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/NOTES/HISTORY
ViewVC logotype

Diff of /sml/trunk/NOTES/HISTORY

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 939, Tue Sep 25 19:25:48 2001 UTC revision 1593, Thu Aug 5 22:19:45 2004 UTC
# Line 8  Line 8 
8  The form of an entry should be:  The form of an entry should be:
9    
10  Name:  Name:
11  Date:  Date: yyyy/mm/dd
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14    
15  ----------------------------------------------------------------------  ----------------------------------------------------------------------
16    Name: Allen Leung (leunga (at) reservoir (dot) com)
17    Date: 2004/08/04 18:17:00 EST
18    Tag: leunga-110_47-ppc-ibm-asm
19    Description:
20    
21      - Added minimal IBM assembly syntax support for PowerPC.
22    
23      - Cygwin: manually changed the file cygwin.def.  Some exported symbols have
24        been altered in the runtime.  We need an automatic way to keep the file
25        in sync.
26    
27    ----------------------------------------------------------------------
28    Name: Matthias Blume (blume (at) tti - c (dot) org)
29    Date: 2004/08/04 14:00:00 CDT
30    Tag: Release_110_47
31    Description:
32    
33    New working version (110.47).  NEW BOOTFILES!
34    
35    ----------------------------------------------------------------------
36    Name: Matthias Blume (blume (at) tti - c (dot) org)
37    Date: 2004/08/03 14:25:00 CDT
38    Tag: blume-20040803-callingconv
39    Description:
40    
41    Added low-level support for choosing C calling conventions by
42    twiddling the type of rawccall.  (See
43    src/compiler/Semant/types/cproto.sml for details.)
44    
45    ----------------------------------------------------------------------
46    Name: Matthias Blume (blume (at) tti - c (dot) org)
47    Date: 2004/08/02 15:55:00 CDT
48    Tag: blume-20040802-backout
49    Description:
50    
51    Backed out of change to win32-filesys.c.  The earlier patch to
52    get_file_time caused CM to produce files with the wrong time stamp.
53    
54    ----------------------------------------------------------------------
55    Name: Matthias Blume (blume (at) tti - c (dot) org)
56    Date: 2004/08/02 14:45:00 CDT
57    Tag: blume-20040802-nlffi-win32
58    Description:
59    
60    Added NLFFI support for Win32, adapted from a patch provided by David
61    Hansel.  This is currently completely untested.  Also, the issue
62    concerning stdcall vs. ccall is still unresolved.
63    
64    ----------------------------------------------------------------------
65    Name: Matthias Blume (blume (at) tti - c (dot) org)
66    Date: 2004/07/30 17:55:00 CDT
67    Tag: blume-20040730-various
68    Description:
69    
70    Gearing up towards 110.47...
71    
72    - various minor bugfixes to ml-nlffigen
73    - a beginning of a manual for nlffi
74    
75    - eliminated 'export name=value' in config/install.sh as this does
76      not work with certain versions of /bin/sh
77      (Thanks to David King at Motorola for catching this.)
78    
79    - several bugfixes provided or suggested by David Hansel at Reactive Systems:
80      - added a test for tm==NULL to gmtime.c and localtime.c
81      - applied patch for incorrect GetFileTime under win32
82      - toSeconds -> toMilliseconds in Win32/win32-process.sml
83    
84    ----------------------------------------------------------------------
85    Name: Matthias Blume (blume (at) tti - c (dot) org)
86    Date: 2004/07/21 18:20:00 CDT
87    Tag: blume-20040721-nlffigen
88    Description:
89    
90    - Fixed minor issue in ml-nlffigen: Now generate structure T_foo
91      for a typedef to an incomplete type, but leave out the "typ" member.
92      (This is just for consistency.)
93    - Started to produce what is supposed to become better (i.e., comprehensive)
94      documentation of what ml-nlffigen does and produces.
95    
96    ----------------------------------------------------------------------
97    Name: Matthias Blume (blume (at) tti - c (dot) org)
98    Date: 2004/07/14 16:25:00 CDT
99    Tag: blume-20040714-union
100    Description:
101    
102    Added C_UNION to c-calls/c-types.sml and updated the machinery
103    (ml-nlffigen, cproto.sml) that conveys C function interface
104    information to the code generator.
