias Blume
1487    Date: 2002/03/25 16:00:00 EST
1488    Tag: blume-20020325-raw-primops
1489    Description:
1490    
1491    I did some cleanup to Allen's new primop code and
1492    replaced yesterday's bootfiles with new ones.
1493    (But they are stored in the same place.)
1494    
1495    ----------------------------------------------------------------------
1496    Name: Matthias Blume
1497    Date: 2002/03/24 22:40:00 EST
1498    Tag: blume-20020324-bootfiles
1499    Description:
1500    
1501    Made the bootfiles that Allen asked for.
1502    
1503    ----------------------------------------------------------------------
1504    Name: Allen Leung
1505    Date: 2002/03/23 15:50:00 EST
1506    Tag: leunga-20020323-flint-cps-rcc-primops
1507    Description:
1508    
1509      1. Changes to FLINT primops:
1510    
1511        (* make a call to a C-function;
1512         * The primop carries C function prototype information and specifies
1513         * which of its (ML-) arguments are floating point. C prototype
1514         * information is for use by the backend, ML information is for
1515         * use by the CPS converter. *)
1516      | RAW_CCALL of { c_proto: CTypes.c_proto,
1517                       ml_args: ccall_type list,
1518                       ml_res_opt: ccall_type option,
1519                       reentrant : bool
1520                     } option
1521       (* Allocate uninitialized storage on the heap.
1522        * The record is meant to hold short-lived C objects, i.e., they
1523        * are not ML pointers.  With the tag, the representation is
1524        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1525        *)
1526      | RAW_RECORD of {tag:bool,sz:int}
1527      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1528    
1529      2.  These CPS primops are now overloaded:
1530    
1531           rawload of {kind:numkind}
1532           rawstore of {kind:numkind}
1533    
1534          The one argument form is:
1535    
1536             rawload {kind} address
1537    
1538          The two argument form is:
1539    
1540             rawload {kind} [ml object, byte-offset]
1541    
1542      3. RAW_CCALL/RCC now takes two extra arguments:
1543    
1544         a. The first is whether the C call is reentrant, i.e., whether
1545            ML state should be saved and restored.
1546         b. The second argument is a string argument specifying the name of
1547            library and the C function.
1548    
1549         These things are currently not handled in the code generator, yet.
1550    
1551      4. In CProto,
1552    
1553         An encoding type of "bool" means "ml object" and is mapped into
1554         C prototype of PTR.  Note that "bool" is different than "string",
1555         even though "string" is also mapped into PTR, because "bool"
1556         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1557    
1558      5. Pickler/unpicker
1559    
1560         Changed to handle RAW_RECORD and newest RAW_CCALL
1561    
1562      6. MLRiscGen,
1563    
1564         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1565         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1566            in the file CodeGen/cpscompile/cps-c-calls.sml
1567    
1568      7. Added the conditional move operator
1569    
1570             condmove of branch
1571    
1572         to cps.  Generation of this is still buggy so it is currently
1573         disabled.
1574    
1575    ----------------------------------------------------------------------
1576    Name: Lal George
1577    Date: 2002/03/22 14:18:25 EST
1578    Tag: george-20020322-cps-branch-prob
1579    Description:
1580    
1581    Implemented the Ball-Larus branch prediction-heuristics, and
1582    incorporated graphical viewers for control flow graphs.
1583    
1584    Ball-Larus Heuristics:
1585    ---------------------
1586    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1587    
1588    By design it uses the Dempster-Shafer theory for combining
1589    probabilities.  For example, in the function:
1590    
1591        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1592    
1593    the ball-larus heuristics predicts that the n=0 is unlikely
1594    (OH-heuristic), and the 'then' branch is unlikely because of the
1595    RH-heuristic -- giving the 'then' branch an even lower combined
1596    probability using the Dempster-Shafer theory.
1597    
1598    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1599    probability of the 'then' branch because of the loop in the else
1600    branch.
1601    
1602    
1603    Graphical Viewing:
1604    ------------------
1605    I merely plugged in Allen's graphical viewers into the compiler. The
1606    additional code is not much. At the top level, saying:
1607    
1608            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1609    
1610    will display the graphical view of the control flow graph just before
1611    back-patching.  daVinci must be in your path for this to work. If
1612    daVinci is not available, then the default viewer can be changed
1613    using:
1614    
1615            Control.MLRISC.getString "viewer"
1616    
1617    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1618    course, these viewers must be in your path.
1619    
1620    The above will display the compilation unit at the level of clusters,
1621    many of which are small, boring, and un-interesting. Also setting:
1622    
1623            Control.MLRISC.getInt "cfg-graphical-view_size"
1624    
1625    will display clusters that are larger than the value set by the above.
1626    
1627    
1628    ----------------------------------------------------------------------
1629    Name: Matthias Blume
1630    Date: 2002/03/21 22:20:00 EST
1631    Tag: blume-20020321-kmp-bugfix
1632    Description:
1633    
1634    Changed the interface to the KMP routine in PreString and fixed
1635    a minor bug in one place where it was used.
1636    
1637    ----------------------------------------------------------------------
1638    Name: Allen Leung
1639    Date: 2002/03/21 20:30:00 EST
1640    Tag: leunga-20020321-cfg
1641    Description:
1642    
1643      Fixed a potential problem in cfg edge splitting.
1644    
1645    ----------------------------------------------------------------------
1646    Name: Allen Leung
1647    Date: 2002/03/21 17:15:00 EST
1648    Tag: leunga-20020321-x86-fp-cfg
1649    Description:
1650    
1651      1. Recoded the buggy parts of x86-fp.
1652    
1653         a. All the block reordering code has been removed.
1654            We now depend on the block placement phases to do this work.
1655    
1656         b. Critical edge splitting code has been simplified and moved into the
1657            CFG modules, as where they belong.
1658    
1659         Both of these were quite buggy and complex.  The code is now much, much
1660         simpler.
1661    
1662      2. X86 backend.
1663    
1664         a. Added instructions for 64-bit support.  Instruction selection for
1665            64-bit has not been committed, however, since that
1666            requires changes to MLTREE which haven't been approved by
1667            Lal and John.
1668    
1669         b. Added support for FUCOMI and FUCOMIP when generating code for
1670            PentiumPro and above.  We only generate these instructions in
1671            the fast-fp mode.
1672    
1673         c. Added cases for JP and JNP in X86FreqProps.
1674    
1675      3. CFG
1676    
1677         CFG now has a bunch of methods for edge splitting and merging.
1678    
1679      4. Machine description.
1680    
1681         John's simplification of MLTREE_BASIS.fcond broke a few machine
1682         description things:
1683    
1684         rtl-build.{sig,sml} and hppa.mdl fixed.
1685    
1686         NOTE: the machine description stuff in the repository is still broken.
1687               Again, I can't put my fixes in because that involves
1688               changes to MLTREE.
1689    
1690    ----------------------------------------------------------------------
1691    Name: Matthias Blume
1692    Date: 2002/03/20 15:55:00 EST
1693    Tag: blume-20020320-kmp
1694    Description:
1695    
1696    Implemented Knuth-Morris-Pratt string matching in PreString and used
1697    it for String.isSubstring, Substring.isSubstring, and
1698    Substring.position.
1699    
1700    (Might need some stress-testing.  Simple examples worked fine.)
1701    
1702    ----------------------------------------------------------------------
1703    Name: Matthias Blume
1704    Date: 2002/03/19 16:37:00 EST
1705    Tag: blume-20020319-witnesses
1706    Description:
1707    
1708    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1709    
1710    This implements a generic mechanism for changing constness qualifiers
1711    anywhere within big C types without resorting to outright "casts".
1712    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1713    modify the constness at the outermost level.)
1714    The implementation of "convert" is based on the idea of "witness"
1715    values -- values that are not used by the operation but whose types
1716    "testify" to their applicability.  On the implementation side, "convert"
1717    is simply a projection (returning its second curried argument).  With
1718    cross-module inlining, it should not result in any machine code being
1719    generated.
1720    
1721    ----------------------------------------------------------------------
1722    Name: Matthias Blume
1723    Date: 2002/03/15 16:40:00 EST
1724    Tag: blume-20020315-basis
1725    Description:
1726    
1727    Provided (preliminary?) implementations for
1728    
1729      {String,Substring}.{concatWith,isSuffix,isSubstring}
1730    
1731    and
1732    
1733      Substring.full
1734    
1735    Those are in the Basis spec but they were missing in SML/NJ.
1736    
1737    ----------------------------------------------------------------------
1738    Name: Matthias Blume
1739    Date: 2002/03/14 21:30:00 EST
1740    Tag: blume-20020314-controls
1741    Description:
1742    
1743    Controls:
1744    ---------
1745    
1746    1. Factored out the recently-added Controls : CONTROLS stuff and put
1747       it into its own library $/controls-lib.cm.  The source tree for
1748       this is under src/smlnj-lib/Controls.
1749    
1750    2. Changed the names of types and functions in this interface, so they
1751       make a bit more "sense":
1752    
1753          module -> registry
1754          'a registry -> 'a group
1755    
1756    3. The interface now deals in ref cells only.  The getter/setter interface
1757       is (mostly) gone.
1758    
1759    4. Added a function that lets one register an already-existing ref cell.
1760    
1761    5. Made the corresponding modifications to the rest of the code so that
1762       everything compiles again.
1763    
1764    6. Changed the implementation of Controls.MLRISC back to something closer
1765       to the original.  In particular, this module (and therefore MLRISC)
1766       does not depend on Controls.  There now is some link-time code in
1767       int-sys.sml that registers the MLRISC controls with the Controls
1768       module.
1769    
1770    CM:
1771    ---
1772    
1773      * One can now specify the lambda-split aggressiveness in init.cmi.
1774    
1775    ----------------------------------------------------------------------
1776    Name: Allen Leung
1777    Date: 2002/03/13 17:30:00 EST
1778    Tag: leunga-20020313-x86-fp-unary
1779    Description:
1780    
1781    Bug fix for:
1782    
1783    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1784    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1785    > - fun f(x,(y,z)) = Real.~ y;
1786    > [autoloading]
1787    > [autoloading done]
1788    >       fchsl   (%eax), 184(%esp)
1789    > Error: MLRisc bug: X86MCEmitter.emitInstr
1790    >
1791    > uncaught exception Error
1792    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1793    
1794    The problem was that the code generator did not generate any fp registers
1795    in this case, and the ra didn't know that it needed to run the X86FP phase to
1796    translate the pseudo fp instruction.   This only happened with unary fp
1797    operators in certain situations.
1798    
1799    ----------------------------------------------------------------------
1800    Name: Matthias Blume
1801    Date: 2002/03/13 14:00:00 EST
1802    Tag: blume-20020313-overload-etc
1803    Description:
1804    
1805    1. Added _overload as a synonym for overload for backward compatibility.
