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 761, Sat Dec 23 05:37:37 2000 UTC revision 1506, Thu Jun 17 22:19:15 2004 UTC
# Line 8  Line 8 
8  The form of an entry should be:  The form of an entry should be:
9    
10  Name:  Name:
11  Date:  Date: yyyy/mm/dd
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14    
15    ----------------------------------------------------------------------
16    Name: Matthias Blume (blume (at) tti - c (dot) org)
17    Date: 2004/06/17 17:20:00 CDT
18    Tag: blume-20040617-timer-unix
19    Description:
20    
21    Changed the interface of structures Timer and Unix to match the most
22    recent Basis spec.
23    
24    In the case of Unix there still seems to be an open/weird issue:
25    
26       The {text,bin}{In,Out}streamOf functions are supposed to create
27       fresh streams whenever they are called -- as opposed to have them
28       return the same stream every time.  This design is supposed to
29       prevent space leaks caused by proc values hanging on to streams.
30    
31       The reap function, on the other hand, is supposed to close the
32       streams.  This cannot be done without having a handle on the
33       stream in proc after all...
34    
35    I took the liberty to implement the following stopgap solution:
36    
37      The proc value hangs on to the most recently created stream(s).
38      Reap closes those.  If either or both of the two streams hadn't
39      been created at all yet, then reap will close the corresponding
40      file descriptors directly.
41    
42    PS: I don't understand the original space leak argument anymore.  If
43    a proc hangs on to the imperative stream, then I/O operations on those
44    will advance the state of the cached stream and avoid the space leak.
45    
46    ----------------------------------------------------------------------
47    Name: Matthias Blume (blume (at) tti - c (dot) org)
48    Date: 2004/05/28 16:45:00 CDT
49    Tag: blume-20040528-basis
50    Description:
51    
52    Added signature PACK_REAL and exported functor PrimIO.
53    
54    ----------------------------------------------------------------------
55    Name: Matthias Blume (blume (at) tti - c (dot) org)
56    Date: 2004/05/25 16:00:00 CDT
57    Tag: blume-20040525-group-owner
58    Description:
59    
60    CM now ignores (but still accepts) the "owner" information in group
61    descriptions.  The owner of a group is its next enclosing
62    library. Each group must have a unique owner.  (There is a virtual
63    "toplevel" library that owns groups which are not nested within a real
64    library.)  Previously, each group had to explicitly declare its owner,
65    and CM would check that such a declaration is correct.  The new scheme
66    is to have CM check that for each group there is precisely one owning
67    library.
68    
69    The advantage of the new scheme is that the programmer no longer needs
70    to maintain the somewhat annoying owner information.  The downside is
71    that CM cannot enforce the ownership rule across multiple runs of
72    CM.make.  Fortunately, enclosing the same group in two different
73    libraries A and B which are not part of the same program does not
74    cause real problems.
75    
76    ----------------------------------------------------------------------
77    Name: Matthias Blume (blume (at) tti - c (dot) org)
78    Date: 2004/05/20 16:00:00 CDT
79    Tag: blume-20040520-win32
80    Description:
81    
82    Made the win32 version work again.  (Strangely, a misplaced comma had
83    slipped into win32-process.c which prevented the runtime from being
84    compiled correctly.)
85    
86    Also, included a minor addition to ml-build.bat analogous to what was
87    done in blume-20040519-ml-build.
88    
89    ----------------------------------------------------------------------
90    Name: Matthias Blume (blume (at) tti - c (dot) org)
91    Date: 2004/05/19 22:10:00 CDT
92    Tag: blume-20040519-ml-build
93    Description:
94    
95    Arranged for ml-build to clean up after itself a little bit better.
96    The script generates a temporary SML source file and compiles it using
97    CM, so CM generates metadata (GUID, SKEL, objectfile) for it.  It now
98    gets rid of those at the end, so they don't accumulate under .cm.
99    
100    This required a minor change to install.sh because the name of the
101    metadata directory (default: .cm) is actually configurable at
102    installation time.
103    
104    ----------------------------------------------------------------------
105    Name: Matthias Blume (blume (at) tti - c (dot) org)
106    Date: 2004/05/18 15:50:00 CDT
107    Tag: blume-20040518-mkreader
108    Description:
109    
110    Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
111    implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
112    
113    ----------------------------------------------------------------------
114    Name: Matthias Blume (blume (at) tti - c (dot) org)
115    Date: 2004/05/11 14:35:00 CDT
116    Tag: blume-20040511-win32sock
117    Description:
118    
119    Added previously missing support for many socket-related functions
120    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
121    for the voluminous patch!
122    
123    (I have not tested this patch under win32 yet.)
124    
125    Here is David's e-mail:
126    
127    Hi,
128    
129    Attached to this email you find a diff against sml/nj 110.45
130    that will enable socket support under Windows.
131    
132    To apply the patch (using unix or cygwin)
133    1) gunzip runtime.diff.gz
134    2) "cd" into "src/runtime" in the source tree of a fresh
135       110.45 installation.
136    3) patch -p 1 < [your/path/to]runtime.diff
137    
138    The code compiles fine but has NOT yet been extensively tested.
139    I only ran a few tests for basic socket client functionality
140    (which worked fine).  Especially the functions that use ioctl
141    are not tested at all and might not work (see below).
142    
143    I implemented this since we want to move to a newer version of sml/nj
144    but need socket support in order to use it.  This is the first time I
145    even had a look at the sml/nj source,  so please review my changes
146    before making this part of the distribution!  Here are a few issues
147    that I think might be better for someone to solve who is more
148    familiar with the sml/nj source (and socket programming):
149    
150    - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
151      exception since I could not figure out what the windows equivalent
152      of these functions is
153    
154    - In sockets-osdep.h there are a some #include statements that are
155      only used in a few files that include sockets-osdep.h
156    
157    - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
158      does not process its return value since I don't know how to
159      report an error upwards.
160    
161    - It would probably be good to have a call to WSACleanup() when
162      the library is unloaded (if there is such a possibility).
163      Otherwise I think Windows will take care of this automatically
164      when the process finishes.
165    
166    - I used ioctlsocket() as a replacement for ioctl() but I have
167      no idea if that is actually the proper replacement on Windows.
168    
169    - All these issues are marked in the code by "FIXME" comments.
170    
171    We use sml/nj extensively in our products and are quite happy
172    with it.  I hope this contribution will help you.
173    
174    Keep up the good work!
175    
176    David
177    
178    ----------------------------------------------------------------------
179    Name: Matthias Blume (blume (at) tti - c (dot) org)
180    Date: 2004/05/11 14:20:00 CDT
181    Tag: blume-20040511-installml
182    Description:
183    
184    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
185    the patch.)
186    
187    ----------------------------------------------------------------------
188    Name: Matthias Blume (blume (at) tti - c (dot) org)
189    Date: 2004/05/11 14:05:00 CDT
190    Tag: blume-20040511-nlffi-netbsd
191    Description:
192    
193    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
194    the patch.)
195    
196    ----------------------------------------------------------------------
197    Name: Matthias Blume (blume (at) tti - c (dot) org)
198    Date: 2004/05/11 12:05:00 CDT
199    Tag: blume-20040511-exports
200    Description:
201    
202    As per request by Adam Chlipala <adam@hcoop.net>, extended various
203    export lists in compiler-related .cm-files.
204    
205    ----------------------------------------------------------------------
206    Name: Matthias Blume (blume (at) tti - c (dot) org)
207    Date: 2004/05/11 11:35:00 CDT
208    Tag: blume-20040511-allsource
209    Description:
210    
211    The installer now honors the "src-smlnj" target again, although its meaning
212    has changed from "all sources required for the compiler" to "all sources
213    the installer knows about".  In other words, if you enable "src-smlnj"
214    in the "targets" file, then the installer will pull in sources for
215    everything.  (Notice that this refers to source code only.  Compiled
216    code is still only installed for modules that were requested explicitly
217    or which are required for other modules that were requested explicitly.)
218    
219    ----------------------------------------------------------------------
220    Name: Matthias Blume (blume (at) tti - c (dot) org)
221    Date: 2004/04/23 17:40:00 CDT
222    Tag: blume-20040423-ieee-scan
223    Description:
224    
225    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
226    in the exponent calculation we get INF or ZERO (depending on the mantissa
227    and the sign of the exponent).
228    
229    ----------------------------------------------------------------------
230    Name: Matthias Blume (blume (at) tti - c (dot) org)
231    Date: 2004/04/23 10:40:00 CDT
232    Tag: blume-20040423-ml-build
233    Description:
234    
235    The ml-build script now terminates with a non-0 status when something
236    goes wrong.
237    
238    ----------------------------------------------------------------------
239    Name: Matthias Blume (blume (at) tti - c (dot) org)
240    Date: 2004/04/22 16:35:00 CDT
241    Tag: blume-20040422-Option
242    Description:
243    
244    Made exception Option to be the same as exception Option.Option
245    (as it should be).
246    
247    ----------------------------------------------------------------------
248    Name: Allen Leung (leunga (at) reservoir (dot) com)
249    Date: 2004/03/19 14:40:00 EST
250    Tag: leunga-20040319-cygwin-nlffi
251    Description:
252    
253    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
254    of SML/NJ.  The problem is that
255    
256       lib = dlopen(NULL, ...)
257       f   = dlsym(lib, "malloc");
258    
259    does not work on Windows unless we explicitly export symbols
260    such as 'malloc' during linking.  We fixed this by explicitly
261    exporting the required symbols with the magic gcc incantation:
262    
263        -Wl,--export-all cygwin.def
264    
265    where cygwin.def is a file containing all the symbols that
266    we wish to export.
267    
268    I suspect this is a Windows problem and we'll have to
269    do the same (somehow with windows compilers) when
270    we build the native win32 version with the system
271    calls LoadLibrary/GetProcAddress.
272    
273    ----------------------------------------------------------------------
274    Name: Matthias Blume (blume (at) tti - c (dot) org)
275    Date: 2004/03/04 16:35:00 CST
276    Tag: blume-20040304-intinf-fmt
277    Description:
278    
279    Fixed problem with IntInf.fmt (sign would show up on the right instead
280    of on the left for BIN, OCT, and HEX).
281    
282    ----------------------------------------------------------------------
283    Name: Matthias Blume (blume (at) tti - c (dot) org)
284    Date: 2004/03/04 11:25:00 CST
285    Tag: blume-20040304-symlinks
286    Description:
287    
288    Fixed problem with installer script (unix only) where bin/ml-yacc and
289    friends pointed (via symlinks) to absolute locations instead of just
290    .run-sml.  This was reported by Vesa A Norrman.
291    
292    ----------------------------------------------------------------------
293    Name: Matthias Blume (blume (at) tti - c (dot) org)
294    Date: 2004/02/13 14:50:00 CST
295    Tag: Release_110_45
296    Description:
297    
298    New working version (110.45). New bootfiles.
299    
300    ----------------------------------------------------------------------
301    Name: Matthias Blume (blume (at) tti - c (dot) org)
302    Date: 2004/01/26 15:15:15 CST
303    Tag: blume-20040126-toplevel
304    Description:
305    
306    Improved handling of exceptions at the interactive toplevel.
307    
308    ----------------------------------------------------------------------
309    Name: Matthias Blume (blume (at) tti - c (dot) org)
310    Date: 2004/01/26 11:25:00
311    Tag: blume-20040126-app
312    Description:
313    
314    Type of top-level "app" corrected.
315    Added code for setting vp_limitPtrMask to Win32-specific runtime.
316    
317    ----------------------------------------------------------------------
318    Name: Matthias Blume (blume (at) tti - c (dot) org)
319    Date: 2003/11/18 17:10 CST
320    Tag: blume-20031118-basis-fiddle
321    Description:
322    
323    - changed Timer interface to what might become the spec
324    - POSIX_FLAGS -> BIT_FLAGS according to spec
325    - some other minor discrepancies wrt. spec eliminated
326    
327    ----------------------------------------------------------------------
328    Name: Matthias Blume (blume (at) tti - c (dot) org)
329    Date: 2003/11/06 12:00:00 CST
330    Tag: Release_110_44
331    Description:
332    
333    New working version (110.44). New bootfiles.
334    
335    ----------------------------------------------------------------------
336    Name: Matthias Blume (blume (at) tti - c (dot) org)
337    Date: 2003/11/04 11:50:00 CST
338    Tag: blume-20031104-move-libraries
339    Description:
340    
341    Eliminated the "dont_move_libraries" directive in config/targets.
342    (The mechanism was broken and could not be fixed easily.  Moreover,
343    there does not seem to be any reason not to move all libraries into
344    lib during installation.  I originally implemented this directive as a
345    backward-compatibility feature when I first introduced the new CM.
346    Now that things have been stable for a long time and going back to the
347    old CM is not an option, there is no reason to keep it around.)
348    
349    ----------------------------------------------------------------------
350    Name: Matthias Blume (blume (at) tti - c (dot) org)
351    Date: 2003/11/03 16:00:00 CST
352    Tag: blume-20031103-installdir
353    Description:
354    
355    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
356    Richards for pointing out the problem and providing the solution.)
357    
358    ----------------------------------------------------------------------
359    Name: Matthias Blume (blume (at) tti - c (dot) org)
360    Date: 2003/10/01 17:05:00 CDT
361    Tag: blume-20031001-lal-mlrisc
362    Description:
363    
364    MLRISC bug fix from Lal.
365    
366    ----------------------------------------------------------------------
367    Name: Matthias Blume (blume (at) tti - c (dot) org)
368    Date: 2003/09/30 16:10:00 CDT
369    Tag: blume-20030930-primio-bat
370    Description:
371    
372    1. Added openVector, nullRd, and nullWr to PRIM_IO.
373    2. Improved .bat files (for Win32 port) to make things work under Win95.
374       (thanks to Aaron S. Hawley for this one)
375    
376    ----------------------------------------------------------------------
377    Name: Matthias Blume (blume (at) tti - c (dot) org)
378    Date: 2003/09/26 16:05:00 CDT
379    Tag: blume-20030926-wrappriv
380    Description:
381    
382    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
383    
384    ----------------------------------------------------------------------
385    Name: Matthias Blume (blume (at) tti - c (dot) org)
386    Date: 2003/09/26 15:00:00 CDT
387    Tag: blume-20030926-110_43_3
388    Description:
389    
390    - additional cleanup
391    - version number bump, NEW BOOTFILES
392    
393    ----------------------------------------------------------------------
394    Name: Matthias Blume (blume (at) tti - c (dot) org)
395    Date: 2003/09/26 12:00:00 CDT
396    Tag: blume-20030926-ppautoload
397    Description:
398    
399    I modified the read-eval-print loop so that the autoloader gets
400    invoked whenever the prettyprinter tries to look up a symbol that
401    is not currently defined in the toplevel environment but which
402    appears in CM's autoload registry.  As a result, we see far fewer of
403    those ?.Foo.Bar.xxx names in the prettyprinter's output.
404    
405    In addition to this I tried to clean up some pieces of the Basis
406    implementation (e.g., Socket, Word8Array) in order to prevent other
407    instances of these ?.Foo.Bar.xxx names from being printed.
408    
409    The mechanism that picks names for types still needs some work, though.
410    (Right now it seems that if there is a type A.t which is defined to
411    be B.u, but B is unavailable at toplevel, then A.t gets printed as
412    "?.B.u" although the perhaps more sensible solution would be to use
413    "A.t" in this case.  In other words, the prettyprinter should follow
414    a chain of DEFtycs not farther than there are corresponding toplevel
415    names in the current environment.)
416    
417    ----------------------------------------------------------------------
418    Name: Matthias Blume (blume (at) tti - c (dot) org)
419    Date: 2003/09/24 16:31:00 CDT
420    Tag: blume-20030924-installer
421    Description:
422    
423    Another installer tweak:  All the ML code for the installer is now
424    compiled during CMB.make and put into a little library called
425    $smlnj/installer.cm.  The installation then simply invokes
426    
427       sml -m $smlnj/installer.cm
428    
429    and everything happens automagically.
430    
431    Win32:  ML code senses value of environment variable SMLNJ_HOME.
432    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
433            and BINDIR.
434    
435    The new scheme guarantees that the ML code responsible for the installation
436    is in sync with the APIs of the main system.  Also, the installer is
437    somewhat faster because the installer script is precompiled.
438    
439    ----------------------------------------------------------------------
440    Name: Matthias Blume (blume (at) tti - c (dot) org)
441    Date: 2003/09/24 15:35:00 CDT
442    Tag: blume-20030924-synsock
443    Description:
444    
445    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
446    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
447    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
448    implementations of non-blocking operations from CML's Socket
449    structure.
450    
451    ----------------------------------------------------------------------
452    Name: Matthias Blume (blume (at) tti - c (dot) org)
453    Date: 2003/09/24 15:10:05 CDT
454    Tag: blume-20030924-sockets
455    Description:
456    
457    1. Fixed SOCKET API and implementation to match Basis spec.
458       This required changing the internal representation of sockets to one
459       that remembers (for each socket file descriptor) whether it is currently
460       blocking or non-blocking.  This state is maintained lazily (i.e., a system
461       call is made only if the state actually needs to change).
462    
463    2. OS-specific details of sockets were moved into separate files, thus
464       making it possible to unify the bulk of the socket implementations
465       between Unix and Win32.
466    
467    3. CML's socket API changed accordingly.
468       (Note that we need to remove non-blocking functions from this API
469       since they are redundant in the case of CML!)