105    
106    However, the actual architecture-specific implementation of function
107    arguments and results that are C unions is still not implemented.
108    
109    ----------------------------------------------------------------------
110    Name: Allen Leung (leunga (at) reservoir (dot) com)
111    Date: 2004/07/14 14:38:00 EST
112    Tag: leunga-110_46_1-ppc-lwzu
113    Description:
114    
115        Added these instructions to the PowerPC architecture:
116        LBZU(X), LHZU(X), LWZU(X),
117        STWU(X), STFDU, STFSU
118    
119        etc...
120    
121        Note: I haven't added their instruction encoding into the description.
122    
123    ----------------------------------------------------------------------
124    Name: Allen Leung (leunga (at) reservoir (dot) com)
125    Date: 2004/07/13 15:04:00 EST
126    Tag: leunga-110_46_1-ppc-lwarx
127    Description:
128    
129       Added the two instructions LWARX and STWCX to the PowerPC
130    instruction set.
131    
132       A (untested) rewrite of loop-structure.sml.  The old version
133    is completely broken.
134    
135    ----------------------------------------------------------------------
136    Name: Matthias Blume (blume (at) tti - c (dot) org)
137    Date: 2004/07/13 13:50:00 CDT
138    Tag: blume-20040713-nlffi
139    Description:
140    
141    - use paramAlloc to report c-calls with too many arguments
142      (for PPC version where parameter area is pre-allocated)
143    - added ccall_maxargspace to machspec (to implement the above)
144    - made "make" commend in CM's "make" tool configurable
145    - added option (default: on) for passing the name of the SML/NJ's "bin"
146      directory to "make"; the call looks like this:
147    
148         make <options> SMLNJ_BINDIR=<dir> <target>
149    
150      This can be used by the Makefile to, e.g., pick the "right" version
151      of ml-nlffigen.
152    - minor code tweaks
153    
154    ----------------------------------------------------------------------
155    Name: Matthias Blume (blume (at) tti - c (dot) org)
156    Date: 2004/07/12 22:50:00 CDT
157    Tag: blume-110_46_1-macosx-nlffi
158    Description:
159    
160    NLFFI under Mac OS X now working (sort of).  This is largely untested,
161    though.
162    
163    Note:  1. You have to make a new, clean build of the runtime system.
164           2. There are new BOOTFILES, you have to use them!
165              (Doing the bootstrap process yourself would be *very* painful!
166               If you absolutely have to do it, build the system under
167               a different architecture and then cross-compile.)
168    
169    Version bumped to 110.46.1 to account for runtime data format changes.
170    
171    ----------------------------------------------------------------------
172    Name: Matthias Blume (blume (at) tti - c (dot) org)
173    Date: 2004/06/18 14:30:00 CDT
174    Tag: blume-20040618-unix
175    Description:
176    
177    Changed the implementation of structure Unix so that the same stream
178    is returned every time one of the {text,bin}{In,Out}streamOf functions
179    is invoked on the same proc.  This is not what the spec currently
180    says -- although IMO it arguably should.  (See discussion below.)
181    
182    ----------------------------------------------------------------------
183    Name: Matthias Blume (blume (at) tti - c (dot) org)
184    Date: 2004/06/17 18:15:00 CDT
185    Tag: Release_110_46
186    Description:
187    
188    New working version (110.46).  NEW BOOTFILES!
189    
190    ----------------------------------------------------------------------
191    Name: Matthias Blume (blume (at) tti - c (dot) org)
192    Date: 2004/06/17 17:20:00 CDT
193    Tag: blume-20040617-timer-unix
194    Description:
195    
196    Changed the interface of structures Timer and Unix to match the most
197    recent Basis spec.
198    
199    In the case of Unix there still seems to be an open/weird issue:
200    
201       The {text,bin}{In,Out}streamOf functions are supposed to create
202       fresh streams whenever they are called -- as opposed to have them
203       return the same stream every time.  This design is supposed to
204       prevent space leaks caused by proc values hanging on to streams.
205    
206       The reap function, on the other hand, is supposed to close the
207       streams.  This cannot be done without having a handle on the
208       stream in proc after all...
209    
210    I took the liberty to implement the following stopgap solution:
211    
212      The proc value hangs on to the most recently created stream(s).
213      Reap closes those.  If either or both of the two streams hadn't
214      been created at all yet, then reap will close the corresponding
215      file descriptors directly.