1806       (Control.overloadKW must be true for either version to be accepted.)
1807    
1808    2. Fixed bug in install script that caused more things to be installed
1809       than what was requested in config/targets.
1810    
1811    3. Made CM aware of the (_)overload construct so that autoloading
1812       works.
1813    
1814    ----------------------------------------------------------------------
1815    Name: Matthias Blume
1816    Date: 2002/03/12 22:03:00 EST
1817    Tag: blume-20020312-url
1818    Description:
1819    
1820    Forgot to update BOOT and srcarchiveurl.
1821    
1822    ----------------------------------------------------------------------
1823    Name: Matthias Blume
1824    Date: 2002/03/12 17:30:00 EST
1825    Tag: blume-20020312-version110392
1826    Description:
1827    
1828    Yet another version number bump (because of small changes to the
1829    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1830    
1831    Changes:
1832    
1833      The new pid generation scheme described a few weeks ago was overly
1834      complicated.  I implemented a new mechanism that is simpler and
1835      provides a bit more "stability":  Once CM has seen a compilation
1836      unit, it keeps its identity constant (as long as you do not delete
1837      those crucial CM/GUID/* files).  This means that when you change
1838      an interface, compile, then go back to the old interface, and
1839      compile again, you arrive at the original pid.
1840    
1841      There now also is a mechanism that instructs CM to use the plain
1842      environment hash as a module's pid (effectively making its GUID
1843      the empty string).  For this, "noguid" must be specified as an
1844      option to the .sml file in question within its .cm file.
1845      This is most useful for code that is being generated by tools such
1846      as ml-nlffigen (because during development programmers tend to
1847      erase the tool's entire output directory tree including CM's cached
1848      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1849      revert to the old, broken behavior of SML/NJ, but in specific cases
1850      where there is no danger of interface confusion, its use is ok
1851      (I think).
1852    
1853      ml-nlffigen by default generates "noguid" annotations.  They can be
1854      turned off by specifying -guid in its command line.
1855    
1856    ----------------------------------------------------------------------
1857    Name: Lal George
1858    Date: 2002/03/12 12 14:42:36 EST
1859    Tag: george-20020312-frequency-computation
1860    Description:
1861    
1862    Integrated jump chaining and static block frequency into the
1863    compiler. More details and numbers later.
1864    
1865    ----------------------------------------------------------------------
1866    Name: Lal George
1867    Date: 2002/03/11 11 22:38:53 EST
1868    Tag: george-20020311-jump-chain-elim
1869    Description:
1870    
1871    Tested the jump chain elimination on all architectures (except the
1872    hppa).  This is on by default right now and is profitable for the
1873    alpha and x86, however, it may not be profitable for the sparc and ppc
1874    when compiling the compiler.
1875    
1876    The gc test will typically jump to a label at the end of the cluster,
1877    where there is another jump to an external cluster containing the actual
1878    code to invoke gc. This is to allow factoring of common gc invocation
1879    sequences. That is to say, we generate:
1880    
1881            f:
1882               testgc
1883               ja   L1      % jump if above to L1
1884    
1885            L1:
1886               jmp L2
1887    
1888    
1889    After jump chain elimination the 'ja L1' instructions is converted to
1890    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1891    up being implemented in their long form (if L2 is far away) using:
1892    
1893            jbe     L3      % jump if below or equal to L3
1894            jmp     L2
1895         L3:
1896            ...
1897    
1898    
1899    For large compilation units L2  may be far away.
1900    
1901    
1902    ----------------------------------------------------------------------
1903    Name: Matthias Blume
1904    Date: 2002/03/11 13:30:00 EST
1905    Tag: blume-20020311-mltreeeval
1906    Description:
1907    
1908    A functor parameter was missing.
1909    
1910    ----------------------------------------------------------------------
1911    Name: Allen Leung
1912    Date: 2002/03/11 10:30:00 EST
1913    Tag: leunga-20020311-runtime-string0
1914    Description:
1915    
1916       The representation of the empty string now points to a
1917    legal null terminated C string instead of unit.  It is now possible
1918    to convert an ML string into C string with InlineT.CharVector.getData.
1919    This compiles into one single machine instruction.
1920    
1921    ----------------------------------------------------------------------
1922    Name: Allen Leung
1923    Date: 2002/03/10 23:55:00 EST
1924    Tag: leunga-20020310-x86-call
1925    Description:
1926    
1927       Added machine generation for CALL instruction (relative displacement mode)
1928    
1929    ----------------------------------------------------------------------
1930    Name: Matthias Blume
1931    Date: 2002/03/08 16:05:00
1932    Tag: blume-20020308-entrypoints
1933    Description:
1934    
1935    Version number bumped to 110.39.1.  NEW BOOTFILES!
1936    
1937    Entrypoints: non-zero offset into a code object where execution should begin.
1938    
1939    - Added the notion of an entrypoint to CodeObj.
1940    - Added reading/writing of entrypoint info to Binfile.
1941    - Made runtime system bootloader aware of entrypoints.
1942    - Use the address of the label of the first function given to mlriscGen
1943      as the entrypoint.  This address is currently always 0, but it will
1944      not be 0 once we turn on block placement.
1945    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
1946      with entry points) from mlriscGen.
1947    
1948    ----------------------------------------------------------------------
1949    Name: Allen Leung
1950    Date: 2002/03/07 20:45:00 EST
1951    Tag: leunga-20020307-x86-cmov
1952    Description:
1953    
1954       Bug fixes for CMOVcc on x86.
1955    
1956       1. Added machine code generation for CMOVcc
1957       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1958       3. CMOVcc cannot have an immediate operand as argument.
1959    
1960    ----------------------------------------------------------------------
1961    Name: Matthias Blume
1962    Date: 2002/03/07 16:15:00 EST
1963    Tag: blume-20020307-controls
1964    Description:
1965    
1966    This is a very large but mostly boring patch which makes (almost)
1967    every tuneable compiler knob (i.e., pretty much everything under
1968    Control.* plus a few other things) configurable via both the command
1969    line and environment variables in the style CM did its configuration
1970    until now.
1971    
1972    Try starting sml with '-h' (or, if you are brave, '-H')
1973    
1974    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1975    implements the underlying generic mechanism.
1976    
1977    The interface to some of the existing such facilities has changed somewhat.
1978    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1979    (The getFoo interface is still there for backward-compatibility, but its
1980    use is deprecated.)
1981    
1982    The ml-build script passes -Cxxx=yyy command-line arguments through so
1983    that one can now twiddle the compiler settings when using this "batch"
1984    compiler.
1985    
1986    TODO items:
1987    
1988    We should go through and throw out all controls that are no longer
1989    connected to anything.  Moreover, we should go through and provide
1990    meaningful (and correct!) documentation strings for those controls
1991    that still are connected.
1992    
1993    Currently, multiple calls to Controls.new are accepted (only the first
1994    has any effect).  Eventually we should make sure that every control
1995    is being made (via Controls.new) exactly once.  Future access can then
1996    be done using Controls.acc.
1997    
1998    Finally, it would probably be a good idea to use the getter-setter
1999    interface to controls rather than ref cells.  For the time being, both
2000    styles are provided by the Controls module, but getter-setter pairs are
2001    better if thread-safety is of any concern because they can be wrapped.
2002    
2003    *****************************************
2004    
2005    One bug fix: The function blockPlacement in three of the MLRISC
2006    backpatch files used to be hard-wired to one of two possibilities at
2007    link time (according to the value of the placementFlag).  But (I
2008    think) it should rather sense the flag every time.
2009    
2010    *****************************************
2011    
2012    Other assorted changes (by other people who did not supply a HISTORY entry):
2013    
2014    1. the cross-module inliner now works much better (Monnier)
2015    2. representation of weights, frequencies, and probabilities in MLRISC
2016       changed in preparation of using those for weighted block placement
2017       (Reppy, George)
2018    
2019    ----------------------------------------------------------------------
2020    Name: Lal George
2021    Date: 2002/03/07 14:44:24 EST 2002
2022    Tag: george-20020307-weighted-block-placement
2023    
2024    Tested the weighted block placement optimization on all architectures
2025    (except the hppa) using AMPL to generate the block and edge frequencies.
2026    Changes were required in the machine properties to correctly
2027    categorize trap instructions. There is an MLRISC flag
2028    "weighted-block-placement" that can be used to enable weighted block
2029    placement, but this will be ineffective without block/edge
2030    frequencies (coming soon).
2031    
2032    
2033    ----------------------------------------------------------------------
2034    Name: Lal George
2035    Date: 2002/03/05 17:24:48 EST
2036    Tag: george-20020305-linkage-cluster
2037    
2038    In order to support the block placement optimization, a new cluster
2039    is generated as the very first cluster (called the linkage cluster).
2040    It contains a single jump to the 'real' entry point for the compilation
2041    unit. Block placement has no effect on the linkage cluster itself, but
2042    all the other clusters  have full freedom in the manner in which they
2043    reorder blocks or functions.
2044    
2045    On the x86 the typical linkage code that is generated is:
2046       ----------------------
2047            .align 2
2048       L0:
2049            addl    $L1-L0, 72(%esp)
2050            jmp     L1
2051    
2052    
2053            .align  2
2054       L1:
2055       ----------------------
2056    
2057    72(%esp) is the memory location for the stdlink register. This
2058    must contain the address of the CPS function being called. In the
2059    above example, it contains the address of  L0; before
2060    calling L1 (the real entry point for the compilation unit), it
2061    must contain the address for L1, and hence
2062    
2063            addl $L1-L0, 72(%esp)
2064    
2065    I have tested this on all architectures except the hppa.The increase
2066    in code size is of course negligible
2067    
2068    ----------------------------------------------------------------------
2069    Name: Allen Leung
2070    Date: 2002/03/03 13:20:00 EST
2071    Tag: leunga-20020303-mlrisc-tools
2072    
2073      Added #[ ... ] expressions to mlrisc tools
2074    
2075    ----------------------------------------------------------------------
2076    Name: Matthias Blume
2077    Date: 2002/02/27 12:29:00 EST
2078    Tag: blume-20020227-cdebug
2079    Description:
2080    
2081    - made types in structure C and C_Debug to be equal
2082    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
2083    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
2084    
2085    ----------------------------------------------------------------------
2086    Name: Matthias Blume
2087    Date: 2002/02/26 12:00:00 EST
2088    Tag: blume-20020226-ffi
2089    Description:
2090    
2091    1. Fixed a minor bug in CM's "noweb" tool:
2092       If numbering is turned off, then truly don't number (i.e., do not
2093       supply the -L option to noweb).  The previous behavior was to supply
2094       -L'' -- which caused noweb to use the "default" line numbering scheme.