470    
471    4. CML's socket implementation now makes use of non-blocking functions
472       provided by Basis, thus removing all OS-dependent code from this part
473       of CML.
474    
475    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
476    
477    ----------------------------------------------------------------------
478    Name: Matthias Blume (blume (at) tti - c (dot) org)
479    Date: 2003/09/22 12:10:00 CDT
480    Tag: blume-20030922-110_43_2
481    Description:
482    
483    Made a new interim version and bootfiles for developer's bootstrapping
484    convenience.
485    
486       110.43.2 -- NEW BOOTFILES
487    
488    ----------------------------------------------------------------------
489    Name: Matthias Blume (blume (at) tti - c (dot) org)
490    Date: 2003/09/19 15:55:00 CDT
491    Tag: blume-20030919-cmdir
492    Description:
493    
494    1. new-install.sh -> install.sh
495    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
496    3. tweaked installer so that another name instead of .cm can be chosen
497       at install time (by setting the CM_DIR_ARC environment variable
498       during installation); once installation is complete, the name is
499       fixed
500    
501    ----------------------------------------------------------------------
502    Name: Matthias Blume (blume (at) tti - c (dot) org)
503    Date: 2003/09/18 16:00:00 CDT
504    Tag: blume-20030918-110_43_1
505    Description:
506    
507    Made a new interim version and bootfiles for developer's bootstrapping
508    convenience.
509    
510       110.43.1 -- NEW BOOTFILES
511    
512    ----------------------------------------------------------------------
513    Name: Matthias Blume (blume (at) tti - c (dot) org)
514    Date: 2003/09/18 15:20:00 CDT
515    Tag: blume-20030918-misc
516    Description:
517    
518    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
519       this seems to be controversial at the moment) and used those in
520       Posix.ProcEnv.times.
521    
522    2. Added Time.{from,to}Nanoseconds to Time.
523    
524    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
525       For example, fromLargeInt never needs to look at more than 3 "big
526       digits" to get its 53 bits of precision.
527    
528    ----------------------------------------------------------------------
529    Name: Matthias Blume (blume (at) tti - c (dot) org)
530    Date: 2003/09/17 16:30:00 CDT
531    Tag: blume-20030917-real32-slices
532    Description:
533    
534    Added an entry to the primitive environment
535    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
536    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
537    
538    Removed some of the "magic" constants in real64.sml and replaced them
539    with code that generates these values from their corresponding
540    integer counterparts.
541    
542    Made all(?) the slice-related changes to the Basis and made everything
543    compile again...
544    
545    ----------------------------------------------------------------------
546    Name: Matthias Blume (blume (at) tti - c (dot) org)
547    Date: 2003/09/15 17:45:00 CDT
548    Tag: blume-20030915-rbase
549    Description:
550    
551    Fixed bug in Real.fromLargeInt.
552    
553    ----------------------------------------------------------------------
554    Name: Matthias Blume (blume (at) tti - c (dot) org)
555    Date: 2003/09/13 18:11:00 CDT
556    Tag: blume-20030913-libinstall
557    Description:
558    
559    Minor bugfix in config/libinstall (set anchor with path to
560    standalone tool after installing it, otherwise libraries that
561    need ml-lex or ml-yacc won't compile the first time the installer
562    runs).
563    
564    ----------------------------------------------------------------------
565    Name: Matthias Blume (blume (at) tti - c (dot) org)
566    Date: 2003/09/12 11:45:00 CDT
567    Tag: blume-20030912-various
568    Description:
569    
570     - fixed bug in Real.toLargeInt
571     - fixed bug in Posix.ProcEnv.times
572     - changed inputLine functions to return an option
573     - minor installer improvements / bugfixes
574     - changed default @SMLalloc parameter for x86/celeron to 64k
575    
576    ----------------------------------------------------------------------
577    Name: Matthias Blume (blume (at) tti - c (dot) org)
578    Date: 2003/09/09 22:00:00 CDT
579    Tag: Release_110_43
580    Description:
581    
582        New working release 110.43.  New bootfiles.
583    
584    ----------------------------------------------------------------------
585    Name: Matthias Blume (blume (at) tti - c (dot) org)
586    Date: 2003/09/09 19:20:00 CDT
587    Tag: blume-20030909-installer
588    Description:
589    
590    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
591    Modified config/install.bat to take advantage of it.  Also modified
592    config/install.sh (and called it config/new-install.sh) to take advantage
593    of it on Unix systems.  (The SML code is (supposed to be) platform-
594    independent.)
595    
596    The installer can now install everything under Win32
597    as well as under *nix as long as it compiles.
598    
599    Other changes:
600    
601      - made CML compile again under Win32
602      - made eXene compile under Win32 (by providing a fake structure UnixSock
603        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
604      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
605        same as type int;  under Win32 it isn't
606      - fixed some slice-related problems in the win32-specific parts of CML
607      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
608        (under Win32, the volume name is case-insensitive, and the
609        OS.Path code compares volume names for equality)
610    
611    ----------------------------------------------------------------------
612    Name: Matthias Blume (blume (at) tti - c (dot) org)
613    Date: 2003/09/08 11:55:00 CDT
614    Tag: blume-20030908-fullpath
615    Description:
616    
617    Made Win32 version of OS.FileSys.fullPath return current directory
618    when given an empty string.  This is what the spec says, and incidentally,
619    CM depends on it.  (CM otherwise goes into an infinite loop in certain
620    cases when presented with the name of a non-existing .cm file.)
621    
622    ----------------------------------------------------------------------
623    Name: Matthias Blume (blume (at) tti - c (dot) org)
624    Date: 2003/09/04 16:30:00 CDT
625    Tag: blume-20030905-slices-etc
626    Description:
627    
628      1. Changed interface to vectors and arrays in Basis to match
629         (draft) Basis spec.
630      2. Added signatures and implementations of slices according to
631         Basis spec.
632      3. Edited source code throughout the system to make it compile again
633         under 1. and 2.  (In some cases code had to be added to have it
634         match the new signatures.)
635      4. MLRISC should be backward-compatible: the copies of the originals
636         of files that needed to change under 3. were retained, the .cm files
637         check the compiler version number and use old versions when
638         appropriate.
639      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
640         dirstream -> string option (in accordance with Basis spec).
641      6. When generating code that counts lines, ml-lex used function
642         CharVector.foldli, taking advantage of its old interface.
643         This has been replaced with the corresponding code from
644         CharVectorSlice.  (html-lex must be re-lexed!)
645      7. BitArray in smlnj-lib/Util has been extended/modified to match the
646         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
647      8. Removed temporary additions (fromInternal, toInternal) from the
648         (now obsolete) IntInf in smlnj-lib/Util.
649      9. Cleaned up structure Byte.
650     10. Added localOffset, scan, and fromString to Date (according to spec).
651         Cleaned/corrected implementation of Date.
652         (Still need to check for correctness; implement better canonicalizeDate.)
653     11. Added "scan" to signature IEEE_REAL.
654     12. Some improvements to IntInf [in particular: efficiency-hack for
655         mod and rem when second operand is 2 (for parity checks).]
656     13. Changed representation of type Time.time, using a single IntInf.int
657         value counting microseconds.  This considerably simplified the
658         implementation of structure Time.  We now support negative time
659         values; scan and fromString handle signs.
660     14. Functor PrimIO now takes two additional arguments (VectorSlice and
661         ArraySlice).
662    
663    ----------------------------------------------------------------------
664    Name: Matthias Blume (blume (at) tti - c (dot) org)
665    Date: 2003/08/28 17:00:00 CDT
666    Tag: blume-20030828-intinf
667    Description:
668    
669    This is a major update which comes with a version number bump
670    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
671    and an implementation of IntInf in the Basis.
672    
673    There are a fairly large number of related changes and updates throughout
674    the system:
675    
676    Basis:
677            - Implemented IntInf.
678            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
679            - Added some missing Real64 operations, most notably Real.toLargeInt.
680            - Added FixedInt as a synonym for Int32.
681    
682    compiler:
683            * Added support for a built-in intinf type.
684              - literals
685              - pattern matching
686              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
687              - overloading on literals and operations
688    
689            This required adding a primitive type intinf, some additional
690            primops, and implementations for several non-trivial intinf
691            operations in Core.  (The intinf type is completely abstract
692            to the compiler; all operations get delegated back to the Core.)
693    
694            * Intinf equality is handled by polyequal.  However, the compiler
695            does not print its usual warning in this case (since polyequal
696            is the right thing to do there).
697    
698            * Improved the organization of structure InlineT.
699    
700            * A word about conversion primops:
701            If conversions involving intinf do not cancel out during
702            CPS contract, then the compiler must insert calls to Core functions.
703            Since all core access must be resolved already during the FLINT
704            translate phase, it would be too late a the time of CPS contract
705            to add new Core calls.  For this reason, conversion primops
706            for intinf carry two arguments:  1. the numeric argument that
707            they are supposed to convert, and 2. the Core function that
708            can help with this conversion if necessary.  If CPS contract
709            eliminates a primop, then the associated Core function becomes
710            dead and goes away.  Intinf conversion primops that do not get
711            eliminated by CPS contract get rewritten into calls of their
712            core functions by a separate, new phase.
713    
714    interactive system:
715            - Control.Print.intinfDepth controls max length of intinf constants
716            being printed. (Analogous to Control.Print.stringDepth.)
717            - Cleanup in printutil and pputil: got rid of unused stuff and
718            duplicates; replaced some of the code with code that makes better
719            use of library functionality.
720    
721    CM:
722            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
723            are no longer silent.
724    
725    CKIT:
726            Fixed mismatched uses of Int32 and LargeInt.  I always decided
727            in favor of LargeInt -- which is now the same as IntInf.
728            CKIT-knowledgable people should check whether this is what's
729            intended and otherwise change things back to using Int32 or
730            FixedInt.
731    
732    Throughout the code:
733            Started using IntInf.int literals and built-in operations
734            (e.g., comparison with 0) where this seems appropriate.
735    
736    
737    ----------------------------------------------------------------------
738    Name: Dave MacQueen (dbm@cs.uchicago.edu)
739    Date: 2003/08/13 11:36:00 CDT
740    Tag: dbm-20030813-mcz-merge1
741    Description:
742    
743    Merging changes from the mcz-branch development branch into trunk.
744    These changes involve replacement of the emulated old prettyprinter
745    interface with direct use of the SML/NJ Lib PP library, and fixing
746    of a couple of bugs (895, 1186) relating to error messages.  A new
747    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
748    has been added.
749    
750    ----------------------------------------------------------------------
751    Name: Matthias Blume (blume (at) tti - c (dot) org)
752    Date: 2003/08/11 15:45:00 CDT
753    Tag: blume-20030811-windows
754    Description:
755    
756    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
757    
758        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
759    
760    This patch restores SML/NJ's ability to run under win32.  There are a
761    number of changes, including fixes for several bugs that had gone
762    unnoticed until now:
763    
764      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
765        trying to run the win32 version from within cygwin.)
766    
767      - There are a number of simple .bat scripts that substitute for their
768        corresponding Unix shell-scripts.  (See below.)
769    
770      - The internals of ml-build have been modified slightly.  The main
771        difference is that instead of calling ".link-sml" (or link-sml.bat)
772        using OS.Process.system, the ML process delegates this task back
773        to the script.  Otherwise problems arise in mixed environments such
774        as Cygwin where scripts look and work like Unix scripts, but
775        where OS.Process.system cannot run them.
776    
777      - In CM, the srcpath pickler used native pathname syntax -- which
778        is incorrect in the case of cross-compilation.  The new pickle format
779        is independent of platform-specific naming conventions.
780    
781      - Path configuration files (such as lib/pathconfig) can now choose
782        between native and standard syntax.  Placing a line of the form
783    
784            standard!
785    
786        into the file causes all subsequent paths to be interpreted using
787        CM standard pathname syntax (= Unix conventions); a line
788    
789            native!
790    
791        switches back to native style.  This was needed so that
792        path config files can be written portably, see src/system/pathconfig.
793    
794      - Runtime system:
795    
796          - win32-filesys.c:  get_file_time and set_file_time now
797            access modification time, not creation time.
798    
799          - I/O code made aware of new array representation.
800    
801          - Bug fixes in X86.prim.masm.
802    
803       - src/system/makeml made aware of win32. (For use under cygwin
804            and other Unix-environments for windows.)
805    
806       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
807         which caused BinIO.inputAll to fail consistently.
808    
809    .bat scripts:
810    
811       Windows .bat scripts assume that SMLNJ_HOME is defined.
812    
813       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
814         applications (sml, ml-yacc, ml-lex).
815       - ml-build.bat: analogous to ml-build.
816       - config\install.bat:  Analogous to config/install.sh.  This requires
817         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
818         (nmake etc. must be on the path, and vcvars32 must have been run.)
819         Moreover, sources for ml-lex and ml-yacc need to exist under src,
820         and the bootfile hierarchy must have been unpacked under
821         sml.boot.x86-win32.
822         The script is very primitive and does a poor job at error checking.
823         It only installs the base system, ml-lex, and ml-yacc.  No other
824         libraries are being installed (i.e., you get only those that
825         are part of the compiler.)
826       - link-sml.bat:  analogous to .link-sml, but not currently used
827    
828    Unrelated bug fixes:
829    
830      - ml-nlffigen now exports structures ST_* corresponding to incomplete
831        types.
832      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
833        otherwise.)
834    
835    ----------------------------------------------------------------------
836    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
837    Date: 2003/06/17
838    Tag: macqueen-20030617-bug895
839    Description:
840    
841    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
842    Tag will be used for new development branch (mcz-branch) for
843    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
844    
845    ----------------------------------------------------------------------
846    Name: Matthias Blume (blume (at) tti - c (dot) org)
847    Date: 2003/05/27 16:55:00 CDT
848    Tag: blume-20030527-polyeq
849    Description:
850    
851    Tried to eliminated most cases of polymorphic equality.
852    
853    ----------------------------------------------------------------------
854    Name: Matthias Blume (blume (at) tti - c (dot) org)
855    Date: 2003/05/21 17:45:00 CDT
856    Tag: blume-20030517-complete
857    Description:
858    
859    Two changes:
860    
861      1. Added a flag for controlling whether non-exhaustive bindings will
862         be treated as errors (default is false).
863      2. Cleaned up the *entire* source tree so that CMB.make goes through
864         without a single non-exhaustive match- or bind warning.
865    
866    ----------------------------------------------------------------------
867    Name: Matthias Blume (blume (at) tti - c (dot) org)
868    Date: 2003/05/17 10:20:00 CDT
869    Tag: blume-20030517-absyn
870    Description:
871    
872    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
873    
874       This mainly affects the quality of error messages.  However, some
875       of the code is now more straightforward than before. (Treatment of
876       the above four constructs in translate.sml is much simpler than
877       the "macro-expansion" that was going on before.  Plus, the mach-
878       compiler no longer gets invoked just to be able to compile an
879       if-expression.)
880    
881    2. The ErrorMsg.Error exception is now caught and absorbed by the
882       interactive loop.
883    
884    ----------------------------------------------------------------------
885    Name: Allen Leung
886    Date: 2003/05/16 13:05:00 CDT
887    Tag: leunga-20030516-cygwin-runtime
888    Description:
889    
890        Ported the runtime system to cygwin, which uses the unix
891    x86-unix bin files.  Missing/buggy features:
892    
893       o getnetbyname, getnetbyaddr: these functions seem to be missing in
894         the Cygwin library.
895       o Ctrl-C handling may be flaky.
896       o Windows system calls and Windows I/O are not supported.
897    
898       A new set of binfiles is located at:
899    
900         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
901    
902       This is only needed for bootstrapping the cygwin version of smlnj.
903       Other x86 versions can use the existing binfiles.
904    
905    ----------------------------------------------------------------------
906    Name: Matthias Blume
907    Date: 2003/04/08 15:42:00 CDT
908    Tag: blume-20030408-listpair
909    Description:
910    
911    1. Added a target 'mlrisc' to installer.
912    
913    2. Added missing elements to structure ListPair.
914    
915    ----------------------------------------------------------------------
916    Name: Allen Leung
917    Date: 2003/01/07 10:40:00 EST
918    Tag: leunga-20030107-int-rem
919    Description:
920    
921    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
922    The arguments to the SUBL instruction were swapped.
923    
924    ----------------------------------------------------------------------
925    Name: Matthias Blume
926    Date: 2002/12/12 16:25:00 EST
927    Tag: blume-20021212-risc-ra
928    Description:
929    
930    Fixed a serious bug in the rewrite code for FP spilling/reloading that
931    sent the RA into an infinite loop when floating point registers get
932    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
933    110.37 and 110.38.)
934    There was another set of potential problems related to the handling of
935    MLRISC annotations (but those did not yet cause real problems, apparently).
936    
937    ----------------------------------------------------------------------
938    Name: Matthias Blume
939    Date: 2002/12/06 22:40:00 EST
940    Tag: blume-20021206-cm-fileid
941    Description:
942    
943    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
944    This fixes the problem of CM getting confused by files that suddenly
945    change their identity (e.g., by getting unlinked and recreated by some
946    text editor such as vi).  There might be a better/cheaper/cleaner way
947    of doing this, but for now this will have to do.
948    
949    ----------------------------------------------------------------------
950    Name: Matthias Blume
951    Date: 2002/10/28 09:50:00 EST
952    Tag: blume-20021028-typecheck
953    Description:
954    
955    Exported structure Typecheck from $smlnj/viscomp/core.cm.
956    
957    ----------------------------------------------------------------------
958    Name: Matthias Blume
959    Date: 2002/10/17 09:10:00 EDT
960    Tag: Release_110_42
961    Description:
962    
963    In good old tradition, there has been a slight hiccup so that we have
964    to patch 110.42 after the fact.   The old release tag has been replaced
965    (see below).