216    
217    PS: I don't understand the original space leak argument anymore.  If
218    a proc hangs on to the imperative stream, then I/O operations on those
219    will advance the state of the cached stream and avoid the space leak.
220    
221    ----------------------------------------------------------------------
222    Name: Matthias Blume (blume (at) tti - c (dot) org)
223    Date: 2004/05/28 16:45:00 CDT
224    Tag: blume-20040528-basis
225    Description:
226    
227    Added signature PACK_REAL and exported functor PrimIO.
228    
229    ----------------------------------------------------------------------
230    Name: Matthias Blume (blume (at) tti - c (dot) org)
231    Date: 2004/05/25 16:00:00 CDT
232    Tag: blume-20040525-group-owner
233    Description:
234    
235    CM now ignores (but still accepts) the "owner" information in group
236    descriptions.  The owner of a group is its next enclosing
237    library. Each group must have a unique owner.  (There is a virtual
238    "toplevel" library that owns groups which are not nested within a real
239    library.)  Previously, each group had to explicitly declare its owner,
240    and CM would check that such a declaration is correct.  The new scheme
241    is to have CM check that for each group there is precisely one owning
242    library.
243    
244    The advantage of the new scheme is that the programmer no longer needs
245    to maintain the somewhat annoying owner information.  The downside is
246    that CM cannot enforce the ownership rule across multiple runs of
247    CM.make.  Fortunately, enclosing the same group in two different
248    libraries A and B which are not part of the same program does not
249    cause real problems.
250    
251    ----------------------------------------------------------------------
252    Name: Matthias Blume (blume (at) tti - c (dot) org)
253    Date: 2004/05/20 16:00:00 CDT
254    Tag: blume-20040520-win32
255    Description:
256    
257    Made the win32 version work again.  (Strangely, a misplaced comma had
258    slipped into win32-process.c which prevented the runtime from being
259    compiled correctly.)
260    
261    Also, included a minor addition to ml-build.bat analogous to what was
262    done in blume-20040519-ml-build.
263    
264    ----------------------------------------------------------------------
265    Name: Matthias Blume (blume (at) tti - c (dot) org)
266    Date: 2004/05/19 22:10:00 CDT
267    Tag: blume-20040519-ml-build
268    Description:
269    
270    Arranged for ml-build to clean up after itself a little bit better.
271    The script generates a temporary SML source file and compiles it using
272    CM, so CM generates metadata (GUID, SKEL, objectfile) for it.  It now
273    gets rid of those at the end, so they don't accumulate under .cm.
274    
275    This required a minor change to install.sh because the name of the
276    metadata directory (default: .cm) is actually configurable at
277    installation time.
278    
279    ----------------------------------------------------------------------
280    Name: Matthias Blume (blume (at) tti - c (dot) org)
281    Date: 2004/05/18 15:50:00 CDT
282    Tag: blume-20040518-mkreader
283    Description:
284    
285    Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
286    implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
287    
288    ----------------------------------------------------------------------
289    Name: Matthias Blume (blume (at) tti - c (dot) org)
290    Date: 2004/05/11 14:35:00 CDT
291    Tag: blume-20040511-win32sock
292    Description:
293    
294    Added previously missing support for many socket-related functions
295    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
296    for the voluminous patch!
297    
298    (I have not tested this patch under win32 yet.)
299    
300    Here is David's e-mail:
301    
302    Hi,
303    
304    Attached to this email you find a diff against sml/nj 110.45
305    that will enable socket support under Windows.
306    
307    To apply the patch (using unix or cygwin)
308    1) gunzip runtime.diff.gz
309    2) "cd" into "src/runtime" in the source tree of a fresh
310       110.45 installation.
311    3) patch -p 1 < [your/path/to]runtime.diff
312    
313    The code compiles fine but has NOT yet been extensively tested.
314    I only ran a few tests for basic socket client functionality
315    (which worked fine).  Especially the functions that use ioctl
316    are not tested at all and might not work (see below).