2095       Thanks to Chris Richards for pointing this out (and supplying the fix).
2096    
2097    2. Once again, I reworked some aspects of the FFI:
2098    
2099       A. The incomplete/complete type business:
2100    
2101       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
2102         gone!
2103       - ML types representing an incomplete type are now *equal* to
2104         ML types representing their corresponding complete types (just like
2105         in C).  This is still safe because ml-nlffigen will not generate
2106         RTTI for incomplete types, nor will it generate functions that
2107         require access to such RTTI.   But when ML code generated from both
2108         incomplete and complete versions of the C type meet, the ML types
2109         are trivially interoperable.
2110    
2111         NOTE:  These changes restore the full generality of the translation
2112         (which was previously lost when I eliminated functorization)!
2113    
2114       B. Enum types:
2115    
2116       - Structure C now has a type constructor "enum" that is similar to
2117         how the "su" constructor works.  However, "enum" is not a phantom
2118         type because each "T enum" has values (and is isomorphic to
2119         MLRep.Signed.int).
2120       - There are generic access operations for enum objects (using
2121         MLRep.Signed.int).
2122       - ml-nlffigen will generate a structure E_foo for each "enum foo".
2123         * The structure contains the definition of type "mlrep" (the ML-side
2124         representation type of the enum).  Normally, mlrep is the same
2125         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
2126         then mlrep will be defined as a datatype -- thus facilitating
2127         pattern matching on mlrep values.
2128         ("-ec" will be suppressed if there are duplicate values in an
2129          enumeration.)
2130         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
2131         will be generated for each C enum constant xxx.
2132         * Conversion functions m2i and i2m convert between mlrep and
2133         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
2134         * Coversion functions c and ml convert between mlrep and "tag enum".
2135         * Access functions (get/set) fetch and store mlrep values.
2136       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
2137         enumerations are merged into one single enumeration represented by
2138         structure E_'.
2139    
2140    ----------------------------------------------------------------------
2141    Name: Allen Leung
2142    Date: 2002/02/25 04:45:00 EST
2143    Tag: leunga-20020225-cps-spill
2144    
2145    This is a new implementation of the CPS spill phase.
2146    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
2147    In case of problems, replace it with the old file spill.sml
2148    
2149    The current compiler runs into some serious performance problems when
2150    constructing a large record.  This can happen when we try to compile a
2151    structure with many items.  Even a very simple structure like the following
2152    makes the compiler slow down.
2153    
2154        structure Foo = struct
2155           val x_1 = 0w1 : Word32.int
2156           val x_2 = 0w2 : Word32.int
2157           val x_3 = 0w3 : Word32.int
2158           ...
2159           val x_N = 0wN : Word32.int
2160        end
2161    
2162    The following table shows the compile time, from N=1000 to N=4000,
2163    with the old compiler:
2164    
2165    N
2166    1000   CPS 100 spill                           0.04u  0.00s  0.00g
2167           MLRISC ra                               0.06u  0.00s  0.05g
2168              (spills = 0 reloads = 0)
2169           TOTAL                                   0.63u  0.07s  0.21g
2170    
2171    1100   CPS 100 spill                           8.25u  0.32s  0.64g
2172           MLRISC ra                               5.68u  0.59s  3.93g
2173              (spills = 0 reloads = 0)
2174           TOTAL                                   14.71u  0.99s  4.81g
2175    
2176    1500   CPS 100 spill                           58.55u  2.34s  1.74g
2177           MLRISC ra                               5.54u  0.65s  3.91g
2178              (spills = 543 reloads = 1082)
2179           TOTAL                                   65.40u  3.13s  6.00g
2180    
2181    2000   CPS 100 spill                           126.69u  4.84s  3.08g
2182           MLRISC ra                               0.80u  0.10s  0.55g
2183              (spills = 42 reloads = 84)
2184           TOTAL                                   129.42u  5.10s  4.13g
2185    
2186    3000   CPS 100 spill                           675.59u  19.03s  11.64g
2187           MLRISC ra                               2.69u  0.27s  1.38g
2188              (spills = 62 reloads = 124)
2189           TOTAL                                   682.48u  19.61s  13.99g
2190    
2191    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
2192           MLRISC ra                               4.96u  0.27s  2.72g
2193              (spills = 85 reloads = 170)
2194           TOTAL                                   2375.26u  57.21s  48.00g
2195    
2196    As you can see the old cps spill module suffers from some serious
2197    performance problem.  But since I cannot decipher the old code fully,
2198    instead of patching the problems up, I'm reimplementing it
2199    with a different algorithm.  The new code is more modular,
2200    smaller when compiled, and substantially faster
2201    (O(n log n) time and O(n) space).  Timing of the new spill module:
2202    
2203    4000  CPS 100 spill                           0.02u  0.00s  0.00g
2204          MLRISC ra                               0.25u  0.02s  0.15g
2205             (spills=1 reloads=3)
2206          TOTAL                                   7.74u  0.34s  1.62g
2207    
2208    Implementation details:
2209    
2210    As far as I can tell, the purpose of the CPS spill module is to make sure the
2211    number of live variables at any program point (the bandwidth)
2212    does not exceed a certain limit, which is determined by the
2213    size of the spill area.
2214    
2215    When the bandwidth is too large, we decrease the register pressure by
2216    packing live variables into spill records.  How we achieve this is
2217    completely different than what we did in the old code.
2218    
2219    First, there is something about the MLRiscGen code generator
2220    that we should be aware of:
2221    
2222    o MLRiscGen performs code motion!
2223    
2224       In particular, it will move floating point computations and
2225       address computations involving only the heap pointer to
2226       their use sites (if there is only a single use).
2227       What this means is that if we have a CPS record construction
2228       statement
2229    
2230           RECORD(k,vl,w,e)
2231    
2232       we should never count the new record address w as live if w
2233       has only one use (which is often the case).
2234    
2235       We should do something similar to floating point, but the transformation
2236       there is much more complex, so I won't deal with that.
2237    
2238    Secondly, there are now two new cps primops at our disposal:
2239    
2240     1. rawrecord of record_kind option
2241        This pure operator allocates some uninitialized storage from the heap.
2242        There are two forms:
2243    
2244         rawrecord NONE [INT n]  allocates a tagless record of length n
2245         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2246                                     and initializes the tag.
2247    
2248     2. rawupdate of cty
2249          rawupdate cty (v,i,x)
2250          Assigns to x to the ith component of record v.
2251          The storelist is not updated.
2252    
2253    We use these new primops for both spilling and increment record construction.
2254    
2255     1. Spilling.
2256    
2257        This is implemented with a linear scan algorithm (but generalized
2258        to trees).  The algorithm will create a single spill record at the
2259        beginning of the cps function and use rawupdate to spill to it,
2260        and SELECT or SELp to reload from it.  So both spills and reloads
2261        are fine-grain operations.  In contrast, in the old algorithm
2262        "spills" have to be bundled together in records.
2263    
2264        Ideally, we should sink the spill record construction to where
2265        it is needed.  We can even split the spill record into multiple ones
2266        at the places where they are needed.  But CPS is not a good
2267        representation for global code motion, so I'll keep it simple and
2268        am not attempting this.
2269    
2270     2. Incremental record construction (aka record splitting).
2271    
2272        Long records with many component values which are simulatenously live
2273        (recall that single use record addresses are not considered to
2274         be live) are constructed with rawrecord and rawupdate.
2275        We allocate space on the heap with rawrecord first, then gradually
2276        fill it in with rawupdate.  This is the technique suggested to me
2277        by Matthias.
2278    
2279        Some restrictions on when this is applicable:
2280        1. It is not a VECTOR record.  The code generator currently does not handle
2281           this case. VECTOR record uses double indirection like arrays.
2282        2. All the record component values are defined in the same "basic block"
2283           as the record constructor.  This is to prevent speculative
2284           record construction.
2285    
2286    ----------------------------------------------------------------------
2287    Name: Allen Leung
2288    Date: 2002/02/22 01:02:00 EST
2289    Tag: leunga-20020222-mlrisc-tools
2290    
2291    Minor bug fixes in the parser and rewriter
2292    
2293    ----------------------------------------------------------------------
2294    Name: Allen Leung
2295    Date: 2002/02/21 20:20:00 EST
2296    Tag: leunga-20020221-peephole
2297    
2298    Regenerated the peephole files.  Some contained typos in the specification
2299    and some didn't compile because of pretty printing bugs in the old version
2300    of 'nowhere'.
2301    
2302    ----------------------------------------------------------------------
2303    Name: Allen Leung
2304    Date: 2002/02/19 20:20:00 EST
2305    Tag: leunga-20020219-mlrisc-tools
2306    Description:
2307    
2308       Minor bug fixes to the mlrisc-tools library:
2309    
2310       1.  Fixed up parsing colon suffixed keywords
2311       2.  Added the ability to shut the error messages up
2312       3.  Reimplemented the pretty printer and fixed up/improved
2313           the pretty printing of handle and -> types.
2314       4.  Fixed up generation of literal symbols in the nowhere tool.
2315       5.  Added some SML keywords to to sml.sty
2316    
2317    ----------------------------------------------------------------------
2318    Name: Matthias Blume
2319    Date: 2002/02/19 16:20:00 EST
2320    Tag: blume-20020219-cmffi
2321    Description:
2322    
2323    A wild mix of changes, some minor, some major:
2324    
2325    * All C FFI-related libraries are now anchored under $c:
2326        $/c.cm      --> $c/c.cm
2327        $/c-int.cm  --> $c/internals/c-int.cm
2328        $/memory.cm --> $c/memory/memory.cm
2329    
2330    * "make" tool (in CM) now treats its argument pathname slightly
2331      differently:
2332        1. If the native expansion is an absolute name, then before invoking
2333           the "make" command on it, CM will apply OS.Path.mkRelative
2334           (with relativeTo = OS.FileSys.getDir()) to it.
2335        2. The argument will be passed through to subsequent phases of CM
2336           processing without "going native".  In particular, if the argument
2337           was an anchored path, then "make" will not lose track of that anchor.
2338    
2339    * Compiler backends now "know" their respective C calling conventions
2340      instead of having to be told about it by ml-nlffigen.  This relieves
2341      ml-nlffigen from one of its burdens.
2342    
2343    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
2344    
2345    * Export C_DEBUG and C_Debug from $c/c.cm.
2346    
2347    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
2348      subtyping relationship between incomplete pointers and their complete
2349      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
2350      with the convention of instantiating 'o with ('t, 'c) obj whenever
2351      the pointer target type is complete.  In the incomplete case, 'o
2352      will be instantiated with some "'c iobj" -- a type obtained by
2353      using one of the functors PointerToIncompleteType or PointerToCompleteType.
2354    
2355      Operations that work on both incomplete and complete pointer types are
2356      typed as taking an 'o ptr while operations that require the target to
2357      be known are typed as taking some ('t, 'c) obj ptr.