966    
967    The change solves a problem with two competing approaches the
968    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
969    each other's way.
970    
971    This change only affects the runtime system code and the installer script.
972    (No new bootfiles.)
973    
974    ----------------------------------------------------------------------
975    Name: Matthias Blume
976    Date: 2002/10/16 12:00:00 EDT
977    Tag: Release_110_42_removed
978    Description:
979    
980    New working release.  New bootfiles.
981    
982    ----------------------------------------------------------------------
983    Name: Matthias Blume
984    Date: 2002/10/10 13:10:00 EDT
985    Tag: blume-20021010-ppc-divs
986    Description:
987    
988    The mltree operator DIVS must be implemented with an overflow check on
989    the PPC because the hardware indicates divide-by-zero using "overflow" as
990    well.
991    
992    ----------------------------------------------------------------------
993    Name: Matthias Blume
994    Date: 2002/07/23 11:45:00
995    Tag: blume-20020723-smlnj-home
996    Description:
997    
998    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
999    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
1000    CM_PATHCONFIG is also set), the path configuration file is assumed
1001    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
1002    the entire tree to some other place and everything will "just work".
1003    
1004    (Companion commands such as ml-build and ml-makedepend also sense this
1005    variable.)
1006    
1007    ----------------------------------------------------------------------
1008    Name: Matthias Blume
1009    Date: 2002/07/12 21:19:00 EDT
1010    Tag: blume-20020712-liveness
1011    Description:
1012    
1013    Exported two useful "step" functions from liveness module (MLRISC).
1014    
1015    ----------------------------------------------------------------------
1016    Name: Matthias Blume
1017    Date: 2002/07/05 16:00 EDT
1018    Tag: Release_110_41
1019    Description:
1020    
1021    New working release.  New bootfiles.
1022    
1023    ----------------------------------------------------------------------
1024    Name: Matthias Blume
1025    Date: 2002/07/05 10:25:00 EDT
1026    Tag: blume-20020705-btimp
1027    Description:
1028    
1029    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
1030    other clients can set up backtracing support.
1031    
1032    ----------------------------------------------------------------------
1033    Name: Matthias Blume
1034    Date: 2002/06/25 17:23:00 EDT
1035    Tag: blume-20020625-fpmax
1036    Description:
1037    
1038    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
1039    case.  (The sense of the isNaN test was reversed -- which made min and
1040    max always return their first argument.)
1041    
1042    ----------------------------------------------------------------------
1043    Name: Matthias Blume
1044    Date: 2002/06/11
1045    Tag: blume-20020611-unixpath
1046    Description:
1047    
1048    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
1049    
1050    ----------------------------------------------------------------------
1051    Name: Matthias Blume
1052    Date: 2002/06/10 16:35:00 EDT
1053    Tag: blume-20020610-ieeereal
1054    Description:
1055    
1056    I back-ported my implementation of IEEEReal.fromString from the
1057    idlbasis-devel branch so that we can test it.
1058    
1059    Another small change is that ppDec tries to give more information
1060    than just "<sig>" in the case of functors.  However, this code is
1061    broken in some mysterious way if the functor's body's signature
1062    has not been declared by ascription but gets inferred from the
1063    implementation.  This needs fixing...
1064    
1065    ----------------------------------------------------------------------
1066    Name: Matthias Blume
1067    Date: 2002/05/31
1068    Tag: blume-20020531-btrace-mode
1069    Description:
1070    
1071    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
1072    the wayside when I switched over to using Controls everywhere.)
1073    
1074    ----------------------------------------------------------------------
1075    Name: Lal George
1076    Date: 2002/05/23 12:21:40 EDT
1077    Tag: george-20020523-visual-labels
1078    Description:
1079    
1080    Labels are now displayed in the graphical output to make
1081    the fall-through and target blocks obvious.
1082    
1083    ----------------------------------------------------------------------
1084    Name: Matthias Blume
1085    Date: 2002/05/22 11:03:00 EDT
1086    Tag: blume-20020522-shrink
1087    Description:
1088    
1089    John tweaked yesterday's fix for 1131 to handle an out-of-memory
1090    situation that comes up when allocating huge arrays.
1091    
1092    ----------------------------------------------------------------------
1093    Name: Matthias Blume
1094    Date: 2002/05/21 16:00:00 EDT
1095    Tag: Release_110_40
1096    Description:
1097    
1098    New working release (110.40).  New bootfiles.
1099    
1100    [Also: John Reppy fixed GC bug 1131.]
1101    
1102    ----------------------------------------------------------------------
1103    Name: Matthias Blume
1104    Date: 2002/05/21 12:35:00 EDT
1105    Tag: blume-20020521-cmdoc
1106    Description:
1107    
1108    CM documentation update.
1109    
1110    ----------------------------------------------------------------------
1111    Name: Matthias Blume
1112    Date: 2002/05/21 10:55:00 EDT
1113    Tag: blume-20020521-misc
1114    Description:
1115    
1116    - John tweaked runtime to be silent on heap export (except when
1117      GC messages are on).
1118    - I added a few more things (cross-compiling versions of CMB) to
1119      config/preloads (as suggestions).
1120    
1121    ----------------------------------------------------------------------
1122    Name: Matthias Blume
1123    Date: 2002/05/20 22:25:00 EDT
1124    Tag: blume-20020520-controls
1125    Description:
1126    
1127    - Added ControlUtil structure to control-lib.cm.
1128    - Use it throughout.
1129    - Used Controls facility to define MLRISC controls (as opposed to
1130      registering MLRISC control ref cells with Controls after the fact)
1131    - Fixed messed-up controls priorities.
1132    
1133    * Removed again all the stuff from config/preloads that one wouldn't
1134      be able to preload at the time the initial heap image is built.
1135      (Many libraries, e.g., CML, do not exist yet at this time.  The
1136       only libraries that can be preloaded via config/preloads are those
1137       that come bundled with the bootfiles.)
1138    
1139    ----------------------------------------------------------------------
1140    Name: Matthias Blume
1141    Date: 2002/05/20 10:59:00 EDT
1142    Tag: blume-20020520-preloads
1143    Description:
1144    
1145    Added a lot of commented-out suggestions for things to be included
1146    in config/preloads.
1147    
1148    ----------------------------------------------------------------------
1149    Name: Allen Leung
1150    Date: 2002/05/18 14:20:00 EDT
1151    Tag: leunga-20020518-mdl
1152    Description:
1153    
1154    o Made the mdl tool stuff compile and run again.
1155    o I've disabled all the stuff that depends on RTL specifications; they
1156      are all badly broken anyway.
1157    
1158    ----------------------------------------------------------------------
1159    Name: Matthias Blume
1160    Date: 2002/05/17 16:49:00 EDT
1161    Tag: blume-20020517-controls
1162    Description:
1163    
1164    0. John Reppy made several modifications to the SML/NJ library.
1165       In particular, there is a shiny new controls-lib.cm.
1166    
1167    1. Pushed new controls interface through compiler so that everything
1168       compiles again.
1169    
1170    2. Added FormatComb and FORMAT_COMB to the CML version of the
1171       SML/NJ library (so that CML compiles again).
1172    
1173    3. Modified init scripts because XXX_DEFAULT environment variables
1174       are no longer with us.  (Boot-time initialization is now done
1175       using the same environment variables that are also used for
1176       startup-time initialization of controls.)
1177    
1178    ----------------------------------------------------------------------
1179    Name: Lal George
1180    Date: 2002/05/15 09:20:10 EDT
1181    Tag: george-20020515-pseudo-op-decls
1182    Description:
1183    
1184            All pseudo-ops emitted before the first segment declaration
1185    such as TEXT, DATA, and BSS directives are assumed to be global
1186    declarations and are emitted first in the assembly file. This is
1187    useful in a number of situations where one has pseudo-ops that are not
1188    specific to any segment, and also works around the constraint that one
1189    cannot have client pseudo-ops in the TEXT segment.
1190    
1191            Because no segment is associated with these declarations it is
1192    an error to allocate any space or objects before the first segment
1193    directive and an exception will be raised. However, we cannot make
1194    this check for client pseudo-ops.
1195    
1196            These top level declarations are a field in the CFG graph_info.
1197    In theory you can continue to add to this field after the CFG has been
1198    built -- provided you know what you are doing;-)
1199    
1200    ----------------------------------------------------------------------
1201    Name: Matthias Blume
1202    Date: 2002/05/13 16:40:00 EDT
1203    Tag: blume-20020513-pp-etc
1204    Description:
1205    
1206    A few minor bugfixes:
1207    
1208      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1209        (Bogus printouts for redefined bindings still occur.  Compiler
1210        bug should no longer occur now.  We need to redo the prettyprinter
1211        from scratch.)
1212    
1213      - CM pathname printer now also adds escape sequences for ( and )
1214    
1215      - commend and docu fixes for ml-nlffi
1216    
1217    ----------------------------------------------------------------------
1218    Name: Matthias Blume
1219    Date: 2002/05/10 16:40:00 EDT
1220    Tag: blume-20020510-erg-textio
1221    Description:
1222    
1223    Applied the following bugfix provided by Emden Gansner:
1224    
1225        Output is corrupted when outputSubstr is used rather than output.
1226    
1227        The problem occurs when a substring
1228    
1229            ss = (s, dataStart, dataLen)
1230    
1231        where dataStart > 0, fills a stream buffer with avail bytes left.
1232        avail bytes of s, starting at index dataStart, are copied into the
1233        buffer, the buffer is flushed, and then the remaining dataLen-avail
1234        bytes of ss are copied into the beginning of the buffer. Instead of
1235        starting this copy at index dataStart+avail in s, the current code
1236        starts the copy at index avail.
1237    
1238        Fix:
1239        In text-io-fn.sml, change line 695 from
1240             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1241        to
1242             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1243    
1244    ----------------------------------------------------------------------
1245    Name: Matthias Blume
1246    Date: 2002/04/12 13:55:00 EDT
1247    Tag: blume-20020412-assyntax
1248    Description:
1249    
1250    1. Grabbed newer assyntax.h from the XFree86 project.
1251    2. Fiddled with how to compile X86.prim.asm without warnings.
1252    3. (Very) Minor cleanup in CM.
1253    
1254    ----------------------------------------------------------------------
1255    Name: Matthias Blume
1256    Date: 2002/04/01 (no joke!) 17:07:00 EST
1257    Tag: blume-20020401-x86div
1258    Description:
1259    
1260    Added full support for div/mod/rem/quot on the x86, using the machine
1261    instruction's two results (without clumsily recomputing the remainder)
1262    directly where appropriate.
1263    
1264    Some more extensive power-of-two support was added to the x86 instruction
1265    selector (avoiding expensive divs, mods, and muls where they can be
1266    replaced with cheaper shifts and masks).  However, this sort of thing
1267    ought to be done earlier, e.g., within the CPS optimizer so that
1268    all architectures benefit from it.
1269    
1270    The compiler compiles to a fixed point, but changes might be somewhat
1271    fragile nevertheless.  Please, report any strange things that you might
1272    see wrt. div/mod/quot/rem...
1273    
1274    ----------------------------------------------------------------------
1275    Name: Matthias Blume
1276    Date: 2002/03/29 17:22:00
1277    Tag: blume-20020329-div
1278    Description:
1279    
1280    Fixed my broken div/mod logic.  Unfortunately, this means that the
1281    inline code for div/mod now has one more comparison than before.
1282    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1283    The problem was with quotient = 0, because that alone does not tell
1284    us which way the rounding went.  One then has to look at whether
1285    remainder and divisor have the same sign...  :(
1286    
1287    Anyway, I replaced the bootfiles with fresh ones...
1288    
1289    ----------------------------------------------------------------------
1290    Name: Matthias Blume
1291    Date: 2002/03/29 14:10:00 EST
1292    Tag: blume-20020329-inlprims
1293    Description:
1294    
1295    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1296    
1297    Primops have changed. This means that the bin/boot-file formats have
1298    changed as well.
1299    
1300    To make sure that there is no confusion, I made a new version.
1301    
1302    
1303    CHANGES:
1304    
1305    * removed REMT from mltree (remainder should never overflow).
1306    
1307    * added primops to deal with divisions of all flavors to the frontend
1308    
1309    * handled these primops all the way through so they map to their respective
1310      MLRISC support
1311    
1312    * used these primops in the implementation of Int, Int32, Word, Word32
1313    
1314    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1315    
1316    * parameterized INLMIN, INLMAX, and INLABS by a numkind
1317    
1318    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1319      floating point
1320    
1321    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1322      and Real (but Real.abs maps to a separate floating-point-only primop)
1323    
1324    
1325    TODO items:
1326    
1327    * Hacked Alpha32 instruction selection, disabling the selection of REMx
1328      instructions because the machine instruction encoder cannot handle
1329      them.  (Hppa, PPC, and Sparc instruction selection did not handle
1330      REM in the first place, and REM is supported by the x86 machine coder.)
1331    
1332    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1333      selection phase.  (The two can be streamlined because the hardware
1334      delivers both quotient and remainder at the same time anyway.)
1335    
1336    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1337      (Currently the behavior is inconsistent both across architectures and
1338      wrt. the draft Basis spec.)
1339    
1340    * Word8 should eventually be handled natively, too.
1341    
1342    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1343      as if there currently is no execution path that could trigger it in
1344      SML/NJ.  (The assumptions underlying functions arith and promotable do not
1345      hold for things like multiplication and division.)
1346    
1347    ----------------------------------------------------------------------
1348    Name: Matthias Blume
1349    Date: 2002/03/27 16:27:00 EST
1350    Tag: blume-20020327-mlrisc-divisions
1351    Description:
1352    
1353    Added support for all four division operations (ML's div, mod, quot,
1354    and rem) to MLRISC.  In the course of doing so, I also rationalized
1355    the naming (no more annoying switch-around of DIV and QUOT), by
1356    parameterizing the operation by div_rounding_mode (which can be either
1357    DIV_TO_ZERO or DIV_TO_NEGINF).
1358    
1359    The generic MLTreeGen functor takes care of compiling all four
1360    operations down to only round-to-zero div.
1361    
1362    Missing pieces:
1363    
1364      * Doing something smarter than relying on MLTreeGen on architectures
1365        like, e.g., the x86 where hardware division delivers both quotient and
1366        remainder at the same time.  With this, the implementation of the
1367        round-to-neginf operations could be further streamlined.
1368    
1369      * Remove inlining support for div/mod/rem from the frontend and replace it
1370        with primops that get carried through to the backend.  Do this for all
1371        int and word types.
1372    
1373    ----------------------------------------------------------------------
1374    Name: Matthias Blume
1375    Date: 2002/03/25 17:25:00 EST
1376    Tag: blume-20020325-divmod
1377    Description:
1378    
1379    I improved (hopefully without breaking them) the implementation of Int.div,
1380    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1381    advantage of the following observations:
1382    
1383      Let  q = x quot y      r = x rem y
1384           d = x div  y      m = x mod y
1385    
1386    where "quot" is the round-to-zero version of integer division that
1387    hardware usually provides.  Then we have:
1388    
1389         r = x - q * y        where neither the * nor the - will overflow
1390         d = if q >= 0 orelse x = q * y then q else q - 1
1391                              where neither the * nor the - will overflow
1392         m = if q >= 0 orelse r = 0 then r else r + y
1393                              where the + will not overflow
1394    
1395    This results in substantial simplification of the generated code.
1396    The following table shows the number of CFG nodes and edges generated
1397    for
1398            fun f (x, y) = x OPER y
1399            (* with OPER \in div, mod, quot, rem *)
1400    
1401    
1402        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1403        --------------------------------------------------------
1404         div |         24 |         39 |         12 |         16
1405         mod |         41 |         71 |         12 |         16
1406        quot |          8 |         10 |          8 |         10
1407         rem |         10 |         14 |          8 |         10
1408    
1409    
1410    ----------------------------------------------------------------------
1411    Name: Matthias Blume
1412    Date: 2002/03/25 22:06:00 EST
1413    Tag: blume-20020325-cprotobug
1414    Description:
1415    
1416    Fixed a bug in cproto (c prototype decoder).
1417    
1418    ----------------------------------------------------------------------
1419    Name: Matthias Blume
1420    Date: 2002/03/25 16:00:00 EST
1421    Tag: blume-20020325-raw-primops
1422    Description:
1423    
1424    I did some cleanup to Allen's new primop code and
1425    replaced yesterday's bootfiles with new ones.
1426    (But they are stored in the same place.)
1427    
1428    ----------------------------------------------------------------------
1429    Name: Matthias Blume
1430    Date: 2002/03/24 22:40:00 EST
1431    Tag: blume-20020324-bootfiles
1432    Description:
1433    
1434    Made the bootfiles that Allen asked for.
1435    
1436    ----------------------------------------------------------------------
1437    Name: Allen Leung
1438    Date: 2002/03/23 15:50:00 EST
1439    Tag: leunga-20020323-flint-cps-rcc-primops
1440    Description:
1441    
1442      1. Changes to FLINT primops:
1443    
1444        (* make a call to a C-function;
1445         * The primop carries C function prototype information and specifies
1446         * which of its (ML-) arguments are floating point. C prototype
1447         * information is for use by the backend, ML information is for
1448         * use by the CPS converter. *)
1449      | RAW_CCALL of { c_proto: CTypes.c_proto,
1450                       ml_args: ccall_type list,
1451                       ml_res_opt: ccall_type option,
1452                       reentrant : bool
1453                     } option
1454       (* Allocate uninitialized storage on the heap.