317    
318    I implemented this since we want to move to a newer version of sml/nj
319    but need socket support in order to use it.  This is the first time I
320    even had a look at the sml/nj source,  so please review my changes
321    before making this part of the distribution!  Here are a few issues
322    that I think might be better for someone to solve who is more
323    familiar with the sml/nj source (and socket programming):
324    
325    - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
326      exception since I could not figure out what the windows equivalent
327      of these functions is
328    
329    - In sockets-osdep.h there are a some #include statements that are
330      only used in a few files that include sockets-osdep.h
331    
332    - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
333      does not process its return value since I don't know how to
334      report an error upwards.
335    
336    - It would probably be good to have a call to WSACleanup() when
337      the library is unloaded (if there is such a possibility).
338      Otherwise I think Windows will take care of this automatically
339      when the process finishes.
340    
341    - I used ioctlsocket() as a replacement for ioctl() but I have
342      no idea if that is actually the proper replacement on Windows.
343    
344    - All these issues are marked in the code by "FIXME" comments.
345    
346    We use sml/nj extensively in our products and are quite happy
347    with it.  I hope this contribution will help you.
348    
349    Keep up the good work!
350    
351    David
352    
353    ----------------------------------------------------------------------
354    Name: Matthias Blume (blume (at) tti - c (dot) org)
355    Date: 2004/05/11 14:20:00 CDT
356    Tag: blume-20040511-installml
357    Description:
358    
359    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
360    the patch.)
361    
362    ----------------------------------------------------------------------
363    Name: Matthias Blume (blume (at) tti - c (dot) org)
364    Date: 2004/05/11 14:05:00 CDT
365    Tag: blume-20040511-nlffi-netbsd
366    Description:
367    
368    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
369    the patch.)
370    
371    ----------------------------------------------------------------------
372    Name: Matthias Blume (blume (at) tti - c (dot) org)
373    Date: 2004/05/11 12:05:00 CDT
374    Tag: blume-20040511-exports
375    Description:
376    
377    As per request by Adam Chlipala <adam@hcoop.net>, extended various
378    export lists in compiler-related .cm-files.
379    
380    ----------------------------------------------------------------------
381    Name: Matthias Blume (blume (at) tti - c (dot) org)
382    Date: 2004/05/11 11:35:00 CDT
383    Tag: blume-20040511-allsource
384    Description:
385    
386    The installer now honors the "src-smlnj" target again, although its meaning
387    has changed from "all sources required for the compiler" to "all sources
388    the installer knows about".  In other words, if you enable "src-smlnj"
389    in the "targets" file, then the installer will pull in sources for
390    everything.  (Notice that this refers to source code only.  Compiled
391    code is still only installed for modules that were requested explicitly
392    or which are required for other modules that were requested explicitly.)
393    
394    ----------------------------------------------------------------------
395    Name: Matthias Blume (blume (at) tti - c (dot) org)
396    Date: 2004/04/23 17:40:00 CDT
397    Tag: blume-20040423-ieee-scan
398    Description:
399    
400    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
401    in the exponent calculation we get INF or ZERO (depending on the mantissa
402    and the sign of the exponent).
403    
404    ----------------------------------------------------------------------
405    Name: Matthias Blume (blume (at) tti - c (dot) org)
406    Date: 2004/04/23 10:40:00 CDT
407    Tag: blume-20040423-ml-build
408    Description:
409    
410    The ml-build script now terminates with a non-0 status when something
411    goes wrong.
412    
413    ----------------------------------------------------------------------
414    Name: Matthias Blume (blume (at) tti - c (dot) org)
415    Date: 2004/04/22 16:35:00 CDT
416    Tag: blume-20040422-Option
417    Description:
418    
419    Made exception Option to be the same as exception Option.Option
420    (as it should be).
421    
422    ----------------------------------------------------------------------
423    Name: Allen Leung (leunga (at) reservoir (dot) com)
424    Date: 2004/03/19 14:40:00 EST
425    Tag: leunga-20040319-cygwin-nlffi
426    Description:
427    
428    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
429    of SML/NJ.  The problem is that
430    
431       lib = dlopen(NULL, ...)
432       f   = dlsym(lib, "malloc");
433    
434    does not work on Windows unless we explicitly export symbols
435    such as 'malloc' during linking.  We fixed this by explicitly
436    exporting the required symbols with the magic gcc incantation:
437    
438        -Wl,--export-all cygwin.def
439    
440    where cygwin.def is a file containing all the symbols that
441    we wish to export.