2358    
2359      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
2360      where void is an eqtype without any values.  This makes it possible
2361      to work on voidptr values using functions meant to operate on light
2362      incomplete pointers.
2363    
2364    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
2365      been vastly simplified.
2366    
2367    ----------------------------------------------------------------------
2368    Name: Matthias Blume
2369    Date: 2002/02/19 10:48:00 EST
2370    Tag: blume-20020219-pqfix
2371    Description:
2372    
2373    Applied Chris Okasaki's bug fix for priority queues.
2374    
2375    ----------------------------------------------------------------------
2376    Name: Matthias Blume
2377    Date: 2002/02/15 17:05:00
2378    Tag: Release_110_39
2379    Description:
2380    
2381    Last-minute retagging is becoming a tradition... :-(
2382    
2383    This is the working release 110.39.
2384    
2385    ----------------------------------------------------------------------
2386    Name: Matthias Blume
2387    Date: 2002/02/15 16:00:00 EST
2388    Tag: Release_110_39-orig
2389    Description:
2390    
2391    Working release 110.39.  New bootfiles.
2392    
2393    (Update: There was a small bug in the installer so it wouldn't work
2394    with all shells.  So I retagged. -Matthias)
2395    
2396    ----------------------------------------------------------------------
2397    Name: Matthias Blume
2398    Date: 2002/02/15 14:17:00 EST
2399    Tag: blume-20020215-showbindings
2400    Description:
2401    
2402    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
2403    the latter can be useful for exploring what bindings are available at
2404    the interactive prompt.  (The first function returns only the list
2405    of symbols that are really bound, the second prints those but also the
2406    ones that CM's autoloading mechanism knows about.)
2407    
2408    ----------------------------------------------------------------------
2409    Name: Matthias Blume
2410    Date: 2002/02/15 12:08:00 EST
2411    Tag: blume-20020215-iptrs
2412    Description:
2413    
2414    Two improvements to ml-nlffigen:
2415    
2416      1. Write files only if they do not exist or if their current contents
2417         do not coincide with what's being written.  (That is, avoid messing
2418         with the time stamps unless absolutely necessary.)
2419    
2420      2. Implement a "repository" mechanism for generated files related
2421         to "incomplete pointer types".   See the README file for details.
2422    
2423    ----------------------------------------------------------------------
2424    Name: Matthias Blume
2425    Date: 2002/02/14 11:50:00 EST
2426    Tag: blume-20020214-quote
2427    Description:
2428    
2429    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
2430    because of the new and improved tag generation scheme.  (Thanks to Allen
2431    Leung for pointing it out.)
2432    
2433    ----------------------------------------------------------------------
2434    Name: Lal George
2435    Date: 2002/02/14 09:55:27 EST 2002
2436    Tag: george-20020214-isabelle-bug
2437    Description:
2438    
2439    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2440    of Isabelle on the x86.
2441    
2442    From Allen:
2443    -----------
2444     I've found the problem:
2445    
2446         in ra-core.sml, I use the counter "blocked" to keep track of the
2447         true number of elements in the freeze queue.  When the counter goes
2448         to zero, I skip examining the queue.  But I've messed up the
2449         bookkeeping in combine():
2450    
2451             else ();
2452             case !ucol of
2453               PSEUDO => (if !cntv > 0 then
2454                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2455                                        ^^^^^^^^^^^^^^^^^^^^^^^
2456                      moveu := mergeMoveList(!movev, !moveu)
2457                     )
2458                  else ();
2459    
2460         combine() is called to coalesce two nodes u and v.
2461         I think I was thinking that if the move counts of u and v are both
2462         greater than zero then after they are coalesced then one node is
2463         removed from the freeze queue.  Apparently I was thinking that
2464         both u and v are of low degree, but that's clearly not necessarily true.
2465    
2466    
2467    02/12/2002:
2468        Here's the patch.  HOL now compiles.
2469    
2470        I don't know how this impact on performance (compile
2471        time or runtime).  This bug caused the RA (especially on the x86)
2472        to go thru the potential spill phase when there are still nodes on the
2473        freeze queue.
2474    
2475    
2476    
2477    
2478    ----------------------------------------------------------------------
2479    Name: Matthias Blume
2480    Date: 2002/02/13 22:40:00 EST
2481    Tag: blume-20020213-fptr-rtti
2482    Description:
2483    
2484    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2485    updates.
2486    
2487    ----------------------------------------------------------------------
2488    Name: Matthias Blume
2489    Date: 2002/02/13 16:41:00 EST
2490    Tag: blume-20020213-cmlpq
2491    Description:
2492    
2493    Added new priority queue export symbols (which have just been added to
2494    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2495    not compile and the installer would choke.)
2496    
2497    ----------------------------------------------------------------------
2498    Name: Matthias Blume
2499    Date: 2002/02/13 16:15:00 EST
2500    Tag: blume-20020213-various
2501    Description:
2502    
2503    1. More tweaks to ml-nlffigen:
2504    
2505       - better internal datastructures (resulting in slight speedup)
2506       - "-match" option requires exact match
2507       - "localized" gensym counters (untagged structs/unions nested within
2508         other structs/unions or within typedefs get a fresh counter; their
2509         tag will be prefixed by a concatenation of their parents' tags)
2510       - bug fixes (related to calculation of transitive closure of types
2511         to be included in the output)
2512    
2513    2. Minor Basis updates:
2514    
2515       - added implementations for List.collate and Option.app
2516    
2517    ----------------------------------------------------------------------
2518    Name: Matthias Blume
2519    Date: 2002/02/11 15:55:00 EST
2520    Tag: blume-20020211-gensym
2521    Description:
2522    
2523    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2524    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2525    names (ML structure names that correspond to unnamed C structs, unions,
2526    and enums), so that separate runs of ml-nlffigen do not clash.
2527    
2528    ----------------------------------------------------------------------
2529    Name: Matthias Blume
2530    Date: 2002/02/11 12:05:00 EST
2531    Tag: blume-20020211-gensml
2532    Description:
2533    
2534    A quick fix for a problem with GenSML (in the pgraph-util library):
2535    Make generation of toplevel "local" optional.  (Strictly speaking,
2536    signature definitions within "local" are not legal SML.)
2537    
2538    Other than that: updates to INSTALL and cm/TODO.
2539    
2540    ----------------------------------------------------------------------
2541    Name: Matthias Blume
2542    Date: 2002/02/08 15:00:00 EST
2543    Tag: blume-20020208-uniquepid
2544    Description:
2545    
2546    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2547    
2548    1. The installer (config/install.sh) has gotten smarter:
2549    
2550         - Configuration options are a bit easier to specify now
2551           (in config/targets).
2552         - Bug in recognizing .tar.bz2 files fixed.
2553         - Installer automatically resolves dependencies between
2554           configuration options (e.g., if you ask for eXene, you will
2555           also get cml -- regardless whether you asked for it or not).
2556         - Installer can run in "quieter mode" by setting the environment
2557           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2558           "completely silent", though.
2559         - Build HashCons library as part of smlnj-lib.
2560    
2561    2. A new scheme for assigning persistent identifiers to compilation
2562       units (and, by extension, to types etc.) has been put into place.
2563       This fixes a long-standing bug where types and even dynamic values
2564       can get internally confused, thereby compromising type safety
2565       (abstraction) and dynamic correctness.  See
2566    
2567         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2568    
2569       for an example of how things could go wrong until now.
2570    
2571       The downside of the new scheme is that pids are not quite as
2572       persistent as they used to be: CM will generate a fresh pid
2573       for every compilation unit that it thinks it sees for the first
2574       time.  That means that if you compile starting from a clean, fresh
2575       source tree at two different times, you end up with different
2576       binaries.
2577    
2578       Cutoff recompilation, however, has not been compromised because
2579       CM keeps pid information in special caches between runs.
2580    
2581    ----------------------------------------------------------------------
2582    Name: Lal George
2583    Date: 2002/02/07 15:34:13 EST 2002
2584    Tag: <none>
2585    Description:
2586    
2587    Compilers that generate assembly code may produce  global labels
2588    whose value is resolved at link time. The various peephole optimization
2589    modules did not take this in account.
2590    
2591    TODO. The Labels.addrOf function should really return an option
2592    type so that clients are forced to deal with this issue, rather
2593    than an exception being raised.
2594    
2595    ----------------------------------------------------------------------
2596    Name: Lal George
2597    Date: 2002/02/06 13:55:02 EST
2598    Tag: george-20020206-ra-breakup
2599    Description:
2600    
2601    1. A bug fix from Allen.
2602    
2603        A typo causes extra fstp %st(0)'s to be generated at compensation
2604        edges, which might cause stack underflow traps at runtime.  This
2605        occurs in fft where there are extraneous fstps right before the 'into'
2606        trap instruction (in this case they are harmless since none of the
2607        integers overflow.)
2608    
2609    2. Pulled out various utility modules that were embedded in the modules
2610       of the register allocator. I need these modules for other purposes, but
2611       they are not complete enough to put into a library (just yet).
2612    ----------------------------------------------------------------------
2613    Name: Matthias Blume
2614    Date: 2002/01/31 16:05:00 EST
2615    Tag: blume-20020131-sparc-ccalls
2616    Description:
2617    
2618    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2619       of extra stack space by mistake.  Fixed.
2620    
2621    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2622    
2623    ----------------------------------------------------------------------
2624    Name: Allen Leung
2625    Date: 2002/01/30
2626    Tag: leunga-20020130-nowhere-bug-fix
2627    Description:
2628    
2629       MLRISC bug fixes:
2630       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2631       2. MachineInt.fromString was negating its value.
2632    
2633    ----------------------------------------------------------------------
2634    Name: Matthias Blume
2635    Date: 2002/01/29
2636    Tag: blume-20020129-INSTALL
2637    Description:
2638    
2639    - Added somewhat detailed installation instructions (file INSTALL).
2640    - Fixed curl-detection bug in config/install.sh.
2641    - It is now possible to select the URL getter using the URLGETTER
2642      environment variable:
2643    
2644          not set / "unknown"      --> automatic detection (script tries wget,
2645                                       curl, and lynx)
2646          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2647                                       how to properly invoke them)
2648          other                    --> use $URLGETTER directly, it must take
2649                                       precisely two command-line arguments
2650                                       (source URL and destination file name)
2651    
2652    ----------------------------------------------------------------------
2653    Name: Matthias Blume
2654    Date: 2002/01/28
2655    Tag: blume-20020128-sparc-ccalls
2656    Description:
2657    
2658    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2659    - Make use of the allocParam argument in sparc-c-calls.
2660    
2661    ----------------------------------------------------------------------
2662    Name: Matthias Blume
2663    Date: 2002/01/28
2664    Tag: blume-20020128-allocParam
2665    Description:
2666    
2667    John Reppy:  Changes c-calls API to accept client-callback for
2668    allocating extra stack space.