1455        * The record is meant to hold short-lived C objects, i.e., they
1456        * are not ML pointers.  With the tag, the representation is
1457        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1458        *)
1459      | RAW_RECORD of {tag:bool,sz:int}
1460      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1461    
1462      2.  These CPS primops are now overloaded:
1463    
1464           rawload of {kind:numkind}
1465           rawstore of {kind:numkind}
1466    
1467          The one argument form is:
1468    
1469             rawload {kind} address
1470    
1471          The two argument form is:
1472    
1473             rawload {kind} [ml object, byte-offset]
1474    
1475      3. RAW_CCALL/RCC now takes two extra arguments:
1476    
1477         a. The first is whether the C call is reentrant, i.e., whether
1478            ML state should be saved and restored.
1479         b. The second argument is a string argument specifying the name of
1480            library and the C function.
1481    
1482         These things are currently not handled in the code generator, yet.
1483    
1484      4. In CProto,
1485    
1486         An encoding type of "bool" means "ml object" and is mapped into
1487         C prototype of PTR.  Note that "bool" is different than "string",
1488         even though "string" is also mapped into PTR, because "bool"
1489         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1490    
1491      5. Pickler/unpicker
1492    
1493         Changed to handle RAW_RECORD and newest RAW_CCALL
1494    
1495      6. MLRiscGen,
1496    
1497         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1498         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1499            in the file CodeGen/cpscompile/cps-c-calls.sml
1500    
1501      7. Added the conditional move operator
1502    
1503             condmove of branch
1504    
1505         to cps.  Generation of this is still buggy so it is currently
1506         disabled.
1507    
1508    ----------------------------------------------------------------------
1509    Name: Lal George
1510    Date: 2002/03/22 14:18:25 EST
1511    Tag: george-20020322-cps-branch-prob
1512    Description:
1513    
1514    Implemented the Ball-Larus branch prediction-heuristics, and
1515    incorporated graphical viewers for control flow graphs.
1516    
1517    Ball-Larus Heuristics:
1518    ---------------------
1519    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1520    
1521    By design it uses the Dempster-Shafer theory for combining
1522    probabilities.  For example, in the function:
1523    
1524        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1525    
1526    the ball-larus heuristics predicts that the n=0 is unlikely
1527    (OH-heuristic), and the 'then' branch is unlikely because of the
1528    RH-heuristic -- giving the 'then' branch an even lower combined
1529    probability using the Dempster-Shafer theory.
1530    
1531    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1532    probability of the 'then' branch because of the loop in the else
1533    branch.
1534    
1535    
1536    Graphical Viewing:
1537    ------------------
1538    I merely plugged in Allen's graphical viewers into the compiler. The
1539    additional code is not much. At the top level, saying:
1540    
1541            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1542    
1543    will display the graphical view of the control flow graph just before
1544    back-patching.  daVinci must be in your path for this to work. If
1545    daVinci is not available, then the default viewer can be changed
1546    using:
1547    
1548            Control.MLRISC.getString "viewer"
1549    
1550    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1551    course, these viewers must be in your path.
1552    
1553    The above will display the compilation unit at the level of clusters,
1554    many of which are small, boring, and un-interesting. Also setting:
1555    
1556            Control.MLRISC.getInt "cfg-graphical-view_size"
1557    
1558    will display clusters that are larger than the value set by the above.
1559    
1560    
1561    ----------------------------------------------------------------------
1562    Name: Matthias Blume
1563    Date: 2002/03/21 22:20:00 EST
1564    Tag: blume-20020321-kmp-bugfix
1565    Description:
1566    
1567    Changed the interface to the KMP routine in PreString and fixed
1568    a minor bug in one place where it was used.
1569    
1570    ----------------------------------------------------------------------
1571    Name: Allen Leung
1572    Date: 2002/03/21 20:30:00 EST
1573    Tag: leunga-20020321-cfg
1574    Description:
1575    
1576      Fixed a potential problem in cfg edge splitting.
1577    
1578    ----------------------------------------------------------------------
1579    Name: Allen Leung
1580    Date: 2002/03/21 17:15:00 EST
1581    Tag: leunga-20020321-x86-fp-cfg
1582    Description:
1583    
1584      1. Recoded the buggy parts of x86-fp.
1585    
1586         a. All the block reordering code has been removed.
1587            We now depend on the block placement phases to do this work.
1588    
1589         b. Critical edge splitting code has been simplified and moved into the
1590            CFG modules, as where they belong.
1591    
1592         Both of these were quite buggy and complex.  The code is now much, much
1593         simpler.
1594    
1595      2. X86 backend.
1596    
1597         a. Added instructions for 64-bit support.  Instruction selection for
1598            64-bit has not been committed, however, since that
1599            requires changes to MLTREE which haven't been approved by
1600            Lal and John.
1601    
1602         b. Added support for FUCOMI and FUCOMIP when generating code for
1603            PentiumPro and above.  We only generate these instructions in
1604            the fast-fp mode.
1605    
1606         c. Added cases for JP and JNP in X86FreqProps.
1607    
1608      3. CFG
1609    
1610         CFG now has a bunch of methods for edge splitting and merging.
1611    
1612      4. Machine description.
1613    
1614         John's simplification of MLTREE_BASIS.fcond broke a few machine
1615         description things:
1616    
1617         rtl-build.{sig,sml} and hppa.mdl fixed.
1618    
1619         NOTE: the machine description stuff in the repository is still broken.
1620               Again, I can't put my fixes in because that involves
1621               changes to MLTREE.
1622    
1623    ----------------------------------------------------------------------
1624    Name: Matthias Blume
1625    Date: 2002/03/20 15:55:00 EST
1626    Tag: blume-20020320-kmp
1627    Description:
1628    
1629    Implemented Knuth-Morris-Pratt string matching in PreString and used
1630    it for String.isSubstring, Substring.isSubstring, and
1631    Substring.position.
1632    
1633    (Might need some stress-testing.  Simple examples worked fine.)
1634    
1635    ----------------------------------------------------------------------
1636    Name: Matthias Blume
1637    Date: 2002/03/19 16:37:00 EST
1638    Tag: blume-20020319-witnesses
1639    Description:
1640    
1641    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1642    
1643    This implements a generic mechanism for changing constness qualifiers
1644    anywhere within big C types without resorting to outright "casts".
1645    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1646    modify the constness at the outermost level.)
1647    The implementation of "convert" is based on the idea of "witness"
1648    values -- values that are not used by the operation but whose types
1649    "testify" to their applicability.  On the implementation side, "convert"
1650    is simply a projection (returning its second curried argument).  With
1651    cross-module inlining, it should not result in any machine code being
1652    generated.
1653    
1654    ----------------------------------------------------------------------
1655    Name: Matthias Blume
1656    Date: 2002/03/15 16:40:00 EST
1657    Tag: blume-20020315-basis
1658    Description:
1659    
1660    Provided (preliminary?) implementations for
1661    
1662      {String,Substring}.{concatWith,isSuffix,isSubstring}
1663    
1664    and
1665    
1666      Substring.full
1667    
1668    Those are in the Basis spec but they were missing in SML/NJ.
1669    
1670    ----------------------------------------------------------------------
1671    Name: Matthias Blume
1672    Date: 2002/03/14 21:30:00 EST
1673    Tag: blume-20020314-controls
1674    Description:
1675    
1676    Controls:
1677    ---------
1678    
1679    1. Factored out the recently-added Controls : CONTROLS stuff and put
1680       it into its own library $/controls-lib.cm.  The source tree for
1681       this is under src/smlnj-lib/Controls.
1682    
1683    2. Changed the names of types and functions in this interface, so they
1684       make a bit more "sense":
1685    
1686          module -> registry
1687          'a registry -> 'a group
1688    
1689    3. The interface now deals in ref cells only.  The getter/setter interface
1690       is (mostly) gone.
1691    
1692    4. Added a function that lets one register an already-existing ref cell.
1693    
1694    5. Made the corresponding modifications to the rest of the code so that
1695       everything compiles again.
1696    
1697    6. Changed the implementation of Controls.MLRISC back to something closer
1698       to the original.  In particular, this module (and therefore MLRISC)
1699       does not depend on Controls.  There now is some link-time code in
1700       int-sys.sml that registers the MLRISC controls with the Controls
1701       module.
1702    
1703    CM:
1704    ---
1705    
1706      * One can now specify the lambda-split aggressiveness in init.cmi.
1707    
1708    ----------------------------------------------------------------------
1709    Name: Allen Leung
1710    Date: 2002/03/13 17:30:00 EST
1711    Tag: leunga-20020313-x86-fp-unary
1712    Description:
1713    
1714    Bug fix for:
1715    
1716    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1717    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1718    > - fun f(x,(y,z)) = Real.~ y;
1719    > [autoloading]
1720    > [autoloading done]
1721    >       fchsl   (%eax), 184(%esp)
1722    > Error: MLRisc bug: X86MCEmitter.emitInstr
1723    >
1724    > uncaught exception Error
1725    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1726    
1727    The problem was that the code generator did not generate any fp registers
1728    in this case, and the ra didn't know that it needed to run the X86FP phase to
1729    translate the pseudo fp instruction.   This only happened with unary fp
1730    operators in certain situations.
1731    
1732    ----------------------------------------------------------------------
1733    Name: Matthias Blume
1734    Date: 2002/03/13 14:00:00 EST
1735    Tag: blume-20020313-overload-etc
1736    Description:
1737    
1738    1. Added _overload as a synonym for overload for backward compatibility.
1739       (Control.overloadKW must be true for either version to be accepted.)
1740    
1741    2. Fixed bug in install script that caused more things to be installed
1742       than what was requested in config/targets.
1743    
1744    3. Made CM aware of the (_)overload construct so that autoloading
1745       works.
1746    
1747    ----------------------------------------------------------------------
1748    Name: Matthias Blume
1749    Date: 2002/03/12 22:03:00 EST
1750    Tag: blume-20020312-url
1751    Description:
1752    
1753    Forgot to update BOOT and srcarchiveurl.
1754    
1755    ----------------------------------------------------------------------
1756    Name: Matthias Blume
1757    Date: 2002/03/12 17:30:00 EST
1758    Tag: blume-20020312-version110392
1759    Description:
1760    
1761    Yet another version number bump (because of small changes to the
1762    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1763    
1764    Changes:
1765    
1766      The new pid generation scheme described a few weeks ago was overly
1767      complicated.  I implemented a new mechanism that is simpler and
1768      provides a bit more "stability":  Once CM has seen a compilation
1769      unit, it keeps its identity constant (as long as you do not delete
1770      those crucial CM/GUID/* files).  This means that when you change
1771      an interface, compile, then go back to the old interface, and
1772      compile again, you arrive at the original pid.
1773    
1774      There now also is a mechanism that instructs CM to use the plain
1775      environment hash as a module's pid (effectively making its GUID
1776      the empty string).  For this, "noguid" must be specified as an
1777      option to the .sml file in question within its .cm file.
1778      This is most useful for code that is being generated by tools such
1779      as ml-nlffigen (because during development programmers tend to
1780      erase the tool's entire output directory tree including CM's cached
1781      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1782      revert to the old, broken behavior of SML/NJ, but in specific cases
1783      where there is no danger of interface confusion, its use is ok
1784      (I think).
1785    
1786      ml-nlffigen by default generates "noguid" annotations.  They can be
1787      turned off by specifying -guid in its command line.
1788    
1789    ----------------------------------------------------------------------
1790    Name: Lal George
1791    Date: 2002/03/12 12 14:42:36 EST
1792    Tag: george-20020312-frequency-computation
1793    Description:
1794    
1795    Integrated jump chaining and static block frequency into the
1796    compiler. More details and numbers later.
1797    
1798    ----------------------------------------------------------------------
1799    Name: Lal George
1800    Date: 2002/03/11 11 22:38:53 EST
1801    Tag: george-20020311-jump-chain-elim
1802    Description:
1803    
1804    Tested the jump chain elimination on all architectures (except the
1805    hppa).  This is on by default right now and is profitable for the
1806    alpha and x86, however, it may not be profitable for the sparc and ppc
1807    when compiling the compiler.
1808    
1809    The gc test will typically jump to a label at the end of the cluster,
1810    where there is another jump to an external cluster containing the actual
1811    code to invoke gc. This is to allow factoring of common gc invocation
1812    sequences. That is to say, we generate:
1813    
1814            f:
1815               testgc
1816               ja   L1      % jump if above to L1
1817    
1818            L1:
1819               jmp L2
1820    
1821    
1822    After jump chain elimination the 'ja L1' instructions is converted to
1823    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1824    up being implemented in their long form (if L2 is far away) using:
1825    
1826            jbe     L3      % jump if below or equal to L3
1827            jmp     L2
1828         L3:
1829            ...
1830    
1831    
1832    For large compilation units L2  may be far away.
1833    
1834    
1835    ----------------------------------------------------------------------
1836    Name: Matthias Blume
1837    Date: 2002/03/11 13:30:00 EST
1838    Tag: blume-20020311-mltreeeval
1839    Description:
1840    
1841    A functor parameter was missing.
1842    
1843    ----------------------------------------------------------------------
1844    Name: Allen Leung
1845    Date: 2002/03/11 10:30:00 EST
1846    Tag: leunga-20020311-runtime-string0
1847    Description:
1848    
1849       The representation of the empty string now points to a
1850    legal null terminated C string instead of unit.  It is now possible
1851    to convert an ML string into C string with InlineT.CharVector.getData.
1852    This compiles into one single machine instruction.
1853    
1854    ----------------------------------------------------------------------
1855    Name: Allen Leung
1856    Date: 2002/03/10 23:55:00 EST
1857    Tag: leunga-20020310-x86-call
1858    Description:
1859    
1860       Added machine generation for CALL instruction (relative displacement mode)
1861    
1862    ----------------------------------------------------------------------
1863    Name: Matthias Blume
1864    Date: 2002/03/08 16:05:00
1865    Tag: blume-20020308-entrypoints
1866    Description:
1867    
1868    Version number bumped to 110.39.1.  NEW BOOTFILES!
1869    
1870    Entrypoints: non-zero offset into a code object where execution should begin.
1871    
1872    - Added the notion of an entrypoint to CodeObj.
1873    - Added reading/writing of entrypoint info to Binfile.
1874    - Made runtime system bootloader aware of entrypoints.
1875    - Use the address of the label of the first function given to mlriscGen
1876      as the entrypoint.  This address is currently always 0, but it will
1877      not be 0 once we turn on block placement.
1878    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
1879      with entry points) from mlriscGen.
1880    
1881    ----------------------------------------------------------------------
1882    Name: Allen Leung
1883    Date: 2002/03/07 20:45:00 EST
1884    Tag: leunga-20020307-x86-cmov
1885    Description:
1886    
1887       Bug fixes for CMOVcc on x86.
1888    
1889       1. Added machine code generation for CMOVcc
1890       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1891       3. CMOVcc cannot have an immediate operand as argument.
1892    
1893    ----------------------------------------------------------------------
1894    Name: Matthias Blume
1895    Date: 2002/03/07 16:15:00 EST
1896    Tag: blume-20020307-controls
1897    Description:
1898    
1899    This is a very large but mostly boring patch which makes (almost)
1900    every tuneable compiler knob (i.e., pretty much everything under
1901    Control.* plus a few other things) configurable via both the command
1902    line and environment variables in the style CM did its configuration
1903    until now.
1904    
1905    Try starting sml with '-h' (or, if you are brave, '-H')
1906    
1907    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1908    implements the underlying generic mechanism.
1909    
1910    The interface to some of the existing such facilities has changed somewhat.
1911    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1912    (The getFoo interface is still there for backward-compatibility, but its
1913    use is deprecated.)
1914    
1915    The ml-build script passes -Cxxx=yyy command-line arguments through so
1916    that one can now twiddle the compiler settings when using this "batch"
1917    compiler.
1918    
1919    TODO items:
1920    
1921    We should go through and throw out all controls that are no longer
1922    connected to anything.  Moreover, we should go through and provide
1923    meaningful (and correct!) documentation strings for those controls
1924    that still are connected.
1925    
1926    Currently, multiple calls to Controls.new are accepted (only the first
1927    has any effect).  Eventually we should make sure that every control
1928    is being made (via Controls.new) exactly once.  Future access can then
1929    be done using Controls.acc.
1930    
1931    Finally, it would probably be a good idea to use the getter-setter
1932    interface to controls rather than ref cells.  For the time being, both
1933    styles are provided by the Controls module, but getter-setter pairs are
1934    better if thread-safety is of any concern because they can be wrapped.
1935    
1936    *****************************************
1937    
1938    One bug fix: The function blockPlacement in three of the MLRISC
1939    backpatch files used to be hard-wired to one of two possibilities at
1940    link time (according to the value of the placementFlag).  But (I
1941    think) it should rather sense the flag every time.
1942    
1943    *****************************************
1944    
1945    Other assorted changes (by other people who did not supply a HISTORY entry):
1946    
1947    1. the cross-module inliner now works much better (Monnier)
1948    2. representation of weights, frequencies, and probabilities in MLRISC
1949       changed in preparation of using those for weighted block placement
1950       (Reppy, George)
1951    
1952    ----------------------------------------------------------------------
1953    Name: Lal George
1954    Date: 2002/03/07 14:44:24 EST 2002
1955    Tag: george-20020307-weighted-block-placement
1956    
1957    Tested the weighted block placement optimization on all architectures
1958    (except the hppa) using AMPL to generate the block and edge frequencies.