442    
443    I suspect this is a Windows problem and we'll have to
444    do the same (somehow with windows compilers) when
445    we build the native win32 version with the system
446    calls LoadLibrary/GetProcAddress.
447    
448    ----------------------------------------------------------------------
449    Name: Matthias Blume (blume (at) tti - c (dot) org)
450    Date: 2004/03/04 16:35:00 CST
451    Tag: blume-20040304-intinf-fmt
452    Description:
453    
454    Fixed problem with IntInf.fmt (sign would show up on the right instead
455    of on the left for BIN, OCT, and HEX).
456    
457    ----------------------------------------------------------------------
458    Name: Matthias Blume (blume (at) tti - c (dot) org)
459    Date: 2004/03/04 11:25:00 CST
460    Tag: blume-20040304-symlinks
461    Description:
462    
463    Fixed problem with installer script (unix only) where bin/ml-yacc and
464    friends pointed (via symlinks) to absolute locations instead of just
465    .run-sml.  This was reported by Vesa A Norrman.
466    
467    ----------------------------------------------------------------------
468    Name: Matthias Blume (blume (at) tti - c (dot) org)
469    Date: 2004/02/13 14:50:00 CST
470    Tag: Release_110_45
471    Description:
472    
473    New working version (110.45). New bootfiles.
474    
475    ----------------------------------------------------------------------
476    Name: Matthias Blume (blume (at) tti - c (dot) org)
477    Date: 2004/01/26 15:15:15 CST
478    Tag: blume-20040126-toplevel
479    Description:
480    
481    Improved handling of exceptions at the interactive toplevel.
482    
483    ----------------------------------------------------------------------
484    Name: Matthias Blume (blume (at) tti - c (dot) org)
485    Date: 2004/01/26 11:25:00
486    Tag: blume-20040126-app
487    Description:
488    
489    Type of top-level "app" corrected.
490    Added code for setting vp_limitPtrMask to Win32-specific runtime.
491    
492    ----------------------------------------------------------------------
493    Name: Matthias Blume (blume (at) tti - c (dot) org)
494    Date: 2003/11/18 17:10 CST
495    Tag: blume-20031118-basis-fiddle
496    Description:
497    
498    - changed Timer interface to what might become the spec
499    - POSIX_FLAGS -> BIT_FLAGS according to spec
500    - some other minor discrepancies wrt. spec eliminated
501    
502    ----------------------------------------------------------------------
503    Name: Matthias Blume (blume (at) tti - c (dot) org)
504    Date: 2003/11/06 12:00:00 CST
505    Tag: Release_110_44
506    Description:
507    
508    New working version (110.44). New bootfiles.
509    
510    ----------------------------------------------------------------------
511    Name: Matthias Blume (blume (at) tti - c (dot) org)
512    Date: 2003/11/04 11:50:00 CST
513    Tag: blume-20031104-move-libraries
514    Description:
515    
516    Eliminated the "dont_move_libraries" directive in config/targets.
517    (The mechanism was broken and could not be fixed easily.  Moreover,
518    there does not seem to be any reason not to move all libraries into
519    lib during installation.  I originally implemented this directive as a
520    backward-compatibility feature when I first introduced the new CM.
521    Now that things have been stable for a long time and going back to the
522    old CM is not an option, there is no reason to keep it around.)
523    
524    ----------------------------------------------------------------------
525    Name: Matthias Blume (blume (at) tti - c (dot) org)
526    Date: 2003/11/03 16:00:00 CST
527    Tag: blume-20031103-installdir
528    Description:
529    
530    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
531    Richards for pointing out the problem and providing the solution.)
532    
533    ----------------------------------------------------------------------
534    Name: Matthias Blume (blume (at) tti - c (dot) org)
535    Date: 2003/10/01 17:05:00 CDT
536    Tag: blume-20031001-lal-mlrisc
537    Description:
538    
539    MLRISC bug fix from Lal.
540    
541    ----------------------------------------------------------------------
542    Name: Matthias Blume (blume (at) tti - c (dot) org)
543    Date: 2003/09/30 16:10:00 CDT
544    Tag: blume-20030930-primio-bat
545    Description:
546    
547    1. Added openVector, nullRd, and nullWr to PRIM_IO.
548    2. Improved .bat files (for Win32 port) to make things work under Win95.