2669    me: Corresponding changes to mlriscGen (using a dummy argument that
2670        does not change the current behavior).
2671    
2672    ----------------------------------------------------------------------
2673    Name: Matthias Blume
2674    Date: 2002/01/28 12:00:00
2675    Tag: Release_110_38
2676    Description:
2677    
2678    This time for real!!!
2679    
2680    ----------------------------------------------------------------------
2681    Name: Matthias Blume
2682    Date: 2002/01/28 10:56:00 EST
2683    Tag: blume-20020128-retraction
2684    Description:
2685    
2686    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
2687       with blume-Release_110_38-retracted.)
2688    
2689    1. Fixed a problem with incorrect rounding modes in real64.sml.
2690       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2691    
2692    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2693       and enums fixed.  The naming of corresponding ML identifiers should
2694       now be consistent again.
2695    
2696    ----------------------------------------------------------------------
2697    Name: Allen Leung
2698    Date: 2002/01/27
2699    Tag: leunga-20020127-nowhere
2700    Description:
2701    
2702       Added a target called nowhere in the configuration scripts.
2703       Enabling this will build the MLRISC 'nowhere' tool (for translating
2704       programs with where-clauses into legal SML code) during installation.
2705    
2706    ----------------------------------------------------------------------
2707    Name: Matthias Blume
2708    Date: 2002/01/25 21:27:00 EST
2709    Tag: blume-Release_110_38-retracted
2710    Description:
2711    
2712    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
2713    
2714    README will be added later.
2715    
2716    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
2717    (Release_110_38) removed.  Reason: Last-minute bug fixes.
2718    
2719    ----------------------------------------------------------------------
2720    Name: Matthias Blume
2721    Date: 2002/01/25
2722    Tag: blume-20020125-ffi
2723    Description:
2724    
2725    A large number of tweaks and improvements to ml-nlffi-lib and
2726    ml-nlffigen:
2727    
2728       - ML represenation types have been streamlined
2729       - getter and setter functions work with concrete values, not abstract
2730         ones where possible
2731       - ml-nlffigen command line more flexible (see README file there)
2732       - some bugs have been fixed (hopefully)
2733    
2734    ----------------------------------------------------------------------
2735    Name: Lal George
2736    Date: 2002/01/24
2737    Tag: george-20020124-risc-ra-interface
2738    Description:
2739    
2740       There is a dramatic simplification in the interface to the
2741       register allocator for RISC architectures as a result of making
2742       parallel copy instructions explicit.
2743    
2744    ----------------------------------------------------------------------
2745    Name: Matthias Blume
2746    Date: 2002/01/22
2747    Tag: blume-20020122-x86-ccalls
2748    Description:
2749    
2750    Bug fix for c-calls on x86 (having to do with how char- and
2751    short-arguments are being handled).
2752    
2753    ----------------------------------------------------------------------
2754    Name: Matthias Blume
2755    Date: 2002/01/21
2756    Tag: blume-20020121-ff
2757    Description:
2758    
2759    Another day of fiddling with the FFI...
2760    
2761    1. Bug fix/workaround:  CKIT does not complain about negative array
2762       dimensions, so ml-nlffigen has to guard itself against this possibility.
2763       (Otherwise a negative dimension would send it into an infinite loop.)
2764    
2765    2. Some of the abstract types (light objects, light pointers, most "base"
2766       types) in structure C are now eqtypes.
2767    
2768    3. Added constructors and test functions for NULL function pointers.
2769    
2770    ----------------------------------------------------------------------
2771    Name: Matthias Blume
2772    Date: 2002/01/18
2773    Tag: blume-20020118-ready-for-new-release
2774    Description:
2775    
2776    Made config/srcarchiveurl point to a new place.  (Will provide boot
2777    files shortly.)
2778    
2779    Maybe we christen this to be 110.38?
2780    
2781    ----------------------------------------------------------------------
2782    Name: Matthias Blume
2783    Date: 2002/01/18
2784    Tag: blume-20020118-more-ffifiddle
2785    Description:
2786    
2787    Today's FFI fiddling:
2788    
2789      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
2790        C.Set.* that use concrete (MLRep.*) arguments and results instead
2791        of abstract ones.
2792    
2793      - Provided word-style bit operations etc. for "int" representation
2794        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
2795        and Long.
2796    
2797    ----------------------------------------------------------------------
2798    Name: Matthias Blume
2799    Date: 2002/01/18
2800    Tag: blume-20020118-use-x86-fp
2801    Description:
2802    
2803    Now that x86-fast-fp seems to be working, I turned it back on again
2804    by default.  (Seems to work fine now, even with the FFI.)
2805    
2806    Other than that, I added some documentation about the FFI to
2807    src/ml-nlffigen/README and updated the FFI test examples in
2808    src/ml-nlffi-lib/Tests/*.
2809    
2810    ----------------------------------------------------------------------
2811    Name: Allen Leung
2812    Date: 2002/01/17
2813    Tag: leunga-20020117-x86-fast-fp-call
2814    Description:
2815    
2816       1. Fixed a problem with handling return fp values when x86's fast fp
2817          mode is turned on.
2818    
2819       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
2820          of %f32.
2821    
2822       3. Added a constructor INT32lit to the ast of MLRISC tools.
2823    
2824    ----------------------------------------------------------------------
2825    Name: Matthias Blume
2826    Date: 2002/01/16
2827    Tag: blume-20020116-ffifiddle
2828    Description:
2829    
2830    More fiddling with the FFI interface:
2831    
2832     - Make constness 'c instead of rw wherever possible.  This eliminates
2833       the need for certain explicit coercions.  (However, due to ML's
2834       value polymorphism, there will still be many cases where explicit
2835       coercions are necessary.  Phantom types are not the whole answer
2836       to modeling a subtyping relationship in ML.)
2837    
2838     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
2839    
2840     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
2841       clumsy workaround for varargs problem.)
2842    
2843    ----------------------------------------------------------------------
2844    Name: Lal George
2845    Date: 2002/01/15
2846    Tag: <none>
2847    Description:
2848    
2849    1. Since COPY instructions are no longer native to the architecture,
2850       a generic functor can be used to implement the expandCopies function.
2851    
2852    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
2853       TEXT segment.
2854    
2855    ----------------------------------------------------------------------
2856    Name: Matthias Blume
2857    Date: 2002/01/15
2858    Tag: blume-20020115-ffiupdates
2859    Description:
2860    
2861    1. Fix for bug resulting in single-precision float values being returned
2862       incorrectly from FFI calls.
2863    
2864    2. Small modifications to C FFI API:
2865    
2866        - memory-allocation routines return straight objects (no options)
2867          and raise an exception in out-of-memory situations
2868        - unsafe extensions to cast between function pointers and pointers
2869          from/to ints
2870        - added structure C_Debug as an alternative to structure C where
2871          pointer-dereferencing (|*| and |*!) always check for null-pointers
2872        - added open_lib' to DynLinkage;  open_lib' works like open_lib
2873          but also takes a (possibly empty) list of existing library handles
2874          that the current library depends on
2875    
2876    ----------------------------------------------------------------------
2877    Name: Matthias Blume
2878    Date: 2002/01/10
2879    Tag: blume-20020110-newffigen
2880    Description:
2881    
2882    1. Updates to portable graph code.
2883    
2884    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
2885       more scalable now so that even huge interfaces such as the one
2886       for GTK compile in finite time and space. :-)
2887       See src/ml-nlffigen/README for details on what's new.
2888    
2889    ----------------------------------------------------------------------
2890    Name: Lal George
2891    Date: 2001/01/09 14:31:35 EST 2002
2892    Tag: george-20011206-rm-native-copy
2893    Description:
2894    
2895            Removed the native COPY and FCOPY instructions
2896            from all the architectures and replaced it with the
2897            explicit COPY instruction from the previous commit.
2898    
2899            It is now possible to simplify many of the optimizations
2900            modules that manipulate copies. This has not been
2901            done in this change.
2902    
2903    ----------------------------------------------------------------------
2904    Name: Lal George
2905    Date: 2001/12/06 16:50:13 EST 2001
2906    Tag: george-20011206-mlrisc-instruction
2907    Description:
2908    
2909    Changed the representation of instructions from being fully abstract
2910    to being partially concrete. That is to say:
2911    
2912      from
2913            type instruction
2914    
2915      to
2916            type instr                              (* machine instruction *)
2917    
2918            datatype instruction =
2919                LIVE of {regs: C.cellset, spilled: C.cellset}
2920              | KILL of {regs: C.cellset, spilled: C.cellset}
2921              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
2922              | ANNOTATION of {i: instruction, a: Annotations.annotation}
2923              | INSTR of instr
2924    
2925    This makes the handling of certain special instructions that appear on
2926    all architectures easier and uniform.
2927    
2928    LIVE and KILL say that a list of registers are live or killed at the
2929    program point where they appear. No spill code is generated when an
2930    element of the 'regs' field is spilled, but the register is moved to
2931    the 'spilled' (which is present, more for debugging than anything else).
2932    
2933    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
2934    We used to generate:
2935    
2936            DEFFREG f1
2937            f1 := f2 + f3
2938            trapb
2939    
2940    but now generate:
2941    
2942            f1 := f2 + f3
2943            trapb
2944            LIVE {regs=[f1,f2,f3], spilled=[]}
2945    
2946    Furthermore, the DEFFREG (hack) required that all floating point instruction
2947    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
2948    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
2949    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
2950    this architecture).
2951    
2952    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
2953    all the architectures. This will result in further simplification of the
2954    register allocator that must be aware of them for coalescing purposes, and
2955    will also simplify certain aspects of the machine description that provides
2956    callbacks related to parallel copies.
2957    
2958    ANNOTATION should be obvious, and now INSTR represents the honest to God
2959    machine instruction set!
2960    
2961    The <arch>/instructions/<arch>Instr.sml files define certain utility
2962    functions for making porting easier -- essentially converting upper case
2963    to lower case. All machine instructions (of type instr) are in upper case,
2964    and the lower case form generates an MLRISC instruction. For example on
2965    the alpha we have:
2966    
2967      datatype instr =
2968         LDA of {r:cell, b:cell, d:operand}
2969       | ...
2970    
2971      val lda : {r:cell, b:cell, d:operand} -> instruction
2972        ...
2973    
2974    where lda is just (INSTR o LDA), etc.