1959    Changes were required in the machine properties to correctly
1960    categorize trap instructions. There is an MLRISC flag
1961    "weighted-block-placement" that can be used to enable weighted block
1962    placement, but this will be ineffective without block/edge
1963    frequencies (coming soon).
1964    
1965    
1966    ----------------------------------------------------------------------
1967    Name: Lal George
1968    Date: 2002/03/05 17:24:48 EST
1969    Tag: george-20020305-linkage-cluster
1970    
1971    In order to support the block placement optimization, a new cluster
1972    is generated as the very first cluster (called the linkage cluster).
1973    It contains a single jump to the 'real' entry point for the compilation
1974    unit. Block placement has no effect on the linkage cluster itself, but
1975    all the other clusters  have full freedom in the manner in which they
1976    reorder blocks or functions.
1977    
1978    On the x86 the typical linkage code that is generated is:
1979       ----------------------
1980            .align 2
1981       L0:
1982            addl    $L1-L0, 72(%esp)
1983            jmp     L1
1984    
1985    
1986            .align  2
1987       L1:
1988       ----------------------
1989    
1990    72(%esp) is the memory location for the stdlink register. This
1991    must contain the address of the CPS function being called. In the
1992    above example, it contains the address of  L0; before
1993    calling L1 (the real entry point for the compilation unit), it
1994    must contain the address for L1, and hence
1995    
1996            addl $L1-L0, 72(%esp)
1997    
1998    I have tested this on all architectures except the hppa.The increase
1999    in code size is of course negligible
2000    
2001    ----------------------------------------------------------------------
2002    Name: Allen Leung
2003    Date: 2002/03/03 13:20:00 EST
2004    Tag: leunga-20020303-mlrisc-tools
2005    
2006      Added #[ ... ] expressions to mlrisc tools
2007    
2008    ----------------------------------------------------------------------
2009    Name: Matthias Blume
2010    Date: 2002/02/27 12:29:00 EST
2011    Tag: blume-20020227-cdebug
2012    Description:
2013    
2014    - made types in structure C and C_Debug to be equal
2015    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
2016    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
2017    
2018    ----------------------------------------------------------------------
2019    Name: Matthias Blume
2020    Date: 2002/02/26 12:00:00 EST
2021    Tag: blume-20020226-ffi
2022    Description:
2023    
2024    1. Fixed a minor bug in CM's "noweb" tool:
2025       If numbering is turned off, then truly don't number (i.e., do not
2026       supply the -L option to noweb).  The previous behavior was to supply
2027       -L'' -- which caused noweb to use the "default" line numbering scheme.
2028       Thanks to Chris Richards for pointing this out (and supplying the fix).
2029    
2030    2. Once again, I reworked some aspects of the FFI:
2031    
2032       A. The incomplete/complete type business:
2033    
2034       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
2035         gone!
2036       - ML types representing an incomplete type are now *equal* to
2037         ML types representing their corresponding complete types (just like
2038         in C).  This is still safe because ml-nlffigen will not generate
2039         RTTI for incomplete types, nor will it generate functions that
2040         require access to such RTTI.   But when ML code generated from both
2041         incomplete and complete versions of the C type meet, the ML types
2042         are trivially interoperable.
2043    
2044         NOTE:  These changes restore the full generality of the translation
2045         (which was previously lost when I eliminated functorization)!
2046    
2047       B. Enum types:
2048    
2049       - Structure C now has a type constructor "enum" that is similar to
2050         how the "su" constructor works.  However, "enum" is not a phantom
2051         type because each "T enum" has values (and is isomorphic to
2052         MLRep.Signed.int).
2053       - There are generic access operations for enum objects (using
2054         MLRep.Signed.int).
2055       - ml-nlffigen will generate a structure E_foo for each "enum foo".
2056         * The structure contains the definition of type "mlrep" (the ML-side
2057         representation type of the enum).  Normally, mlrep is the same
2058         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
2059         then mlrep will be defined as a datatype -- thus facilitating
2060         pattern matching on mlrep values.
2061         ("-ec" will be suppressed if there are duplicate values in an
2062          enumeration.)
2063         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
2064         will be generated for each C enum constant xxx.
2065         * Conversion functions m2i and i2m convert between mlrep and
2066         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
2067         * Coversion functions c and ml convert between mlrep and "tag enum".
2068         * Access functions (get/set) fetch and store mlrep values.
2069       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
2070         enumerations are merged into one single enumeration represented by
2071         structure E_'.
2072    
2073    ----------------------------------------------------------------------
2074    Name: Allen Leung
2075    Date: 2002/02/25 04:45:00 EST
2076    Tag: leunga-20020225-cps-spill
2077    
2078    This is a new implementation of the CPS spill phase.
2079    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
2080    In case of problems, replace it with the old file spill.sml
2081    
2082    The current compiler runs into some serious performance problems when
2083    constructing a large record.  This can happen when we try to compile a
2084    structure with many items.  Even a very simple structure like the following
2085    makes the compiler slow down.
2086    
2087        structure Foo = struct
2088           val x_1 = 0w1 : Word32.int
2089           val x_2 = 0w2 : Word32.int
2090           val x_3 = 0w3 : Word32.int
2091           ...
2092           val x_N = 0wN : Word32.int
2093        end
2094    
2095    The following table shows the compile time, from N=1000 to N=4000,
2096    with the old compiler:
2097    
2098    N
2099    1000   CPS 100 spill                           0.04u  0.00s  0.00g
2100           MLRISC ra                               0.06u  0.00s  0.05g
2101              (spills = 0 reloads = 0)
2102           TOTAL                                   0.63u  0.07s  0.21g
2103    
2104    1100   CPS 100 spill                           8.25u  0.32s  0.64g
2105           MLRISC ra                               5.68u  0.59s  3.93g
2106              (spills = 0 reloads = 0)
2107           TOTAL                                   14.71u  0.99s  4.81g
2108    
2109    1500   CPS 100 spill                           58.55u  2.34s  1.74g
2110           MLRISC ra                               5.54u  0.65s  3.91g
2111              (spills = 543 reloads = 1082)
2112           TOTAL                                   65.40u  3.13s  6.00g
2113    
2114    2000   CPS 100 spill                           126.69u  4.84s  3.08g
2115           MLRISC ra                               0.80u  0.10s  0.55g
2116              (spills = 42 reloads = 84)
2117           TOTAL                                   129.42u  5.10s  4.13g
2118    
2119    3000   CPS 100 spill                           675.59u  19.03s  11.64g
2120           MLRISC ra                               2.69u  0.27s  1.38g
2121              (spills = 62 reloads = 124)
2122           TOTAL                                   682.48u  19.61s  13.99g
2123    
2124    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
2125           MLRISC ra                               4.96u  0.27s  2.72g
2126              (spills = 85 reloads = 170)
2127           TOTAL                                   2375.26u  57.21s  48.00g
2128    
2129    As you can see the old cps spill module suffers from some serious
2130    performance problem.  But since I cannot decipher the old code fully,
2131    instead of patching the problems up, I'm reimplementing it
2132    with a different algorithm.  The new code is more modular,
2133    smaller when compiled, and substantially faster
2134    (O(n log n) time and O(n) space).  Timing of the new spill module:
2135    
2136    4000  CPS 100 spill                           0.02u  0.00s  0.00g
2137          MLRISC ra                               0.25u  0.02s  0.15g
2138             (spills=1 reloads=3)
2139          TOTAL                                   7.74u  0.34s  1.62g
2140    
2141    Implementation details:
2142    
2143    As far as I can tell, the purpose of the CPS spill module is to make sure the
2144    number of live variables at any program point (the bandwidth)
2145    does not exceed a certain limit, which is determined by the
2146    size of the spill area.
2147    
2148    When the bandwidth is too large, we decrease the register pressure by
2149    packing live variables into spill records.  How we achieve this is
2150    completely different than what we did in the old code.
2151    
2152    First, there is something about the MLRiscGen code generator
2153    that we should be aware of:
2154    
2155    o MLRiscGen performs code motion!
2156    
2157       In particular, it will move floating point computations and
2158       address computations involving only the heap pointer to
2159       their use sites (if there is only a single use).
2160       What this means is that if we have a CPS record construction
2161       statement
2162    
2163           RECORD(k,vl,w,e)
2164    
2165       we should never count the new record address w as live if w
2166       has only one use (which is often the case).
2167    
2168       We should do something similar to floating point, but the transformation
2169       there is much more complex, so I won't deal with that.
2170    
2171    Secondly, there are now two new cps primops at our disposal:
2172    
2173     1. rawrecord of record_kind option
2174        This pure operator allocates some uninitialized storage from the heap.
2175        There are two forms:
2176    
2177         rawrecord NONE [INT n]  allocates a tagless record of length n
2178         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2179                                     and initializes the tag.
2180    
2181     2. rawupdate of cty
2182          rawupdate cty (v,i,x)
2183          Assigns to x to the ith component of record v.
2184          The storelist is not updated.
2185    
2186    We use these new primops for both spilling and increment record construction.
2187    
2188     1. Spilling.
2189    
2190        This is implemented with a linear scan algorithm (but generalized
2191        to trees).  The algorithm will create a single spill record at the
2192        beginning of the cps function and use rawupdate to spill to it,
2193        and SELECT or SELp to reload from it.  So both spills and reloads
2194        are fine-grain operations.  In contrast, in the old algorithm
2195        "spills" have to be bundled together in records.
2196    
2197        Ideally, we should sink the spill record construction to where
2198        it is needed.  We can even split the spill record into multiple ones
2199        at the places where they are needed.  But CPS is not a good
2200        representation for global code motion, so I'll keep it simple and
2201        am not attempting this.
2202    
2203     2. Incremental record construction (aka record splitting).
2204    
2205        Long records with many component values which are simulatenously live
2206        (recall that single use record addresses are not considered to
2207         be live) are constructed with rawrecord and rawupdate.
2208        We allocate space on the heap with rawrecord first, then gradually
2209        fill it in with rawupdate.  This is the technique suggested to me
2210        by Matthias.
2211    
2212        Some restrictions on when this is applicable:
2213        1. It is not a VECTOR record.  The code generator currently does not handle
2214           this case. VECTOR record uses double indirection like arrays.
2215        2. All the record component values are defined in the same "basic block"
2216           as the record constructor.  This is to prevent speculative
2217           record construction.
2218    
2219    ----------------------------------------------------------------------
2220    Name: Allen Leung
2221    Date: 2002/02/22 01:02:00 EST
2222    Tag: leunga-20020222-mlrisc-tools
2223    
2224    Minor bug fixes in the parser and rewriter
2225    
2226    ----------------------------------------------------------------------
2227    Name: Allen Leung
2228    Date: 2002/02/21 20:20:00 EST
2229    Tag: leunga-20020221-peephole
2230    
2231    Regenerated the peephole files.  Some contained typos in the specification
2232    and some didn't compile because of pretty printing bugs in the old version
2233    of 'nowhere'.
2234    
2235    ----------------------------------------------------------------------
2236    Name: Allen Leung
2237    Date: 2002/02/19 20:20:00 EST
2238    Tag: leunga-20020219-mlrisc-tools
2239    Description:
2240    
2241       Minor bug fixes to the mlrisc-tools library:
2242    
2243       1.  Fixed up parsing colon suffixed keywords
2244       2.  Added the ability to shut the error messages up
2245       3.  Reimplemented the pretty printer and fixed up/improved
2246           the pretty printing of handle and -> types.
2247       4.  Fixed up generation of literal symbols in the nowhere tool.
2248       5.  Added some SML keywords to to sml.sty
2249    
2250    ----------------------------------------------------------------------
2251    Name: Matthias Blume
2252    Date: 2002/02/19 16:20:00 EST
2253    Tag: blume-20020219-cmffi
2254    Description:
2255    
2256    A wild mix of changes, some minor, some major:
2257    
2258    * All C FFI-related libraries are now anchored under $c:
2259        $/c.cm      --> $c/c.cm
2260        $/c-int.cm  --> $c/internals/c-int.cm
2261        $/memory.cm --> $c/memory/memory.cm
2262    
2263    * "make" tool (in CM) now treats its argument pathname slightly
2264      differently:
2265        1. If the native expansion is an absolute name, then before invoking
2266           the "make" command on it, CM will apply OS.Path.mkRelative
2267           (with relativeTo = OS.FileSys.getDir()) to it.
2268        2. The argument will be passed through to subsequent phases of CM
2269           processing without "going native".  In particular, if the argument
2270           was an anchored path, then "make" will not lose track of that anchor.
2271    
2272    * Compiler backends now "know" their respective C calling conventions
2273      instead of having to be told about it by ml-nlffigen.  This relieves
2274      ml-nlffigen from one of its burdens.
2275    
2276    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
2277    
2278    * Export C_DEBUG and C_Debug from $c/c.cm.
2279    
2280    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
2281      subtyping relationship between incomplete pointers and their complete
2282      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
2283      with the convention of instantiating 'o with ('t, 'c) obj whenever
2284      the pointer target type is complete.  In the incomplete case, 'o
2285      will be instantiated with some "'c iobj" -- a type obtained by
2286      using one of the functors PointerToIncompleteType or PointerToCompleteType.
2287    
2288      Operations that work on both incomplete and complete pointer types are
2289      typed as taking an 'o ptr while operations that require the target to
2290      be known are typed as taking some ('t, 'c) obj ptr.
2291    
2292      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
2293      where void is an eqtype without any values.  This makes it possible
2294      to work on voidptr values using functions meant to operate on light
2295      incomplete pointers.
2296    
2297    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
2298      been vastly simplified.
2299    
2300    ----------------------------------------------------------------------
2301    Name: Matthias Blume
2302    Date: 2002/02/19 10:48:00 EST
2303    Tag: blume-20020219-pqfix
2304    Description:
2305    
2306    Applied Chris Okasaki's bug fix for priority queues.
2307    
2308    ----------------------------------------------------------------------
2309    Name: Matthias Blume
2310    Date: 2002/02/15 17:05:00
2311    Tag: Release_110_39
2312    Description:
2313    
2314    Last-minute retagging is becoming a tradition... :-(
2315    
2316    This is the working release 110.39.
2317    
2318    ----------------------------------------------------------------------
2319    Name: Matthias Blume
2320    Date: 2002/02/15 16:00:00 EST
2321    Tag: Release_110_39-orig
2322    Description:
2323    
2324    Working release 110.39.  New bootfiles.
2325    
2326    (Update: There was a small bug in the installer so it wouldn't work
2327    with all shells.  So I retagged. -Matthias)
2328    
2329    ----------------------------------------------------------------------
2330    Name: Matthias Blume
2331    Date: 2002/02/15 14:17:00 EST
2332    Tag: blume-20020215-showbindings
2333    Description:
2334    
2335    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
2336    the latter can be useful for exploring what bindings are available at
2337    the interactive prompt.  (The first function returns only the list
2338    of symbols that are really bound, the second prints those but also the
2339    ones that CM's autoloading mechanism knows about.)
2340    
2341    ----------------------------------------------------------------------
2342    Name: Matthias Blume
2343    Date: 2002/02/15 12:08:00 EST
2344    Tag: blume-20020215-iptrs
2345    Description:
2346    
2347    Two improvements to ml-nlffigen:
2348    
2349      1. Write files only if they do not exist or if their current contents
2350         do not coincide with what's being written.  (That is, avoid messing
2351         with the time stamps unless absolutely necessary.)
2352    
2353      2. Implement a "repository" mechanism for generated files related
2354         to "incomplete pointer types".   See the README file for details.
2355    
2356    ----------------------------------------------------------------------
2357    Name: Matthias Blume
2358    Date: 2002/02/14 11:50:00 EST
2359    Tag: blume-20020214-quote
2360    Description:
2361    
2362    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
2363    because of the new and improved tag generation scheme.  (Thanks to Allen
2364    Leung for pointing it out.)
2365    
2366    ----------------------------------------------------------------------
2367    Name: Lal George
2368    Date: 2002/02/14 09:55:27 EST 2002
2369    Tag: george-20020214-isabelle-bug
2370    Description:
2371    
2372    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2373    of Isabelle on the x86.
2374    
2375    From Allen:
2376    -----------
2377     I've found the problem:
2378    
2379         in ra-core.sml, I use the counter "blocked" to keep track of the
2380         true number of elements in the freeze queue.  When the counter goes
2381         to zero, I skip examining the queue.  But I've messed up the
2382         bookkeeping in combine():
2383    
2384             else ();
2385             case !ucol of
2386               PSEUDO => (if !cntv > 0 then
2387                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2388                                        ^^^^^^^^^^^^^^^^^^^^^^^
2389                      moveu := mergeMoveList(!movev, !moveu)
2390                     )
2391                  else ();
2392    
2393         combine() is called to coalesce two nodes u and v.
2394         I think I was thinking that if the move counts of u and v are both
2395         greater than zero then after they are coalesced then one node is
2396         removed from the freeze queue.  Apparently I was thinking that
2397         both u and v are of low degree, but that's clearly not necessarily true.
2398    
2399    
2400    02/12/2002:
2401        Here's the patch.  HOL now compiles.
2402    
2403        I don't know how this impact on performance (compile
2404        time or runtime).  This bug caused the RA (especially on the x86)
2405        to go thru the potential spill phase when there are still nodes on the
2406        freeze queue.