549       (thanks to Aaron S. Hawley for this one)
550    
551    ----------------------------------------------------------------------
552    Name: Matthias Blume (blume (at) tti - c (dot) org)
553    Date: 2003/09/26 16:05:00 CDT
554    Tag: blume-20030926-wrappriv
555    Description:
556    
557    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
558    
559    ----------------------------------------------------------------------
560    Name: Matthias Blume (blume (at) tti - c (dot) org)
561    Date: 2003/09/26 15:00:00 CDT
562    Tag: blume-20030926-110_43_3
563    Description:
564    
565    - additional cleanup
566    - version number bump, NEW BOOTFILES
567    
568    ----------------------------------------------------------------------
569    Name: Matthias Blume (blume (at) tti - c (dot) org)
570    Date: 2003/09/26 12:00:00 CDT
571    Tag: blume-20030926-ppautoload
572    Description:
573    
574    I modified the read-eval-print loop so that the autoloader gets
575    invoked whenever the prettyprinter tries to look up a symbol that
576    is not currently defined in the toplevel environment but which
577    appears in CM's autoload registry.  As a result, we see far fewer of
578    those ?.Foo.Bar.xxx names in the prettyprinter's output.
579    
580    In addition to this I tried to clean up some pieces of the Basis
581    implementation (e.g., Socket, Word8Array) in order to prevent other
582    instances of these ?.Foo.Bar.xxx names from being printed.
583    
584    The mechanism that picks names for types still needs some work, though.
585    (Right now it seems that if there is a type A.t which is defined to
586    be B.u, but B is unavailable at toplevel, then A.t gets printed as
587    "?.B.u" although the perhaps more sensible solution would be to use
588    "A.t" in this case.  In other words, the prettyprinter should follow
589    a chain of DEFtycs not farther than there are corresponding toplevel
590    names in the current environment.)
591    
592    ----------------------------------------------------------------------
593    Name: Matthias Blume (blume (at) tti - c (dot) org)
594    Date: 2003/09/24 16:31:00 CDT
595    Tag: blume-20030924-installer
596    Description:
597    
598    Another installer tweak:  All the ML code for the installer is now
599    compiled during CMB.make and put into a little library called
600    $smlnj/installer.cm.  The installation then simply invokes
601    
602       sml -m $smlnj/installer.cm
603    
604    and everything happens automagically.
605    
606    Win32:  ML code senses value of environment variable SMLNJ_HOME.
607    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
608            and BINDIR.
609    
610    The new scheme guarantees that the ML code responsible for the installation
611    is in sync with the APIs of the main system.  Also, the installer is
612    somewhat faster because the installer script is precompiled.
613    
614    ----------------------------------------------------------------------
615    Name: Matthias Blume (blume (at) tti - c (dot) org)
616    Date: 2003/09/24 15:35:00 CDT
617    Tag: blume-20030924-synsock
618    Description:
619    
620    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
621    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
622    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
623    implementations of non-blocking operations from CML's Socket
624    structure.
625    
626    ----------------------------------------------------------------------
627    Name: Matthias Blume (blume (at) tti - c (dot) org)
628    Date: 2003/09/24 15:10:05 CDT
629    Tag: blume-20030924-sockets
630    Description:
631    
632    1. Fixed SOCKET API and implementation to match Basis spec.
633       This required changing the internal representation of sockets to one
634       that remembers (for each socket file descriptor) whether it is currently
635       blocking or non-blocking.  This state is maintained lazily (i.e., a system
636       call is made only if the state actually needs to change).
637    
638    2. OS-specific details of sockets were moved into separate files, thus
639       making it possible to unify the bulk of the socket implementations
640       between Unix and Win32.
641    
642    3. CML's socket API changed accordingly.
643       (Note that we need to remove non-blocking functions from this API
644       since they are redundant in the case of CML!)
645    
646    4. CML's socket implementation now makes use of non-blocking functions
647       provided by Basis, thus removing all OS-dependent code from this part
648       of CML.
649    
650    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
651    
652    ----------------------------------------------------------------------
653    Name: Matthias Blume (blume (at) tti - c (dot) org)
654    Date: 2003/09/22 12:10:00 CDT
655    Tag: blume-20030922-110_43_2
656    Description:
657    
658    Made a new interim version and bootfiles for developer's bootstrapping
659    convenience.