2975    
2976    ----------------------------------------------------------------------
2977    Name: Matthias Blume
2978    Date: 2001/11/22 21:40:00 EST
2979    Tag: Release_110_37
2980    Description:
2981    
2982    Release 110.37.  This time for real.
2983    
2984    ----------------------------------------------------------------------
2985    Name: Matthias Blume
2986    Date: 2001/11/21 16:35:00 EST
2987    Tag: blume-20011121-foot-in-mouth
2988    Description:
2989    
2990    Removed the "Release_110_37" tag because of a serious bug.
2991    This will be re-tagged once the bug is fixed.
2992    
2993    ----------------------------------------------------------------------
2994    Name: Matthias Blume
2995    Date: 2001/11/21 16:14:00 EST
2996    Tag: blume-20011121-forgottenfile
2997    Description:
2998    
2999    Forgot to add a file.  (Just a .tex-file -- part of
3000    the CM manual source.)
3001    
3002    ----------------------------------------------------------------------
3003    Name: Matthias Blume
3004    Date: 2001/11/21 16:10:00 EST
3005    Tag: blume-20011121-invalid_110_37
3006    Description:
3007    
3008    Note: I removed the original tag "Release_110_37" from this commit
3009          because we found a serious bug in all non-x86 backends.
3010          - Matthias
3011    
3012    1. Modifications to the SML/NJ code generator and to the runtime system
3013       so that code object name strings are directly inserted into code
3014       objects at code generation time.  The only business the runtime system
3015       has with this is now to read the name strings on occasions.
3016       (The encoding of the name string has also changed somewhat.)
3017    
3018    2. CM now implements a simple "set calculus" for specifying export lists.
3019       In particular, it is now possible to refer to the export lists of
3020       other libraries/groups/sources and form unions as well as differences.
3021       See the latest CM manual for details.
3022    
3023    3. An separate notion of "proxy" libraries has again be eliminated from
3024       CM's model.  (Proxy libraries are now simply a special case of using
3025       the export list calculus.)
3026    
3027    4. Some of the existing libraries now take advantage of the new set
3028       calculus.
3029       (Notice that not all libraries have been converted because some
3030       of the existing .cm-files are supposed to be backward compatible
3031       with 110.0.x.)
3032    
3033    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
3034       "General.exnMessage"!  The former relies on a certain hook to be
3035       initialized, and that often does not happen in the stand-alone case.)
3036    
3037    ----------------------------------------------------------------------
3038    Name: Lal George
3039    Date: 2001/11/21  13:56:18 EST
3040    Tag: george-2001121-pseudo-ops
3041    Description:
3042    
3043      Implemented a complete redesign of MLRISC pseudo-ops. Now there
3044      ought to never be any question of incompatabilities with
3045      pseudo-op syntax expected by host assemblers.
3046    
3047      For now, only modules supporting GAS syntax are implemented
3048      but more should follow, such as MASM, and vendor assembler
3049      syntax, e.g. IBM as, Sun as, etc.
3050    
3051    ----------------------------------------------------------------------
3052    Name: Matthias Blume
3053    Date: 2001/11/14 11:52:00 EST
3054    Tag: blume-20011114-srcname
3055    Description:
3056    
3057    1. Routed the name of the current source file to mlriscgen where it
3058       should be directly emitted into the code object.  (This last part
3059       is yet to be done.)
3060    
3061    2. Some cleanup of the pgraph code to make it match the proposal that
3062       I put out the other day.  (The proposal notwithstanding, things are
3063       still in flux here.)
3064    
3065    ----------------------------------------------------------------------
3066    Name: Lal George
3067    Date: 2001/11/14 09:44:04 EST
3068    Tag:
3069    Description:
3070    
3071      Fix for a backpatching bug reported by Allen.
3072    
3073      Because the boundary between short and long span-dependent
3074      instructions is +/- 128, there are an astounding number of
3075      span-dependent instructions whose size is over estimated.
3076    
3077      Allen came up with the idea of letting the size of span
3078      dependent instructions be non-monotonic, for a maxIter
3079      number of times, after which the size must be monotonically
3080      increasing.
3081    
3082      This table shows the number of span-dependent instructions
3083      whose size was over-estimated as a function of maxIter, for the
3084      file Parse/parse/ml.grm.sml:
3085    
3086         maxIter            # of instructions:
3087            10                      687
3088            20                      438
3089            30                      198
3090            40                        0
3091    
3092      In compiling the compiler, there is no significant difference in
3093      compilation speed between maxIter=10 and maxIter=40. Actually,
3094      my measurements showed that maxIter=40 was a tad faster than
3095      maxIter=10! Also 96% of the  files in the compiler reach a fix
3096      point within 13 iterations, so fixing maxIter at 40, while high,
3097      is okay.
3098    
3099    ----------------------------------------------------------------------
3100    Name: Matthias Blume
3101    Date: 2001/10/31 15:25:00 EST
3102    Tag: blume-20011031-pgraph
3103    Description:
3104    
3105    CKIT:
3106    * Changed the "Function" constructor of type Ast.ctype to carry optional
3107      argument identifiers.
3108    * Changed the return type of TypeUtil.getFunction accordingly.
3109    * Type equality ignores the argument names.
3110    * TypeUtil.composite tries to preserve argument names but gives up quickly
3111      if there is a mismatch.
3112    
3113    installation script:
3114    * attempts to use "curl" if available (unless "wget" is available as well)
3115    
3116    CM:
3117    * has an experimental implementation of "portable graphs" which I will
3118      soon propose as an implementation-independent library format
3119    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
3120    
3121    NLFFI-LIB:
3122    * some cleanup (all cosmetic)
3123    
3124    NLFFIGEN:
3125    * temporarily disabled the mechanism that suppresses ML output for
3126      C definitions whose identifiers start with an underscore character
3127    * generate val bindings for enum constants
3128    * user can request that only one style (light or heavy) is being used;
3129      default is to use both (command-line arguments: -heavy and -light)
3130    * fixed bug in handling of function types involving incomplete pointers
3131    * generate ML entry points that take record arguments (i.e., using
3132      named arguments) for C functions that have a prototype with named
3133      arguments
3134      (see changes to CKIT)
3135    
3136    ----------------------------------------------------------------------
3137    Name: Allen Leung
3138    Date: 2001/10/27 20:34:00 EDT
3139    Tag: leunga-20011027-x86-fast-fp-call
3140    Description:
3141    
3142       Fixed the bug described in blume-20010920-slowfp.
3143    
3144       The fix involves
3145          1. generating FCOPYs in FSTP in ia32-svid
3146          2. marking a CALL with the appropriate annotation
3147    
3148    ----------------------------------------------------------------------
3149    Name: Matthias Blume
3150    Date: 2001/10/16 11:32:00 EDT
3151    Tag: blume-20011016-netbsd
3152    Description:
3153    
3154    Underscore patch from Chris Richards (fixing problem with compiling
3155    runtime system under recent NetBSD).
3156    
3157    ----------------------------------------------------------------------
3158    Name: Allen Leung
3159    Date: 2001/10/12 17:18:32 EDT 2001
3160    Tag: leung-20011012-x86-printflowgraph
3161    Description:
3162    
3163    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
3164    
3165    ----------------------------------------------------------------------
3166    Name: Lal George
3167    Date: 2001/10/11 23:51:34 EDT
3168    Tag: george-20011011-too-many-instrs
3169    Description:
3170    
3171    The representation of a program point never expected to see more
3172    than 65536 instructions in a basic block!
3173    
3174    ----------------------------------------------------------------------
3175    Name: Lal George
3176    Date: 2001/10/09 09:41:37 EDT
3177    Tag: george-20011008-mlrisc-labels
3178    Description:
3179    
3180    Changed the machine description files to support printing of
3181    local and global labels in assembly code, based on host assembler
3182    conventions.
3183    
3184    ----------------------------------------------------------------------
3185    Name: Matthias Blume
3186    Date: 2001/09/25 15:25:00 EDT
3187    Tag: blume-20010925-exninfo
3188    Description:
3189    
3190    I provided a non-hook implementation of exnName (at the toplevel) and
3191    made the "dummy" implementation of exnMessage (at the toplevel) more
3192    useful: if nothing gets "hooked in", then at least you are going to
3193    see the exception name and a message indicating why you don't see more.
3194    
3195    [For the time being, programs that need exnMessage and want to use
3196    ml-build should either use General.exnMessage (strongly recommended) or
3197    refer to structure General at some other point so that CM sees a
3198    static dependency.]
3199    
3200    [Similar remarks go for "print" and "use":  If you want to use their
3201    functionality in stand-alone programs generated by ml-build, then use
3202    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
3203    
3204    ----------------------------------------------------------------------
3205    Name: Matthias Blume
3206    Date: 2001/09/20 17:28:00 EDT
3207    Tag: blume-20010920-slowfp
3208    Description:
3209    
3210    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
3211    
3212    ----------------------------------------------------------------------
3213    Name: Matthias Blume
3214    Date: 2001/09/20 17:20:00 EDT
3215    Tag: blume-20010920-canonicalpaths
3216    Description:
3217    
3218    0. Updated the BOOT file (something that I forgot to do earlier).
3219    
3220    1. Small internal change to CM so that it avoids "/../" in filenames
3221       as much as possible (but only where it is safe).
3222    
3223    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
3224       that arguments that contain delimiters are passed through correctly.
3225       This change also means that all "special" arguments of the form
3226       @SMLxxx... must come first.
3227    
3228    3. Changed install script to put relative anchor names for tool commands
3229       into pathconfig.
3230    
3231    ----------------------------------------------------------------------
3232    Name: Matthias Blume
3233    Date: 2001/09/18 15:35:00 EDT
3234    Tag: blume-20010918-readme11036
3235    Description:
3236    
3237    Added README files.
3238    
3239    ----------------------------------------------------------------------
3240    Name: Matthias Blume
3241    Date: 2001/09/18 11:45:00 EDT
3242    Tag: Release_110_36 (retag)
3243    Description:
3244    
3245    Fixed mistake in config/preloads. Retagged as 110.36.
3246    
3247    ----------------------------------------------------------------------
3248    Name: Matthias Blume
3249    Date: 2001/09/18 09:40:00 EDT
3250    Tag: Release_110_36_orig (tag changed)
3251    Description:
3252    
3253    New version (110.36).  New bootfiles.
3254    
3255    ----------------------------------------------------------------------
3256    Name: Matthias Blume
3257    Date: 2001/09/14 16:15:00 EDT
3258    Tag: blume-20010914-x86fastfp
3259    Description:
3260    
3261    John committed some changes that Allen made, in particular a (hopefully)
3262    correctly working version of the x86-fp module.
3263    
3264    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
3265    flag to "true".  Everything seems to compile to a fixpoint ok, and
3266    "mandelbrot" speeds up by about 15%.
3267    
3268    ----------------------------------------------------------------------
3269    Name: Matthias Blume
3270    Date: 2001/09/13 11:20:00 EDT
3271    Tag: blume-20010913-minimal
3272    Description:
3273    
3274    1. Stefan Monnier's patch to fix a miscompilation problem that
3275       was brought to light by John Reppy's work on Moby.