2407    
2408    
2409    
2410    
2411    ----------------------------------------------------------------------
2412    Name: Matthias Blume
2413    Date: 2002/02/13 22:40:00 EST
2414    Tag: blume-20020213-fptr-rtti
2415    Description:
2416    
2417    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2418    updates.
2419    
2420    ----------------------------------------------------------------------
2421    Name: Matthias Blume
2422    Date: 2002/02/13 16:41:00 EST
2423    Tag: blume-20020213-cmlpq
2424    Description:
2425    
2426    Added new priority queue export symbols (which have just been added to
2427    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2428    not compile and the installer would choke.)
2429    
2430    ----------------------------------------------------------------------
2431    Name: Matthias Blume
2432    Date: 2002/02/13 16:15:00 EST
2433    Tag: blume-20020213-various
2434    Description:
2435    
2436    1. More tweaks to ml-nlffigen:
2437    
2438       - better internal datastructures (resulting in slight speedup)
2439       - "-match" option requires exact match
2440       - "localized" gensym counters (untagged structs/unions nested within
2441         other structs/unions or within typedefs get a fresh counter; their
2442         tag will be prefixed by a concatenation of their parents' tags)
2443       - bug fixes (related to calculation of transitive closure of types
2444         to be included in the output)
2445    
2446    2. Minor Basis updates:
2447    
2448       - added implementations for List.collate and Option.app
2449    
2450    ----------------------------------------------------------------------
2451    Name: Matthias Blume
2452    Date: 2002/02/11 15:55:00 EST
2453    Tag: blume-20020211-gensym
2454    Description:
2455    
2456    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2457    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2458    names (ML structure names that correspond to unnamed C structs, unions,
2459    and enums), so that separate runs of ml-nlffigen do not clash.
2460    
2461    ----------------------------------------------------------------------
2462    Name: Matthias Blume
2463    Date: 2002/02/11 12:05:00 EST
2464    Tag: blume-20020211-gensml
2465    Description:
2466    
2467    A quick fix for a problem with GenSML (in the pgraph-util library):
2468    Make generation of toplevel "local" optional.  (Strictly speaking,
2469    signature definitions within "local" are not legal SML.)
2470    
2471    Other than that: updates to INSTALL and cm/TODO.
2472    
2473    ----------------------------------------------------------------------
2474    Name: Matthias Blume
2475    Date: 2002/02/08 15:00:00 EST
2476    Tag: blume-20020208-uniquepid
2477    Description:
2478    
2479    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2480    
2481    1. The installer (config/install.sh) has gotten smarter:
2482    
2483         - Configuration options are a bit easier to specify now
2484           (in config/targets).
2485         - Bug in recognizing .tar.bz2 files fixed.
2486         - Installer automatically resolves dependencies between
2487           configuration options (e.g., if you ask for eXene, you will
2488           also get cml -- regardless whether you asked for it or not).
2489         - Installer can run in "quieter mode" by setting the environment
2490           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2491           "completely silent", though.
2492         - Build HashCons library as part of smlnj-lib.
2493    
2494    2. A new scheme for assigning persistent identifiers to compilation
2495       units (and, by extension, to types etc.) has been put into place.
2496       This fixes a long-standing bug where types and even dynamic values
2497       can get internally confused, thereby compromising type safety
2498       (abstraction) and dynamic correctness.  See
2499    
2500         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2501    
2502       for an example of how things could go wrong until now.
2503    
2504       The downside of the new scheme is that pids are not quite as
2505       persistent as they used to be: CM will generate a fresh pid
2506       for every compilation unit that it thinks it sees for the first
2507       time.  That means that if you compile starting from a clean, fresh
2508       source tree at two different times, you end up with different
2509       binaries.
2510    
2511       Cutoff recompilation, however, has not been compromised because
2512       CM keeps pid information in special caches between runs.
2513    
2514    ----------------------------------------------------------------------
2515    Name: Lal George
2516    Date: 2002/02/07 15:34:13 EST 2002
2517    Tag: <none>
2518    Description:
2519    
2520    Compilers that generate assembly code may produce  global labels
2521    whose value is resolved at link time. The various peephole optimization
2522    modules did not take this in account.
2523    
2524    TODO. The Labels.addrOf function should really return an option
2525    type so that clients are forced to deal with this issue, rather
2526    than an exception being raised.
2527    
2528    ----------------------------------------------------------------------
2529    Name: Lal George
2530    Date: 2002/02/06 13:55:02 EST
2531    Tag: george-20020206-ra-breakup
2532    Description:
2533    
2534    1. A bug fix from Allen.
2535    
2536        A typo causes extra fstp %st(0)'s to be generated at compensation
2537        edges, which might cause stack underflow traps at runtime.  This
2538        occurs in fft where there are extraneous fstps right before the 'into'
2539        trap instruction (in this case they are harmless since none of the
2540        integers overflow.)
2541    
2542    2. Pulled out various utility modules that were embedded in the modules
2543       of the register allocator. I need these modules for other purposes, but
2544       they are not complete enough to put into a library (just yet).
2545    ----------------------------------------------------------------------
2546    Name: Matthias Blume
2547    Date: 2002/01/31 16:05:00 EST
2548    Tag: blume-20020131-sparc-ccalls
2549    Description:
2550    
2551    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2552       of extra stack space by mistake.  Fixed.
2553    
2554    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2555    
2556    ----------------------------------------------------------------------
2557    Name: Allen Leung
2558    Date: 2002/01/30
2559    Tag: leunga-20020130-nowhere-bug-fix
2560    Description:
2561    
2562       MLRISC bug fixes:
2563       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2564       2. MachineInt.fromString was negating its value.
2565    
2566    ----------------------------------------------------------------------
2567    Name: Matthias Blume
2568    Date: 2002/01/29
2569    Tag: blume-20020129-INSTALL
2570    Description:
2571    
2572    - Added somewhat detailed installation instructions (file INSTALL).
2573    - Fixed curl-detection bug in config/install.sh.
2574    - It is now possible to select the URL getter using the URLGETTER
2575      environment variable:
2576    
2577          not set / "unknown"      --> automatic detection (script tries wget,
2578                                       curl, and lynx)
2579          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2580                                       how to properly invoke them)
2581          other                    --> use $URLGETTER directly, it must take
2582                                       precisely two command-line arguments
2583                                       (source URL and destination file name)
2584    
2585    ----------------------------------------------------------------------
2586    Name: Matthias Blume
2587    Date: 2002/01/28
2588    Tag: blume-20020128-sparc-ccalls
2589    Description:
2590    
2591    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2592    - Make use of the allocParam argument in sparc-c-calls.
2593    
2594    ----------------------------------------------------------------------
2595    Name: Matthias Blume
2596    Date: 2002/01/28
2597    Tag: blume-20020128-allocParam
2598    Description:
2599    
2600    John Reppy:  Changes c-calls API to accept client-callback for
2601    allocating extra stack space.
2602    me: Corresponding changes to mlriscGen (using a dummy argument that
2603        does not change the current behavior).
2604    
2605    ----------------------------------------------------------------------
2606    Name: Matthias Blume
2607    Date: 2002/01/28 12:00:00
2608    Tag: Release_110_38
2609    Description:
2610    
2611    This time for real!!!
2612    
2613    ----------------------------------------------------------------------
2614    Name: Matthias Blume
2615    Date: 2002/01/28 10:56:00 EST
2616    Tag: blume-20020128-retraction
2617    Description:
2618    
2619    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
2620       with blume-Release_110_38-retracted.)
2621    
2622    1. Fixed a problem with incorrect rounding modes in real64.sml.
2623       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2624    
2625    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2626       and enums fixed.  The naming of corresponding ML identifiers should
2627       now be consistent again.
2628    
2629    ----------------------------------------------------------------------
2630    Name: Allen Leung
2631    Date: 2002/01/27
2632    Tag: leunga-20020127-nowhere
2633    Description:
2634    
2635       Added a target called nowhere in the configuration scripts.
2636       Enabling this will build the MLRISC 'nowhere' tool (for translating
2637       programs with where-clauses into legal SML code) during installation.
2638    
2639    ----------------------------------------------------------------------
2640    Name: Matthias Blume
2641    Date: 2002/01/25 21:27:00 EST
2642    Tag: blume-Release_110_38-retracted
2643    Description:
2644    
2645    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
2646    
2647    README will be added later.
2648    
2649    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
2650    (Release_110_38) removed.  Reason: Last-minute bug fixes.
2651    
2652    ----------------------------------------------------------------------
2653    Name: Matthias Blume
2654    Date: 2002/01/25
2655    Tag: blume-20020125-ffi
2656    Description:
2657    
2658    A large number of tweaks and improvements to ml-nlffi-lib and
2659    ml-nlffigen:
2660    
2661       - ML represenation types have been streamlined
2662       - getter and setter functions work with concrete values, not abstract
2663         ones where possible
2664       - ml-nlffigen command line more flexible (see README file there)
2665       - some bugs have been fixed (hopefully)
2666    
2667    ----------------------------------------------------------------------
2668    Name: Lal George
2669    Date: 2002/01/24
2670    Tag: george-20020124-risc-ra-interface
2671    Description:
2672    
2673       There is a dramatic simplification in the interface to the
2674       register allocator for RISC architectures as a result of making
2675       parallel copy instructions explicit.
2676    
2677    ----------------------------------------------------------------------
2678    Name: Matthias Blume
2679    Date: 2002/01/22
2680    Tag: blume-20020122-x86-ccalls
2681    Description:
2682    
2683    Bug fix for c-calls on x86 (having to do with how char- and
2684    short-arguments are being handled).
2685    
2686    ----------------------------------------------------------------------
2687    Name: Matthias Blume
2688    Date: 2002/01/21
2689    Tag: blume-20020121-ff
2690    Description:
2691    
2692    Another day of fiddling with the FFI...
2693    
2694    1. Bug fix/workaround:  CKIT does not complain about negative array
2695       dimensions, so ml-nlffigen has to guard itself against this possibility.
2696       (Otherwise a negative dimension would send it into an infinite loop.)
2697    
2698    2. Some of the abstract types (light objects, light pointers, most "base"
2699       types) in structure C are now eqtypes.
2700    
2701    3. Added constructors and test functions for NULL function pointers.
2702    
2703    ----------------------------------------------------------------------
2704    Name: Matthias Blume
2705    Date: 2002/01/18
2706    Tag: blume-20020118-ready-for-new-release
2707    Description:
2708    
2709    Made config/srcarchiveurl point to a new place.  (Will provide boot
2710    files shortly.)
2711    
2712    Maybe we christen this to be 110.38?
2713    
2714    ----------------------------------------------------------------------
2715    Name: Matthias Blume
2716    Date: 2002/01/18
2717    Tag: blume-20020118-more-ffifiddle
2718    Description:
2719    
2720    Today's FFI fiddling:
2721    
2722      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
2723        C.Set.* that use concrete (MLRep.*) arguments and results instead
2724        of abstract ones.
2725    
2726      - Provided word-style bit operations etc. for "int" representation
2727        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
2728        and Long.
2729    
2730    ----------------------------------------------------------------------
2731    Name: Matthias Blume
2732    Date: 2002/01/18
2733    Tag: blume-20020118-use-x86-fp
2734    Description:
2735    
2736    Now that x86-fast-fp seems to be working, I turned it back on again
2737    by default.  (Seems to work fine now, even with the FFI.)
2738    
2739    Other than that, I added some documentation about the FFI to
2740    src/ml-nlffigen/README and updated the FFI test examples in
2741    src/ml-nlffi-lib/Tests/*.
2742    
2743    ----------------------------------------------------------------------
2744    Name: Allen Leung
2745    Date: 2002/01/17
2746    Tag: leunga-20020117-x86-fast-fp-call
2747    Description:
2748    
2749       1. Fixed a problem with handling return fp values when x86's fast fp
2750          mode is turned on.
2751    
2752       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
2753          of %f32.
2754    
2755       3. Added a constructor INT32lit to the ast of MLRISC tools.
2756    
2757    ----------------------------------------------------------------------
2758    Name: Matthias Blume
2759    Date: 2002/01/16
2760    Tag: blume-20020116-ffifiddle
2761    Description:
2762    
2763    More fiddling with the FFI interface:
2764    
2765     - Make constness 'c instead of rw wherever possible.  This eliminates
2766       the need for certain explicit coercions.  (However, due to ML's
2767       value polymorphism, there will still be many cases where explicit
2768       coercions are necessary.  Phantom types are not the whole answer
2769       to modeling a subtyping relationship in ML.)
2770    
2771     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
2772    
2773     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
2774       clumsy workaround for varargs problem.)
2775    
2776    ----------------------------------------------------------------------
2777    Name: Lal George
2778    Date: 2002/01/15
2779    Tag: <none>
2780    Description:
2781    
2782    1. Since COPY instructions are no longer native to the architecture,
2783       a generic functor can be used to implement the expandCopies function.
2784    
2785    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
2786       TEXT segment.
2787    
2788    ----------------------------------------------------------------------
2789    Name: Matthias Blume
2790    Date: 2002/01/15
2791    Tag: blume-20020115-ffiupdates
2792    Description:
2793    
2794    1. Fix for bug resulting in single-precision float values being returned
2795       incorrectly from FFI calls.
2796    
2797    2. Small modifications to C FFI API:
2798    
2799        - memory-allocation routines return straight objects (no options)
2800          and raise an exception in out-of-memory situations
2801        - unsafe extensions to cast between function pointers and pointers
2802          from/to ints
2803        - added structure C_Debug as an alternative to structure C where
2804          pointer-dereferencing (|*| and |*!) always check for null-pointers
2805        - added open_lib' to DynLinkage;  open_lib' works like open_lib
2806          but also takes a (possibly empty) list of existing library handles
2807          that the current library depends on
2808    
2809    ----------------------------------------------------------------------
2810    Name: Matthias Blume
2811    Date: 2002/01/10
2812    Tag: blume-20020110-newffigen
2813    Description:
2814    
2815    1. Updates to portable graph code.
2816    
2817    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
2818       more scalable now so that even huge interfaces such as the one
2819       for GTK compile in finite time and space. :-)
2820       See src/ml-nlffigen/README for details on what's new.
2821    
2822    ----------------------------------------------------------------------
2823    Name: Lal George
2824    Date: 2001/01/09 14:31:35 EST 2002
2825    Tag: george-20011206-rm-native-copy
2826    Description:
2827    
2828            Removed the native COPY and FCOPY instructions
2829            from all the architectures and replaced it with the
2830            explicit COPY instruction from the previous commit.
2831    
2832            It is now possible to simplify many of the optimizations
2833            modules that manipulate copies. This has not been
2834            done in this change.
2835    
2836    ----------------------------------------------------------------------
2837    Name: Lal George
2838    Date: 2001/12/06 16:50:13 EST 2001
2839    Tag: george-20011206-mlrisc-instruction
2840    Description:
2841    
2842    Changed the representation of instructions from being fully abstract
2843    to being partially concrete. That is to say:
2844    
2845      from
2846            type instruction
2847    
2848      to
2849            type instr                              (* machine instruction *)
2850    
2851            datatype instruction =
2852                LIVE of {regs: C.cellset, spilled: C.cellset}
2853              | KILL of {regs: C.cellset, spilled: C.cellset}
2854              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
2855              | ANNOTATION of {i: instruction, a: Annotations.annotation}
2856              | INSTR of instr
2857    
2858    This makes the handling of certain special instructions that appear on
2859    all architectures easier and uniform.
2860    
2861    LIVE and KILL say that a list of registers are live or killed at the
2862    program point where they appear. No spill code is generated when an
2863    element of the 'regs' field is spilled, but the register is moved to
2864    the 'spilled' (which is present, more for debugging than anything else).
2865    
2866    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
2867    We used to generate:
2868    
2869            DEFFREG f1
2870            f1 := f2 + f3
2871            trapb
2872    
2873    but now generate:
2874    
2875            f1 := f2 + f3
2876            trapb
2877            LIVE {regs=[f1,f2,f3], spilled=[]}
2878    
2879    Furthermore, the DEFFREG (hack) required that all floating point instruction
2880    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
2881    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
2882    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
2883    this architecture).
2884    
2885    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
2886    all the architectures. This will result in further simplification of the
2887    register allocator that must be aware of them for coalescing purposes, and
2888    will also simplify certain aspects of the machine description that provides
2889    callbacks related to parallel copies.
2890    
2891    ANNOTATION should be obvious, and now INSTR represents the honest to God
2892    machine instruction set!
2893    
2894    The <arch>/instructions/<arch>Instr.sml files define certain utility
2895    functions for making porting easier -- essentially converting upper case
2896    to lower case. All machine instructions (of type instr) are in upper case,
2897    and the lower case form generates an MLRISC instruction. For example on
2898    the alpha we have:
2899    
2900      datatype instr =
2901         LDA of {r:cell, b:cell, d:operand}
2902       | ...
2903    
2904      val lda : {r:cell, b:cell, d:operand} -> instruction
2905        ...
2906    
2907    where lda is just (INSTR o LDA), etc.