660    
661       110.43.2 -- NEW BOOTFILES
662    
663    ----------------------------------------------------------------------
664    Name: Matthias Blume (blume (at) tti - c (dot) org)
665    Date: 2003/09/19 15:55:00 CDT
666    Tag: blume-20030919-cmdir
667    Description:
668    
669    1. new-install.sh -> install.sh
670    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
671    3. tweaked installer so that another name instead of .cm can be chosen
672       at install time (by setting the CM_DIR_ARC environment variable
673       during installation); once installation is complete, the name is
674       fixed
675    
676    ----------------------------------------------------------------------
677    Name: Matthias Blume (blume (at) tti - c (dot) org)
678    Date: 2003/09/18 16:00:00 CDT
679    Tag: blume-20030918-110_43_1
680    Description:
681    
682    Made a new interim version and bootfiles for developer's bootstrapping
683    convenience.
684    
685       110.43.1 -- NEW BOOTFILES
686    
687    ----------------------------------------------------------------------
688    Name: Matthias Blume (blume (at) tti - c (dot) org)
689    Date: 2003/09/18 15:20:00 CDT
690    Tag: blume-20030918-misc
691    Description:
692    
693    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
694       this seems to be controversial at the moment) and used those in
695       Posix.ProcEnv.times.
696    
697    2. Added Time.{from,to}Nanoseconds to Time.
698    
699    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
700       For example, fromLargeInt never needs to look at more than 3 "big
701       digits" to get its 53 bits of precision.
702    
703    ----------------------------------------------------------------------
704    Name: Matthias Blume (blume (at) tti - c (dot) org)
705    Date: 2003/09/17 16:30:00 CDT
706    Tag: blume-20030917-real32-slices
707    Description:
708    
709    Added an entry to the primitive environment
710    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
711    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
712    
713    Removed some of the "magic" constants in real64.sml and replaced them
714    with code that generates these values from their corresponding
715    integer counterparts.
716    
717    Made all(?) the slice-related changes to the Basis and made everything
718    compile again...
719    
720    ----------------------------------------------------------------------
721    Name: Matthias Blume (blume (at) tti - c (dot) org)
722    Date: 2003/09/15 17:45:00 CDT
723    Tag: blume-20030915-rbase
724    Description:
725    
726    Fixed bug in Real.fromLargeInt.
727    
728    ----------------------------------------------------------------------
729    Name: Matthias Blume (blume (at) tti - c (dot) org)
730    Date: 2003/09/13 18:11:00 CDT
731    Tag: blume-20030913-libinstall
732    Description:
733    
734    Minor bugfix in config/libinstall (set anchor with path to
735    standalone tool after installing it, otherwise libraries that
736    need ml-lex or ml-yacc won't compile the first time the installer
737    runs).
738    
739    ----------------------------------------------------------------------
740    Name: Matthias Blume (blume (at) tti - c (dot) org)
741    Date: 2003/09/12 11:45:00 CDT
742    Tag: blume-20030912-various
743    Description:
744    
745     - fixed bug in Real.toLargeInt
746     - fixed bug in Posix.ProcEnv.times
747     - changed inputLine functions to return an option
748     - minor installer improvements / bugfixes
749     - changed default @SMLalloc parameter for x86/celeron to 64k
750    
751    ----------------------------------------------------------------------
752    Name: Matthias Blume (blume (at) tti - c (dot) org)
753    Date: 2003/09/09 22:00:00 CDT
754    Tag: Release_110_43
755    Description:
756    
757        New working release 110.43.  New bootfiles.
758    
759    ----------------------------------------------------------------------
760    Name: Matthias Blume (blume (at) tti - c (dot) org)
761    Date: 2003/09/09 19:20:00 CDT
762    Tag: blume-20030909-installer
763    Description:
764    
765    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
766    Modified config/install.bat to take advantage of it.  Also modified
767    config/install.sh (and called it config/new-install.sh) to take advantage
768    of it on Unix systems.  (The SML code is (supposed to be) platform-
769    independent.)
770    
771    The installer can now install everything under Win32
772    as well as under *nix as long as it compiles.