3276    
3277    2. Implemented a minimal "structure Compiler" that contains just
3278       "version" and "architecture".  The minimal version will be
3279       available when the full version is not.  This is for backward-
3280       compatibility with code that wants to test Compiler.version.
3281    
3282    ----------------------------------------------------------------------
3283    Name: Matthias Blume
3284    Date: 2001/08/28 14:03:00 EDT
3285    Tag: blume-20010828-ml-lex
3286    Description:
3287    
3288    Fix for bug 1581, received from Neophytos Michael.
3289    
3290    ----------------------------------------------------------------------
3291    Name: Matthias Blume
3292    Date: 2001/08/27 11:20:00 EDT
3293    Tag: blume-20010827-readme11035
3294    Description:
3295    
3296    Fleshed out the README file for 110.35.
3297    
3298    ----------------------------------------------------------------------
3299    Name: Matthias Blume
3300    Date: 2001/08/24 17:10:00 EDT
3301    Tag: Release_110_35
3302    Description:
3303    
3304    New version number (110.35).  New bootfiles.
3305    
3306    ----------------------------------------------------------------------
3307    Name: Lal George
3308    Date: 2001/08/24 13:47:18 EDT 2001
3309    Tag: george-20010824-MLRISC-graphs
3310    Description:
3311    
3312     removed clusters from MLRISC completely and replaced with graphs.
3313    
3314    ----------------------------------------------------------------------
3315    Name: Matthias Blume
3316    Date: 2001/08/23 17:50:00 EDT
3317    Tag: blume-20010823-toplevel
3318    Description:
3319    
3320    - some reorganization of the code that implements various kinds of
3321      environments in the compiler (static, dynamic, symbolic, combined)
3322    - re-implemented the EnvRef module so that evalStream works properly
3323      (if the stream contains references to "use", "CM.make", etc.)
3324    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
3325    
3326    ----------------------------------------------------------------------
3327    Name: Matthias Blume
3328    Date: 2001/08/20 15:50 EDT
3329    Tag: blume20010820-slipup
3330    Description:
3331    
3332    I forgot to commit a few files.  Here they are...
3333    
3334    ----------------------------------------------------------------------
3335    Name: Matthias Blume
3336    Date: 2001/08/20 15:35:00 EDT
3337    Tag: blume-20010820-debugprof
3338    Description:
3339    
3340    !!!! NEW BOOTFILES !!!!
3341    
3342    This is another round of reorganizing the compiler sources.  This
3343    time the main goal was to factor out all the "instrumentation"
3344    passes (for profiling and backtracing) into their own library.
3345    The difficulty was to do it in such a way that it does not depend
3346    on elaborate.cm but only on elabdata.cm.
3347    
3348    Therefore there have been further changes to both elaborate.cm and
3349    elabdata.cm -- more "generic" things have been moved from the former
3350    to the latter.  As a result, I was forced to split the assignment
3351    of numbers indicating "primtyc"s into two portions: SML-generic and
3352    SML/NJ-specific.  Since it would have been awkward to maintain,
3353    I bit the bullet and actually _changed_ the mapping between these
3354    numbers and primtycs.  The bottom line of this is that you need
3355    a new set of bin- and bootfiles.
3356    
3357    I have built new bootfiles for all architectures, so doing a fresh
3358    checkout and config/install.sh should be all you need.
3359    
3360    The newly created library's name is
3361    
3362        $smlnj/viscomp/debugprof.cm
3363    
3364    and its sources live under
3365    
3366        src/compiler/DebugProf
3367    
3368    ----------------------------------------------------------------------
3369    Name: Matthias Blume
3370    Date: 2001/08/15 17:15:00 EDT
3371    Tag: blume-20010815-compreorg
3372    Description:
3373    
3374    This is a first cut at reorganizing the CM libraries that make up the
3375    core of the compiler.  The idea is to separate out pieces that could
3376    be used independently by tools, e.g., the parser, the typechecker, etc.
3377    
3378    The current status is a step in this direction, but it is not quite
3379    satisfactory yet.  Expect more changes in the future.
3380    
3381    Here is the current (new) organization...
3382    
3383        What used to be $smlnj/viscomp/core.cm is now divided into
3384        six CM libraries:
3385    
3386             $smlnj/viscomp/basics.cm
3387                           /parser.cm
3388                           /elabdata.cm
3389                           /elaborate.cm
3390                           /execute.cm
3391                           /core.cm
3392    
3393        The CM files for these libraries live under src/system/smlnj/viscomp.
3394        All these libraries are proxy libraries that contain precisely
3395        one CM library component.  Here are the locations of the components
3396        (all within the src/compiler tree):
3397    
3398             Basics/basics.cm
3399             Parse/parser.cm
3400             ElabData/elabdata.cm
3401             Elaborator/elaborate.cm
3402             Execution/execute.cm
3403             core.cm
3404    
3405         [This organization is the same that has been used already
3406         for a while for the architecture-specific parts of the visible
3407         compiler and for the old version of core.cm.]
3408    
3409         As you will notice, many source files have been moved from their
3410         respective original locations to a new home in one of the above
3411         subtrees.
3412    
3413         The division of labor between the new libraries is the following:
3414    
3415             basics.cm:
3416                - Simple, basic definitions that pertain to many (or all) of
3417                  the other libraries.
3418             parser.cm:
3419                - The SML parser, producing output of type Ast.dec.
3420                - The type family for Ast is also defined and exported here.
3421             elabdata.cm:
3422                - The datatypes that describe input and output of the elaborator.
3423                  This includes types, absyn, and static environments.
3424             elaborator.cm:
3425                - The SML/NJ type checker and elaborator.
3426                  This maps an Ast.dec (with a given static environment) to
3427                  an Absyn.dec (with a new static environment).
3428                - This libraries implements certain modules that used to be
3429                  structures as functors (to remove dependencies on FLINT).
3430             execute.cm:
3431                - Everything having to do with executing binary code objects.
3432                - Dynamic environments.
3433             core.cm:
3434                - SML/NJ-specific instantiations of the elaborator and MLRISC.
3435                - Top-level modules.
3436                - FLINT (this should eventually become its own library)
3437    
3438    Notes:
3439    
3440    I am not 100% happy with the way I separated the elaborator (and its
3441    data structures) from FLINT.  Two instances of the same problem:
3442    
3443        1. Data structures contain certain fields that carry FLINT-specific
3444           information.  I hacked around this using exn and the property list
3445           module from smlnj-lib.  But the fact that there are middle-end
3446           specific fields around at all is a bit annoying.
3447    
3448        2. The elaborator calculates certain FLINT-related information.  I tried
3449           to make this as abstract as I could using functorization, but, again,
3450           the fact that the elaborator has to perform calculations on behalf
3451           of the middle-end at all is not nice.
3452    
3453        3. Having to used exn and property lists is unfortunate because it
3454           weakens type checking.  The other alternative (parameterizing
3455           nearly *everything*) is not appealing, though.
3456    
3457    I removed the "rebinding =" warning hack because due to the new organization
3458    it was awkward to maintain it.  As a result, the compiler now issues some of
3459    these warnings when compiling init.cmi during bootstrap compilation. On
3460    the plus side, you also get a warning when you do, for example:
3461       val op = = Int32.+
3462    which was not the case up to now.
3463    
3464    I placed "assign" and "deref" into the _Core structure so that the
3465    code that deals with the "lazy" keyword can find them there.  This
3466    removes the need for having access to the primitive environment
3467    during elaboration.
3468    
3469    ----------------------------------------------------------------------
3470    Name: Matthias Blume
3471    Date: 2001/08/13
3472    Tag: blume-20010813-closures
3473    Description:
3474    
3475    This fix was sent to us by Zhong Shao.  It is supposed to improve the
3476    performance of certain loops by avoiding needless closure allocation.
3477    
3478    ----------------------------------------------------------------------
3479    Name: Lal George
3480    Date: 2001/07/31 10:03:23 EDT 2001
3481    Tag: george-20010731-x86-fmalloc
3482    Description: Fixed bug in x86 calls
3483    
3484        There was a bug where call instructions would mysteriously
3485        vanish. The call instruction had to be one that returned
3486        a floating point value.
3487    
3488    ----------------------------------------------------------------------
3489    Name: Lal George
3490    Date: 2001/07/19 16:36:29 EDT 2001
3491    Tag: george-20010719-simple-cells
3492    Description:
3493    
3494    I have dramatically simplified the interface for CELLS in MLRISC.
3495    
3496    In summary, the cells interface is broken up into three parts:
3497    
3498      1. CellsBasis : CELLS_BASIS
3499    
3500            CellsBasis is a top level structure and common for all
3501            architectures.  it contains the definitions of basic datatypes
3502            and utility  functions over these types.
3503    
3504      2. functor Cells() : CELLS
3505    
3506            Cells generates an interface for CELLS that incorporates the
3507            specific resources on the target architecture, such as the
3508            presence of special register classes, their number and size,
3509            and various useful substructures.
3510    
3511      3. <ARCH>CELLS
3512    
3513            e.g. SparcCells: SPARCCELLS
3514    
3515            <ARCH>CELLS usually contains additional bindings for special
3516            registers  on the architecture, such as:
3517    
3518                    val r0 : cell           (* register zero *)
3519                    val y : cell            (* Y register *)
3520                    val psr : cell          (* processor status register *)
3521                    ...
3522    
3523            The structure returned by applying the Cells functor is opened
3524            in this interface.
3525    
3526    The main implication of all this is that the datatypes for cells is
3527    split between CellsBasis and CELLS -- a fairly simple change for user
3528    code.
3529    
3530    In the old scheme the CELLS interface had a definitional binding of
3531    the form:
3532    
3533            signature CELLS = sig
3534    
3535               structure CellsBasis = CellsBasis
3536    
3537               ...
3538    
3539            end
3540    
3541    With all the sharing constraints that goes on in MLRISC, this old
3542    design  quickly leads to errors such as:
3543    
3544            "structure definition spec inside of sharing ... "
3545    
3546    
3547    and appears to require an unacceptable amount of sharing and where
3548    constraint hackery.
3549    
3550    I think this error message (the interaction of definitional specs and
3551    sharing) requires more explanation on our web page.
3552    
3553    ----------------------------------------------------------------------
3554    Name: Matthias Blume
3555    Date: 2001/07/19 15:00:00 EDT
3556    Tag: blume-20010719-libreorg
3557    Description:
3558    
3559    This update puts together a fairly extensive but straightforward change
3560    to the way the libraries that implement the interactive system are
3561    organized:
3562    
3563       The biggest change is the elimination of structure Compiler.  As a
3564       replacement for this structure, there is now a CM library
3565       (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
3566       that exports all the substructures of the original structure Compiler
3567       directly.  So instead of saying Compiler.Foo.bar one now simply
3568       says Foo.bar.  (The CM libraries actually export a collection of
3569       structures that is richer than the collection of substructures of
3570       structure Compiler.)