2908    
2909    ----------------------------------------------------------------------
2910    Name: Matthias Blume
2911    Date: 2001/11/22 21:40:00 EST
2912    Tag: Release_110_37
2913    Description:
2914    
2915    Release 110.37.  This time for real.
2916    
2917    ----------------------------------------------------------------------
2918    Name: Matthias Blume
2919    Date: 2001/11/21 16:35:00 EST
2920    Tag: blume-20011121-foot-in-mouth
2921    Description:
2922    
2923    Removed the "Release_110_37" tag because of a serious bug.
2924    This will be re-tagged once the bug is fixed.
2925    
2926    ----------------------------------------------------------------------
2927    Name: Matthias Blume
2928    Date: 2001/11/21 16:14:00 EST
2929    Tag: blume-20011121-forgottenfile
2930    Description:
2931    
2932    Forgot to add a file.  (Just a .tex-file -- part of
2933    the CM manual source.)
2934    
2935    ----------------------------------------------------------------------
2936    Name: Matthias Blume
2937    Date: 2001/11/21 16:10:00 EST
2938    Tag: blume-20011121-invalid_110_37
2939    Description:
2940    
2941    Note: I removed the original tag "Release_110_37" from this commit
2942          because we found a serious bug in all non-x86 backends.
2943          - Matthias
2944    
2945    1. Modifications to the SML/NJ code generator and to the runtime system
2946       so that code object name strings are directly inserted into code
2947       objects at code generation time.  The only business the runtime system
2948       has with this is now to read the name strings on occasions.
2949       (The encoding of the name string has also changed somewhat.)
2950    
2951    2. CM now implements a simple "set calculus" for specifying export lists.
2952       In particular, it is now possible to refer to the export lists of
2953       other libraries/groups/sources and form unions as well as differences.
2954       See the latest CM manual for details.
2955    
2956    3. An separate notion of "proxy" libraries has again be eliminated from
2957       CM's model.  (Proxy libraries are now simply a special case of using
2958       the export list calculus.)
2959    
2960    4. Some of the existing libraries now take advantage of the new set
2961       calculus.
2962       (Notice that not all libraries have been converted because some
2963       of the existing .cm-files are supposed to be backward compatible
2964       with 110.0.x.)
2965    
2966    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
2967       "General.exnMessage"!  The former relies on a certain hook to be
2968       initialized, and that often does not happen in the stand-alone case.)
2969    
2970    ----------------------------------------------------------------------
2971    Name: Lal George
2972    Date: 2001/11/21  13:56:18 EST
2973    Tag: george-2001121-pseudo-ops
2974    Description:
2975    
2976      Implemented a complete redesign of MLRISC pseudo-ops. Now there
2977      ought to never be any question of incompatabilities with
2978      pseudo-op syntax expected by host assemblers.
2979    
2980      For now, only modules supporting GAS syntax are implemented
2981      but more should follow, such as MASM, and vendor assembler
2982      syntax, e.g. IBM as, Sun as, etc.
2983    
2984    ----------------------------------------------------------------------
2985    Name: Matthias Blume
2986    Date: 2001/11/14 11:52:00 EST
2987    Tag: blume-20011114-srcname
2988    Description:
2989    
2990    1. Routed the name of the current source file to mlriscgen where it
2991       should be directly emitted into the code object.  (This last part
2992       is yet to be done.)
2993    
2994    2. Some cleanup of the pgraph code to make it match the proposal that
2995       I put out the other day.  (The proposal notwithstanding, things are
2996       still in flux here.)
2997    
2998    ----------------------------------------------------------------------
2999    Name: Lal George
3000    Date: 2001/11/14 09:44:04 EST
3001    Tag:
3002    Description:
3003    
3004      Fix for a backpatching bug reported by Allen.
3005    
3006      Because the boundary between short and long span-dependent
3007      instructions is +/- 128, there are an astounding number of
3008      span-dependent instructions whose size is over estimated.
3009    
3010      Allen came up with the idea of letting the size of span
3011      dependent instructions be non-monotonic, for a maxIter
3012      number of times, after which the size must be monotonically
3013      increasing.
3014    
3015      This table shows the number of span-dependent instructions
3016      whose size was over-estimated as a function of maxIter, for the
3017      file Parse/parse/ml.grm.sml:
3018    
3019         maxIter            # of instructions:
3020            10                      687
3021            20                      438
3022            30                      198
3023            40                        0
3024    
3025      In compiling the compiler, there is no significant difference in
3026      compilation speed between maxIter=10 and maxIter=40. Actually,
3027      my measurements showed that maxIter=40 was a tad faster than
3028      maxIter=10! Also 96% of the  files in the compiler reach a fix
3029      point within 13 iterations, so fixing maxIter at 40, while high,
3030      is okay.
3031    
3032    ----------------------------------------------------------------------
3033    Name: Matthias Blume
3034    Date: 2001/10/31 15:25:00 EST
3035    Tag: blume-20011031-pgraph
3036    Description:
3037    
3038    CKIT:
3039    * Changed the "Function" constructor of type Ast.ctype to carry optional
3040      argument identifiers.
3041    * Changed the return type of TypeUtil.getFunction accordingly.
3042    * Type equality ignores the argument names.
3043    * TypeUtil.composite tries to preserve argument names but gives up quickly
3044      if there is a mismatch.
3045    
3046    installation script:
3047    * attempts to use "curl" if available (unless "wget" is available as well)
3048    
3049    CM:
3050    * has an experimental implementation of "portable graphs" which I will
3051      soon propose as an implementation-independent library format
3052    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
3053    
3054    NLFFI-LIB:
3055    * some cleanup (all cosmetic)
3056    
3057    NLFFIGEN:
3058    * temporarily disabled the mechanism that suppresses ML output for
3059      C definitions whose identifiers start with an underscore character
3060    * generate val bindings for enum constants
3061    * user can request that only one style (light or heavy) is being used;
3062      default is to use both (command-line arguments: -heavy and -light)
3063    * fixed bug in handling of function types involving incomplete pointers
3064    * generate ML entry points that take record arguments (i.e., using
3065      named arguments) for C functions that have a prototype with named
3066      arguments
3067      (see changes to CKIT)
3068    
3069    ----------------------------------------------------------------------
3070    Name: Allen Leung
3071    Date: 2001/10/27 20:34:00 EDT
3072    Tag: leunga-20011027-x86-fast-fp-call
3073    Description:
3074    
3075       Fixed the bug described in blume-20010920-slowfp.
3076    
3077       The fix involves
3078          1. generating FCOPYs in FSTP in ia32-svid
3079          2. marking a CALL with the appropriate annotation
3080    
3081    ----------------------------------------------------------------------
3082    Name: Matthias Blume
3083    Date: 2001/10/16 11:32:00 EDT
3084    Tag: blume-20011016-netbsd
3085    Description:
3086    
3087    Underscore patch from Chris Richards (fixing problem with compiling
3088    runtime system under recent NetBSD).
3089    
3090    ----------------------------------------------------------------------
3091    Name: Allen Leung
3092    Date: 2001/10/12 17:18:32 EDT 2001
3093    Tag: leung-20011012-x86-printflowgraph
3094    Description:
3095    
3096    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
3097    
3098    ----------------------------------------------------------------------
3099    Name: Lal George
3100    Date: 2001/10/11 23:51:34 EDT
3101    Tag: george-20011011-too-many-instrs
3102    Description:
3103    
3104    The representation of a program point never expected to see more
3105    than 65536 instructions in a basic block!
3106    
3107    ----------------------------------------------------------------------
3108    Name: Lal George
3109    Date: 2001/10/09 09:41:37 EDT
3110    Tag: george-20011008-mlrisc-labels
3111    Description:
3112    
3113    Changed the machine description files to support printing of
3114    local and global labels in assembly code, based on host assembler
3115    conventions.
3116    
3117    ----------------------------------------------------------------------
3118    Name: Matthias Blume
3119    Date: 2001/09/25 15:25:00 EDT
3120    Tag: blume-20010925-exninfo
3121    Description:
3122    
3123    I provided a non-hook implementation of exnName (at the toplevel) and
3124    made the "dummy" implementation of exnMessage (at the toplevel) more
3125    useful: if nothing gets "hooked in", then at least you are going to
3126    see the exception name and a message indicating why you don't see more.
3127    
3128    [For the time being, programs that need exnMessage and want to use
3129    ml-build should either use General.exnMessage (strongly recommended) or
3130    refer to structure General at some other point so that CM sees a
3131    static dependency.]
3132    
3133    [Similar remarks go for "print" and "use":  If you want to use their
3134    functionality in stand-alone programs generated by ml-build, then use
3135    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
3136    
3137    ----------------------------------------------------------------------
3138    Name: Matthias Blume
3139    Date: 2001/09/20 17:28:00 EDT
3140    Tag: blume-20010920-slowfp
3141    Description:
3142    
3143    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
3144    
3145    ----------------------------------------------------------------------
3146    Name: Matthias Blume
3147    Date: 2001/09/20 17:20:00 EDT
3148    Tag: blume-20010920-canonicalpaths
3149    Description:
3150    
3151    0. Updated the BOOT file (something that I forgot to do earlier).
3152    
3153    1. Small internal change to CM so that it avoids "/../" in filenames
3154       as much as possible (but only where it is safe).
3155    
3156    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
3157       that arguments that contain delimiters are passed through correctly.
3158       This change also means that all "special" arguments of the form
3159       @SMLxxx... must come first.
3160    
3161    3. Changed install script to put relative anchor names for tool commands
3162       into pathconfig.
3163    
3164    ----------------------------------------------------------------------
3165    Name: Matthias Blume
3166    Date: 2001/09/18 15:35:00 EDT
3167    Tag: blume-20010918-readme11036
3168    Description:
3169    
3170    Added README files.
3171    
3172    ----------------------------------------------------------------------
3173    Name: Matthias Blume
3174    Date: 2001/09/18 11:45:00 EDT
3175    Tag: Release_110_36 (retag)
3176    Description:
3177    
3178    Fixed mistake in config/preloads. Retagged as 110.36.
3179    
3180    ----------------------------------------------------------------------
3181    Name: Matthias Blume
3182    Date: 2001/09/18 09:40:00 EDT
3183    Tag: Release_110_36_orig (tag changed)
3184    Description:
3185    
3186    New version (110.36).  New bootfiles.
3187    
3188    ----------------------------------------------------------------------
3189    Name: Matthias Blume
3190    Date: 2001/09/14 16:15:00 EDT
3191    Tag: blume-20010914-x86fastfp
3192    Description:
3193    
3194    John committed some changes that Allen made, in particular a (hopefully)
3195    correctly working version of the x86-fp module.
3196    
3197    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
3198    flag to "true".  Everything seems to compile to a fixpoint ok, and
3199    "mandelbrot" speeds up by about 15%.
3200    
3201    ----------------------------------------------------------------------
3202    Name: Matthias Blume
3203    Date: 2001/09/13 11:20:00 EDT
3204    Tag: blume-20010913-minimal
3205    Description:
3206    
3207    1. Stefan Monnier's patch to fix a miscompilation problem that
3208       was brought to light by John Reppy's work on Moby.
3209    
3210    2. Implemented a minimal "structure Compiler" that contains just
3211       "version" and "architecture".  The minimal version will be
3212       available when the full version is not.  This is for backward-
3213       compatibility with code that wants to test Compiler.version.
3214    
3215    ----------------------------------------------------------------------
3216    Name: Matthias Blume
3217    Date: 2001/08/28 14:03:00 EDT
3218    Tag: blume-20010828-ml-lex
3219    Description:
3220    
3221    Fix for bug 1581, received from Neophytos Michael.
3222    
3223    ----------------------------------------------------------------------
3224    Name: Matthias Blume
3225    Date: 2001/08/27 11:20:00 EDT
3226    Tag: blume-20010827-readme11035
3227    Description:
3228    
3229    Fleshed out the README file for 110.35.
3230    
3231    ----------------------------------------------------------------------
3232    Name: Matthias Blume
3233    Date: 2001/08/24 17:10:00 EDT
3234    Tag: Release_110_35
3235    Description:
3236    
3237    New version number (110.35).  New bootfiles.
3238    
3239    ----------------------------------------------------------------------
3240    Name: Lal George
3241    Date: 2001/08/24 13:47:18 EDT 2001
3242    Tag: george-20010824-MLRISC-graphs
3243    Description:
3244    
3245     removed clusters from MLRISC completely and replaced with graphs.
3246    
3247    ----------------------------------------------------------------------
3248    Name: Matthias Blume
3249    Date: 2001/08/23 17:50:00 EDT
3250    Tag: blume-20010823-toplevel
3251    Description:
3252    
3253    - some reorganization of the code that implements various kinds of
3254      environments in the compiler (static, dynamic, symbolic, combined)
3255    - re-implemented the EnvRef module so that evalStream works properly
3256      (if the stream contains references to "use", "CM.make", etc.)
3257    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
3258    
3259    ----------------------------------------------------------------------
3260    Name: Matthias Blume
3261    Date: 2001/08/20 15:50 EDT
3262    Tag: blume20010820-slipup
3263    Description:
3264    
3265    I forgot to commit a few files.  Here they are...
3266    
3267    ----------------------------------------------------------------------
3268    Name: Matthias Blume
3269    Date: 2001/08/20 15:35:00 EDT
3270    Tag: blume-20010820-debugprof
3271    Description:
3272    
3273    !!!! NEW BOOTFILES !!!!
3274    
3275    This is another round of reorganizing the compiler sources.  This
3276    time the main goal was to factor out all the "instrumentation"
3277    passes (for profiling and backtracing) into their own library.
3278    The difficulty was to do it in such a way that it does not depend
3279    on elaborate.cm but only on elabdata.cm.
3280    
3281    Therefore there have been further changes to both elaborate.cm and
3282    elabdata.cm -- more "generic" things have been moved from the former
3283    to the latter.  As a result, I was forced to split the assignment
3284    of numbers indicating "primtyc"s into two portions: SML-generic and
3285    SML/NJ-specific.  Since it would have been awkward to maintain,
3286    I bit the bullet and actually _changed_ the mapping between these
3287    numbers and primtycs.  The bottom line of this is that you need
3288    a new set of bin- and bootfiles.
3289    
3290    I have built new bootfiles for all architectures, so doing a fresh
3291    checkout and config/install.sh should be all you need.
3292    
3293    The newly created library's name is
3294    
3295        $smlnj/viscomp/debugprof.cm
3296    
3297    and its sources live under
3298    
3299        src/compiler/DebugProf
3300    
3301    ----------------------------------------------------------------------
3302    Name: Matthias Blume
3303    Date: 2001/08/15 17:15:00 EDT
3304    Tag: blume-20010815-compreorg
3305    Description:
3306    
3307    This is a first cut at reorganizing the CM libraries that make up the
3308    core of the compiler.  The idea is to separate out pieces that could
3309    be used independently by tools, e.g., the parser, the typechecker, etc.
3310    
3311    The current status is a step in this direction, but it is not quite
3312    satisfactory yet.  Expect more changes in the future.
3313    
3314    Here is the current (new) organization...
3315    
3316        What used to be $smlnj/viscomp/core.cm is now divided into
3317        six CM libraries:
3318    
3319             $smlnj/viscomp/basics.cm
3320                           /parser.cm
3321                           /elabdata.cm
3322                           /elaborate.cm
3323                           /execute.cm
3324                           /core.cm
3325    
3326        The CM files for these libraries live under src/system/smlnj/viscomp.
3327        All these libraries are proxy libraries that contain precisely
3328        one CM library component.  Here are the locations of the components
3329        (all within the src/compiler tree):
3330    
3331             Basics/basics.cm
3332             Parse/parser.cm
3333             ElabData/elabdata.cm
3334             Elaborator/elaborate.cm
3335             Execution/execute.cm
3336             core.cm
3337    
3338         [This organization is the same that has been used already
3339         for a while for the architecture-specific parts of the visible
3340         compiler and for the old version of core.cm.]
3341    
3342         As you will notice, many source files have been moved from their
3343         respective original locations to a new home in one of the above
3344         subtrees.
3345    
3346         The division of labor between the new libraries is the following:
3347    
3348             basics.cm:
3349                - Simple, basic definitions that pertain to many (or all) of
3350                  the other libraries.
3351             parser.cm:
3352                - The SML parser, producing output of type Ast.dec.
3353                - The type family for Ast is also defined and exported here.
3354             elabdata.cm:
3355                - The datatypes that describe input and output of the elaborator.
3356                  This includes types, absyn, and static environments.
3357             elaborator.cm:
3358                - The SML/NJ type checker and elaborator.
3359                  This maps an Ast.dec (with a given static environment) to
3360                  an Absyn.dec (with a new static environment).
3361                - This libraries implements certain modules that used to be
3362                  structures as functors (to remove dependencies on FLINT).
3363             execute.cm:
3364                - Everything having to do with executing binary code objects.
3365                - Dynamic environments.
3366             core.cm:
3367                - SML/NJ-specific instantiations of the elaborator and MLRISC.
3368                - Top-level modules.
3369                - FLINT (this should eventually become its own library)
3370    
3371    Notes:
3372    
3373    I am not 100% happy with the way I separated the elaborator (and its
3374    data structures) from FLINT.  Two instances of the same problem:
3375    
3376        1. Data structures contain certain fields that carry FLINT-specific
3377           information.  I hacked around this using exn and the property list
3378           module from smlnj-lib.  But the fact that there are middle-end
3379           specific fields around at all is a bit annoying.
3380    
3381        2. The elaborator calculates certain FLINT-related information.  I tried
3382           to make this as abstract as I could using functorization, but, again,
3383           the fact that the elaborator has to perform calculations on behalf
3384           of the middle-end at all is not nice.
3385    
3386        3. Having to used exn and property lists is unfortunate because it
3387           weakens type checking.  The other alternative (parameterizing
3388           nearly *everything*) is not appealing, though.
3389    
3390    I removed the "rebinding =" warning hack because due to the new organization
3391    it was awkward to maintain it.  As a result, the compiler now issues some of
3392    these warnings when compiling init.cmi during bootstrap compilation. On
3393    the plus side, you also get a warning when you do, for example:
3394       val op = = Int32.+
3395    which was not the case up to now.
3396    
3397    I placed "assign" and "deref" into the _Core structure so that the
3398    code that deals with the "lazy" keyword can find them there.  This
3399    removes the need for having access to the primitive environment
3400    during elaboration.
3401    
3402    ----------------------------------------------------------------------
3403    Name: Matthias Blume
3404    Date: 2001/08/13
3405    Tag: blume-20010813-closures
3406    Description:
3407    
3408    This fix was sent to us by Zhong Shao.  It is supposed to improve the
3409    performance of certain loops by avoiding needless closure allocation.
3410    
3411    ----------------------------------------------------------------------
3412    Name: Lal George
3413    Date: 2001/07/31 10:03:23 EDT 2001
3414    Tag: george-20010731-x86-fmalloc
3415    Description: Fixed bug in x86 calls
3416    
3417        There was a bug where call instructions would mysteriously
3418        vanish. The call instruction had to be one that returned
3419        a floating point value.
3420    
3421    ----------------------------------------------------------------------
3422    Name: Lal George
3423    Date: 2001/07/19 16:36:29 EDT 2001
3424    Tag: george-20010719-simple-cells
3425    Description:
3426    
3427    I have dramatically simplified the interface for CELLS in MLRISC.
3428    
3429    In summary, the cells interface is broken up into three parts:
3430    
3431      1. CellsBasis : CELLS_BASIS
3432    
3433            CellsBasis is a top level structure and common for all
3434            architectures.  it contains the definitions of basic datatypes
3435            and utility  functions over these types.
3436    
3437      2. functor Cells() : CELLS
3438    
3439            Cells generates an interface for CELLS that incorporates the
3440            specific resources on the target architecture, such as the
3441            presence of special register classes, their number and size,
3442            and various useful substructures.
3443    
3444      3. <ARCH>CELLS
3445    
3446            e.g. SparcCells: SPARCCELLS
3447    
3448            <ARCH>CELLS usually contains additional bindings for special
3449            registers  on the architecture, such as:
3450    
3451                    val r0 : cell           (* register zero *)
3452                    val y : cell            (* Y register *)
3453                    val psr : cell          (* processor status register *)
3454                    ...
3455    
3456            The structure returned by applying the Cells functor is opened
3457            in this interface.
3458    
3459    The main implication of all this is that the datatypes for cells is
3460    split between CellsBasis and CELLS -- a fairly simple change for user
3461    code.
3462    
3463    In the old scheme the CELLS interface had a definitional binding of
3464    the form:
3465    
3466            signature CELLS = sig
3467    
3468               structure CellsBasis = CellsBasis
3469    
3470               ...
3471    
3472            end
3473    
3474    With all the sharing constraints that goes on in MLRISC, this old
3475    design  quickly leads to errors such as:
3476    
3477            "structure definition spec inside of sharing ... "
3478    
3479    
3480    and appears to require an unacceptable amount of sharing and where
3481    constraint hackery.
3482    
3483    I think this error message (the interaction of definitional specs and
3484    sharing) requires more explanation on our web page.
3485    
3486    ----------------------------------------------------------------------
3487    Name: Matthias Blume
3488    Date: 2001/07/19 15:00:00 EDT
3489    Tag: blume-20010719-libreorg
3490    Description:
3491    
3492    This update puts together a fairly extensive but straightforward change
3493    to the way the libraries that implement the interactive system are
3494    organized:
3495    
3496       The biggest change is the elimination of structure Compiler.  As a
3497       replacement for this structure, there is now a CM library
3498       (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
3499       that exports all the substructures of the original structure Compiler
3500       directly.  So instead of saying Compiler.Foo.bar one now simply
3501       says Foo.bar.  (The CM libraries actually export a collection of
3502       structures that is richer than the collection of substructures of
3503       structure Compiler.)
3504    
3505       To make the transition smooth, there is a separate library called
3506       $smlnj/compiler/compiler.cm which puts together and exports the
3507       original structure Compiler (or at least something very close to it).
3508    
3509       There are five members of the original structure Compiler
3510       that are not exported directly but which instead became members
3511       of a new structure Backend (described by signature BACKEND).  These are:
3512       structure Profile (: PROFILE), structure Compile (: COMPILE), structure
3513       Interact (: INTERACT), structure Machine (: MACHINE), and val
3514       architecture (: string).
3515    
3516       Structure Compiler.Version has become structure CompilerVersion.
3517    
3518       Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
3519       by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
3520       or x86, respectively.
3521       Each of these exports the same frontend structures that
3522       $smlnj/compiler.cm exports.  But they do not have a structure Backend
3523       and instead export some structure <Arch>Backend where <Arch> is Alpha32,
3524       Hppa, PPC, Sparc, or X86, respectively.
3525    
3526       Library $smlnj/compiler/all.cm exports the union of the exports of
3527       $smlnj/compiler/<arch>.cm
3528    
3529       There are no structures <Arch>Compiler anymore, use
3530       $smlnj/compiler/<arch>.cm instead.
3531    
3532       Library host-compiler-0.cm is gone.  Instead, the internal library
3533       that instantiates CM is now called cm0.cm.  Selection of the host
3534       compiler (backend) is no longer done here but. (Responsibility for it
3535       now lies with $smlnj/compiler/current.cm.  This seems to be more
3536       logical.)
3537    
3538       Many individual files have been moved or renamed.  Some files have
3539       been split into multiple files, and some "dead" files have been deleted.
3540    
3541    Aside from these changes to library organization, there are also changes
3542    to the way the code itself is organized:
3543    
3544       Structure Binfile has been re-implemented in such a way that it no
3545       longer needs any knowledge of the compiler.  It exclusively deals
3546       with the details of binfile layout.  It no longer invokes the
3547       compiler (for the purpose of creating new prospective binfile
3548       content), and it no longer has any knowledge of how to interpret
3549       pickles.
3550    
3551       Structure Compile (: COMPILE) has been stripped down to the bare
3552       essentials of compilation.  It no longer deals with linking/execution.
3553       The interface has been cleaned up considerably.
3554    
3555       Utility routines for dealing with linking and execution have been
3556       moved into their own substructures.
3557    
3558       (The ultimate goal of these changes is to provide a light-weight
3559       binfile loader/linker (at least for, e.g., stable libraries) that
3560       does not require CM or the compiler to be present.)
3561    
3562    CM documentation has been updated to reflect the changes to library
3563    organization.
3564    
3565    ----------------------------------------------------------------------
3566    Name: Matthias Blume
3567    Date: 2001/07/10 17:30:00 EDT
3568    Tag: Release_110_34
3569    Description:
3570    
3571    Minor tweak to 110.34 (re-tagged):
3572    
3573      - README.html file added to CVS repository
3574      - runtime compiles properly under FreeBSD 3.X and 4.X
3575    
3576    ----------------------------------------------------------------------
3577    Name: Matthias Blume
3578    Date: 2001/07/10 17:30:00 EDT
3579    Tag: Release_110_34
3580    Description:
3581    
3582    New version number (110.34). New bootfiles.
3583    
3584    ----------------------------------------------------------------------
3585    Name: Matthias Blume
3586    Date: 2001/07/09 16:00:00 EDT
3587    Tag: blume-20010709-more-varargs
3588    Description:
3589    
3590    I changed the handling of varargs in ml-nlffigen again:
3591    The ellipsis ... will now simply be ignored (with an accompanying warning).
3592    
3593    The immediate effect is that you can actually call a varargs function
3594    from ML -- but you can't actually supply any arguments beyond the ones
3595    specified explicitly.  (For example, you can call printf with its format
3596    string, but you cannot pass additional arguments.)
3597    
3598    This behavior is only marginally more useful than the one before, but
3599    it has the advantage that a function or, more importantly, a function
3600    type never gets dropped on the floor, thus avoiding follow-up problems with
3601    other types that refer to the offending one.
3602    
3603    ----------------------------------------------------------------------
3604    Name: Matthias Blume
3605    Date: 2001/07/09 11:25:00 EDT
3606    Tag: blume-20010709-varargs
3607    Description:
3608    
3609    1. ckit-lib.cm now exports structure Error
3610    2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
3611       with a warning accompanied by a source location.  Moreover, it
3612       merely skips the offending function or type and proceeds with the
3613       rest of its work.u  As a result, one can safely feed C code containing
3614       "..." to ml-nlffigen.
3615    3. There are some internal improvements to CM, providing slightly
3616       more general string substitutions in the tools subsystem.
3617    
3618    ----------------------------------------------------------------------
3619    Name: Matthias Blume
3620    Date: 2001/06/27 15:10:00 EDT
3621    Tag: blume-20010627-concur
3622    Description:
3623    
3624    Fixed a small bug in CM's handling of parallel compilation.
3625    (You could observe the bug by Control-C-interrupting an ordinary
3626    CMB.make or CM.stabilize and then attaching some compile servers.
3627    The result was that all of a sudden the previously interrupted
3628    compilation would continue on its own.  This was because of
3629    an over-optimization: CM did not bother to clean out certain queues
3630    when no servers were attached "anyway", resulting in the contents
3631    of these queues to grab control when new servers did get attached.)
3632    
3633    There is also another minor update to the CM manual.
3634    
3635    ----------------------------------------------------------------------
3636    Name: Matthias Blume
3637    Date: 2001/06/26 16:15:00 EDT
3638    Tag: blume-20010626-cmdoc
3639    Description:
3640    
3641    Minor typo fixed in CM manual (syntax diagram for libraries).
3642    
3643    ----------------------------------------------------------------------
3644    Name: Matthias Blume
3645    Date: 2001/06/25 22:55:00 EDT
3646    Tag: blume-20010625-x86pc
3647    Description:
3648    
3649    Fixed a nasty bug in the X86 assembly code that caused signal
3650    handlers to fail (crash) randomly.
3651    
3652    ----------------------------------------------------------------------
3653    Name: Matthias Blume
3654    Date: 2001/06/25 12:05:00 EDT
3655    Tag: blume-20010625-nlffigen
3656    Description:
3657    
3658    This update fixes a number of minor bugs in ml-nlffigen as reported by
3659    Nick Carter <nbc@andrew.cmu.edu>.
3660    
3661      1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
3662      2. Default names for generated files are now derived from the name of
3663         the C file *without its directory*.  In particular, this causes generated
3664         files to be placed locally even if the C file is in some system directory.
3665      3. Default names for generated signatures and structures are also derived
3666         from the C file name without its directory.  This avoids silly things
3667         like "structure GL/GL".
3668         (Other silly names are still possible because ml-nlffigen does not do
3669          a thorough check of whether generated names are legal ML identifiers.
3670          When in doubt, use command line arguments to force particular names.)
3671    
3672    ----------------------------------------------------------------------
3673    Name: Matthias Blume
3674    Date: 2001/06/21 12:25:00 EDT
3675    Tag: blume-20010621-eXene
3676    Description:
3677    
3678    eXene now compiles and (sort of) works again.
3679    
3680    The library name (for version > 110.33) is $/eXene.cm.
3681    
3682    I also added an new example in src/eXene/examples/nbody.  See the
3683    README file there for details.
3684    
3685    ----------------------------------------------------------------------
3686    Name: Matthias Blume
3687    Date: 2001/06/20 16:40:00 EDT
3688    Tag: blume-20010620-cml
3689    Description:
3690    
3691    CML now compiles and works again.
3692    
3693    Libraries (for version > 110.33):
3694    
3695      $cml/cml.cm            Main CML library.
3696      $cml/basis.cm          CML's version of $/basis.cm.
3697      $cml/cml-internal.cm   Internal helper library.
3698      $cml/core-cml.cm       Internal helper library.
3699      $cml-lib/trace-cml.cm  Tracing facility.
3700      $cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
3701    
3702    The installer (config/install.sh) has been taught how to properly
3703    install this stuff.
3704    
3705    ----------------------------------------------------------------------
3706    Name: Matthias Blume
3707    Date: 2001/06/19 17:55:00 EDT
3708    Tag: blume-20010619-instantiate
3709    Description:
3710    
3711    This un-breaks the fix for bug 1432.
3712    (The bug was originally fixed in 110.9 but I broke it again some
3713    time after that.)
3714    
3715    ----------------------------------------------------------------------
3716    Name: Matthias Blume
3717    Date: 2001/06/19 17:25:00 EDT
3718    Tag: blume-20010619-signals
3719    Description:
3720    
3721    This should (hopefully) fix the long-standing signal handling bug.
3722    (The runtime system was constructing a continuation record with an
3723    incorrect descriptor which would cause the GC to drop data on the floor...)
3724    
3725    ----------------------------------------------------------------------
3726    Name: Matthias Blume
3727    Date: 2001/06/15 15:05:00 EDT
3728    Tag: blume-20010615-moresparc
3729    Description:
3730    
3731    Here is a short late-hour update related to Sparc c-calls:
3732    
3733     -- made handling of double-word arguments a bit smarter
3734    
3735     -- instruction selection phase tries to collapse certain clumsily
3736        constructed ML-Trees; typical example:
3737    
3738            ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
3739    
3740        This currently has no further impact on SML/NJ since mlriscGen does
3741        not seem to generate such patterns in the first place, and c-calls
3742        (which did generate them in the beginning) has meanwhile been fixed
3743        so as to avoid them as well.
3744    
3745    ----------------------------------------------------------------------
3746    Name: Matthias Blume
3747    Date: 2001/06/15 15:05:00 EDT
3748    Tag: blume-20010615-sparc
3749    Description:
3750    
3751    The purpose of this update is to provide an implementation of NLFFI
3752    on Sparc machines.
3753    
3754    Here are the changes in detail:
3755    
3756       * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
3757       the Sparc implementation of the c-calls API.
3758       * The Sparc backend of SML/NJ has been modified to uniformely use %fp
3759       for accessing the ML frame.  Thus, we have a real frame pointer and
3760       can freely modify %sp without need for an omit-frame-ptr phase.
3761       The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
3762       this case.
3763       * ml-nlffigen has been taught to produce code for different architectures
3764       and calling conventions.
3765       * In a way similar to what was done in the x86 case, the Sparc
3766       backend uses its own specific extension to mltree.  (For example,
3767       it needs to be able to generate UNIMP instructions which are part
3768       of the calling convention.)
3769       * ml-nlffi-lib was reorganized to make it more modular (in particular,
3770       to make it easier to plug in new machine- and os-dependent parts).
3771    
3772    There are some other fairly unrelated bug fixes and cleanups as well:
3773    
3774       * I further hacked the .cm files for MLRISC tools (like MDLGen) so
3775       that they properly share their libraries with existing SML/NJ libraries.
3776       * I fixed a minor cosmetic bug in CM, supressing certain spurious
3777       follow-up error messages.
3778       * Updates to CM/CMB documentation.
3779    
3780    TODO items:
3781    
3782       * MLRISC should use a different register as its asmTemp on the Sparc.
3783         (The current %o2 is a really bad choice because it is part of the
3784         calling conventions, so things might interfere in unexpected ways.)
3785    
3786    ----------------------------------------------------------------------
3787    Name: Matthias Blume
3788    Date: 2001/06/07
3789    Tag: blume-20010607-calls
3790    Description:
3791    
3792    A number of internal changes related to C calls and calling conventions:
3793    
3794    1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
3795       number of bytes popped implicitly (by the callee).  In most cases
3796       this field is 0 but on x86/win32 it is some non-zero value.  This
3797       is information provided for the benefit of the "omit-frameptr" pass.
3798    2. The CALL instruction on the x86 carries a similar "pops" field.
3799       The instruction selection phase copies its value from the ML-Tree
3800       CALL statement.
3801    3. On all other architectures, the instruction selection phase checks
3802       whether "pops=0" and complains if not.
3803    4. The c-calls implementation for x86 now accepts two calling conventions:
3804       "ccall" and "stdcall".  When "ccall" is selected, the caller cleans
3805       up after the call and pops is set to 0.  For "stdcall", the caller
3806       does nothing, leaving the cleanup to the callee; pops is set to
3807       the number of bytes that were pushed onto the stack.
3808    5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
3809       distinguish between "ccall" and "stdcall".
3810    6. The UNIMP instruction has been added to the supported Sparc instruction
3811       set. (This is needed for implementing the official C calling convention
3812       on this architecture.)
3813    7. I fixed some of the .cm files under src/MLRISC/Tools to make them
3814       work with the latest CM.
3815    
3816    ----------------------------------------------------------------------
3817    Name: Matthias Blume
3818    Date: 2001/06/05 15:10:00 EDT
3819    Tag: blume-20010605-cm-index
3820    Description:
3821    
3822    0. The "lambdasplit" parameter for class "sml" in CM has been documented.
3823    
3824    1. CM can now generate "index files".  These are human-readable files
3825       that list on a per-.cm-file basis each toplevel symbol defined or
3826       imported.  The location of the index file for
3827       <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
3828       To enable index-file generation, set CM.Control.generate_index to true
3829       or export an environment-symbol: export CM_GENERATE_INDEX=true.
3830    
3831       The CM manual has been updated accordingly.
3832    
3833    2. I made some slight modifications to the c-calls API in MLRISC.
3834    
3835         a) There is now a callback to support saving/restoring of
3836            dedicated but caller-save registers around the actual call
3837            instruction.
3838         b) One can o