3571    
3572       To make the transition smooth, there is a separate library called
3573       $smlnj/compiler/compiler.cm which puts together and exports the
3574       original structure Compiler (or at least something very close to it).
3575    
3576       There are five members of the original structure Compiler
3577       that are not exported directly but which instead became members
3578       of a new structure Backend (described by signature BACKEND).  These are:
3579       structure Profile (: PROFILE), structure Compile (: COMPILE), structure
3580       Interact (: INTERACT), structure Machine (: MACHINE), and val
3581       architecture (: string).
3582    
3583       Structure Compiler.Version has become structure CompilerVersion.
3584    
3585       Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
3586       by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
3587       or x86, respectively.
3588       Each of these exports the same frontend structures that
3589       $smlnj/compiler.cm exports.  But they do not have a structure Backend
3590       and instead export some structure <Arch>Backend where <Arch> is Alpha32,
3591       Hppa, PPC, Sparc, or X86, respectively.
3592    
3593       Library $smlnj/compiler/all.cm exports the union of the exports of
3594       $smlnj/compiler/<arch>.cm
3595    
3596       There are no structures <Arch>Compiler anymore, use
3597       $smlnj/compiler/<arch>.cm instead.
3598    
3599       Library host-compiler-0.cm is gone.  Instead, the internal library
3600       that instantiates CM is now called cm0.cm.  Selection of the host
3601       compiler (backend) is no longer done here but. (Responsibility for it
3602       now lies with $smlnj/compiler/current.cm.  This seems to be more
3603       logical.)
3604    
3605       Many individual files have been moved or renamed.  Some files have
3606       been split into multiple files, and some "dead" files have been deleted.
3607    
3608    Aside from these changes to library organization, there are also changes
3609    to the way the code itself is organized:
3610    
3611       Structure Binfile has been re-implemented in such a way that it no
3612       longer needs any knowledge of the compiler.  It exclusively deals
3613       with the details of binfile layout.  It no longer invokes the
3614       compiler (for the purpose of creating new prospective binfile
3615       content), and it no longer has any knowledge of how to interpret
3616       pickles.
3617    
3618       Structure Compile (: COMPILE) has been stripped down to the bare
3619       essentials of compilation.  It no longer deals with linking/execution.
3620       The interface has been cleaned up considerably.
3621    
3622       Utility routines for dealing with linking and execution have been
3623       moved into their own substructures.
3624    
3625       (The ultimate goal of these changes is to provide a light-weight
3626       binfile loader/linker (at least for, e.g., stable libraries) that
3627       does not require CM or the compiler to be present.)
3628    
3629    CM documentation has been updated to reflect the changes to library
3630    organization.
3631    
3632    ----------------------------------------------------------------------
3633    Name: Matthias Blume
3634    Date: 2001/07/10 17:30:00 EDT
3635    Tag: Release_110_34
3636    Description:
3637    
3638    Minor tweak to 110.34 (re-tagged):
3639    
3640      - README.html file added to CVS repository
3641      - runtime compiles properly under FreeBSD 3.X and 4.X
3642    
3643    ----------------------------------------------------------------------
3644    Name: Matthias Blume
3645    Date: 2001/07/10 17:30:00 EDT
3646    Tag: Release_110_34
3647    Description:
3648    
3649    New version number (110.34). New bootfiles.
3650    
3651    ----------------------------------------------------------------------
3652    Name: Matthias Blume
3653    Date: 2001/07/09 16:00:00 EDT
3654    Tag: blume-20010709-more-varargs
3655    Description:
3656    
3657    I changed the handling of varargs in ml-nlffigen again:
3658    The ellipsis ... will now simply be ignored (with an accompanying warning).
3659    
3660    The immediate effect is that you can actually call a varargs function
3661    from ML -- but you can't actually supply any arguments beyond the ones
3662    specified explicitly.  (For example, you can call printf with its format
3663    string, but you cannot pass additional arguments.)
3664    
3665    This behavior is only marginally more useful than the one before, but
3666    it has the advantage that a function or, more importantly, a function
3667    type never gets dropped on the floor, thus avoiding follow-up problems with
3668    other types that refer to the offending one.
3669    
3670    ----------------------------------------------------------------------
3671    Name: Matthias Blume
3672    Date: 2001/07/09 11:25:00 EDT
3673    Tag: blume-20010709-varargs
3674    Description:
3675    
3676    1. ckit-lib.cm now exports structure Error
3677    2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
3678       with a warning accompanied by a source location.  Moreover, it
3679       merely skips the offending function or type and proceeds with the
3680       rest of its work.u  As a result, one can safely feed C code containing
3681       "..." to ml-nlffigen.
3682    3. There are some internal improvements to CM, providing slightly
3683       more general string substitutions in the tools subsystem.
3684    
3685    ----------------------------------------------------------------------
3686    Name: Matthias Blume
3687    Date: 2001/06/27 15:10:00 EDT
3688    Tag: blume-20010627-concur
3689    Description:
3690    
3691    Fixed a small bug in CM's handling of parallel compilation.
3692    (You could observe the bug by Control-C-interrupting an ordinary
3693    CMB.make or CM.stabilize and then attaching some compile servers.
3694    The result was that all of a sudden the previously interrupted
3695    compilation would continue on its own.  This was because of
3696    an over-optimization: CM did not bother to clean out certain queues
3697    when no servers were attached "anyway", resulting in the contents
3698    of these queues to grab control when new servers did get attached.)
3699    
3700    There is also another minor update to the CM manual.
3701    
3702    ----------------------------------------------------------------------
3703    Name: Matthias Blume
3704    Date: 2001/06/26 16:15:00 EDT
3705    Tag: blume-20010626-cmdoc
3706    Description:
3707    
3708    Minor typo fixed in CM manual (syntax diagram for libraries).
3709    
3710    ----------------------------------------------------------------------
3711    Name: Matthias Blume
3712    Date: 2001/06/25 22:55:00 EDT
3713    Tag: blume-20010625-x86pc
3714    Description:
3715    
3716    Fixed a nasty bug in the X86 assembly code that caused signal
3717    handlers to fail (crash) randomly.
3718    
3719    ----------------------------------------------------------------------
3720    Name: Matthias Blume
3721    Date: 2001/06/25 12:05:00 EDT
3722    Tag: blume-20010625-nlffigen
3723    Description:
3724    
3725    This update fixes a number of minor bugs in ml-nlffigen as reported by
3726    Nick Carter <nbc@andrew.cmu.edu>.
3727    
3728      1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
3729      2. Default names for generated files are now derived from the name of
3730         the C file *without its directory*.  In particular, this causes generated
3731         files to be placed locally even if the C file is in some system directory.
3732      3. Default names for generated signatures and structures are also derived
3733         from the C file name without its directory.  This avoids silly things
3734         like "structure GL/GL".
3735         (Other silly names are still possible because ml-nlffigen does not do
3736          a thorough check of whether generated names are legal ML identifiers.
3737          When in doubt, use command line arguments to force particular names.)
3738    
3739    ----------------------------------------------------------------------
3740    Name: Matthias Blume
3741    Date: 2001/06/21 12:25:00 EDT
3742    Tag: blume-20010621-eXene
3743    Description:
3744    
3745    eXene now compiles and (sort of) works again.
3746    
3747    The library name (for version > 110.33) is $/eXene.cm.
3748    
3749    I also added an new example in src/eXene/examples/nbody.  See the
3750    README file there for details.
3751    
3752    ----------------------------------------------------------------------
3753    Name: Matthias Blume
3754    Date: 2001/06/20 16:40:00 EDT
3755    Tag: blume-20010620-cml
3756    Description:
3757    
3758    CML now compiles and works again.
3759    
3760    Libraries (for version > 110.33):
3761    
3762      $cml/cml.cm            Main CML library.
3763      $cml/basis.cm          CML's version of $/basis.cm.
3764      $cml/cml-internal.cm   Internal helper library.
3765      $cml/core-cml.cm       Internal helper library.
3766      $cml-lib/trace-cml.cm  Tracing facility.
3767      $cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
3768    
3769    The installer (config/install.sh) has been taught how to properly
3770    install this stuff.
3771    
3772    ----------------------------------------------------------------------
3773    Name: Matthias Blume
3774    Date: 2001/06/19 17:55:00 EDT
3775    Tag: blume-20010619-instantiate
3776    Description:
3777    
3778    This un-breaks the fix for bug 1432.
3779    (The bug was originally fixed in 110.9 but I broke it again some
3780    time after that.)
3781    
3782    ----------------------------------------------------------------------
3783    Name: Matthias Blume
3784    Date: 2001/06/19 17:25:00 EDT
3785    Tag: blume-20010619-signals
3786    Description:
3787    
3788    This should (hopefully) fix the long-standing signal handling bug.
3789    (The runtime system was constructing a continuation record with an
3790    incorrect descriptor which would cause the GC to drop data on the floor...)
3791    
3792    ----------------------------------------------------------------------
3793    Name: Matthias Blume
3794    Date: 2001/06/15 15:05:00 EDT
3795    Tag: blume-20010615-moresparc
3796    Description:
3797    
3798    Here is a short late-hour update related to Sparc c-calls:
3799    
3800     -- made handling of double-word arguments a bit smarter
3801    
3802     -- instruction selection phase tries to collapse certain clumsily
3803        constructed ML-Trees; typical example:
3804    
3805            ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
3806    
3807        This currently has no further impact on SML/NJ since mlriscGen does
3808        not seem to generate such patterns in the first place, and c-calls
3809        (which did generate them in the beginning) has meanwhile been fixed
3810        so as to avoid them as well.
3811    
3812    ----------------------------------------------------------------------
3813    Name: Matthias Blume
3814    Date: 2001/06/15 15:05:00 EDT
3815    Tag: blume-20010615-sparc
3816    Description:
3817    
3818    The purpose of this update is to provide an implementation of NLFFI
3819    on Sparc machines.
3820    
3821    Here are the changes in detail:
3822    
3823       * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
3824       the Sparc implementation of the c-calls API.
3825       * The Sparc backend of SML/NJ has been modified to uniformely use %fp
3826       for accessing the ML frame.  Thus, we have a real frame pointer and
3827       can freely modify %sp without need for an omit-frame-ptr phase.
3828       The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
3829       this case.
3830       * ml-nlffigen has been taught to produce code for different architectures
3831       and calling conventions.
3832       * In a way similar to what was done in the x86 case, the Sparc
3833       backend uses its own specific extension to mltree.  (For example,
3834       it needs to be able to generate UNIMP instructions which are part
3835       of the calling convention.)
3836       * ml-nlffi-lib was reorganized to make it more modular (in particular,
3837       to make it easier to plug in new machine- and os-dependent parts).
3838    
3839    There are some other fairly unrelated bug fixes and cleanups as well:
3840