773    
774    Other changes:
775    
776      - made CML compile again under Win32
777      - made eXene compile under Win32 (by providing a fake structure UnixSock
778        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
779      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
780        same as type int;  under Win32 it isn't
781      - fixed some slice-related problems in the win32-specific parts of CML
782      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
783        (under Win32, the volume name is case-insensitive, and the
784        OS.Path code compares volume names for equality)
785    
786    ----------------------------------------------------------------------
787    Name: Matthias Blume (blume (at) tti - c (dot) org)
788    Date: 2003/09/08 11:55:00 CDT
789    Tag: blume-20030908-fullpath
790    Description:
791    
792    Made Win32 version of OS.FileSys.fullPath return current directory
793    when given an empty string.  This is what the spec says, and incidentally,
794    CM depends on it.  (CM otherwise goes into an infinite loop in certain
795    cases when presented with the name of a non-existing .cm file.)
796    
797    ----------------------------------------------------------------------
798    Name: Matthias Blume (blume (at) tti - c (dot) org)
799    Date: 2003/09/04 16:30:00 CDT
800    Tag: blume-20030905-slices-etc
801    Description:
802    
803      1. Changed interface to vectors and arrays in Basis to match
804         (draft) Basis spec.
805      2. Added signatures and implementations of slices according to
806         Basis spec.
807      3. Edited source code throughout the system to make it compile again
808         under 1. and 2.  (In some cases code had to be added to have it
809         match the new signatures.)
810      4. MLRISC should be backward-compatible: the copies of the originals
811         of files that needed to change under 3. were retained, the .cm files
812         check the compiler version number and use old versions when
813         appropriate.
814      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
815         dirstream -> string option (in accordance with Basis spec).
816      6. When generating code that counts lines, ml-lex used function
817         CharVector.foldli, taking advantage of its old interface.
818         This has been replaced with the corresponding code from
819         CharVectorSlice.  (html-lex must be re-lexed!)
820      7. BitArray in smlnj-lib/Util has been extended/modified to match the
821         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
822      8. Removed temporary additions (fromInternal, toInternal) from the
823         (now obsolete) IntInf in smlnj-lib/Util.
824      9. Cleaned up structure Byte.
825     10. Added localOffset, scan, and fromString to Date (according to spec).
826         Cleaned/corrected implementation of Date.
827         (Still need to check for correctness; implement better canonicalizeDate.)
828     11. Added "scan" to signature IEEE_REAL.
829     12. Some improvements to IntInf [in particular: efficiency-hack for
830         mod and rem when second operand is 2 (for parity checks).]
831     13. Changed representation of type Time.time, using a single IntInf.int
832         value counting microseconds.  This considerably simplified the
833         implementation of structure Time.  We now support negative time
834         values; scan and fromString handle signs.
835     14. Functor PrimIO now takes two additional arguments (VectorSlice and
836         ArraySlice).
837    
838    ----------------------------------------------------------------------
839    Name: Matthias Blume (blume (at) tti - c (dot) org)
840    Date: 2003/08/28 17:00:00 CDT
841    Tag: blume-20030828-intinf
842    Description:
843    
844    This is a major update which comes with a version number bump
845    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
846    and an implementation of IntInf in the Basis.
847    
848    There are a fairly large number of related changes and updates throughout
849    the system:
850    
851    Basis:
852            - Implemented IntInf.
853            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
854            - Added some missing Real64 operations, most notably Real.toLargeInt.
855            - Added FixedInt as a synonym for Int32.
856    
857    compiler:
858            * Added support for a built-in intinf type.
859              - literals
860              - pattern matching
861              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
862              - overloading on literals and operations
863    
864            This required adding a primitive type intinf, some additional
865            primops, and implementations for several non-trivial intinf
866            operations in Core.  (The intinf type is completely abstract
867            to the compiler; all operations get delegated back to the Core.)
868    
869            * Intinf equality is handled by polyequal.  However, the compiler
870            does not print its usual warning in this case (since polyequal
871            is the right thing to do there).
872    
873            * Improved the organization of structure InlineT.
874    
875            * A word about conversion primops:
876            If conversions involving intinf do not cancel out during
877            CPS contract, then the compiler must insert calls to Core functions.
878            Since all core access must be resolved already during the FLINT
879            translate phase, it would be too late a the time of CPS contract
880            to add new Core calls.  For this reason, conversion primops
881            for intinf carry two arguments:  1. the numeric argument that
882            they are supposed to convert, and 2. the Core function that
883            can help with this conversion if necessary.  If CPS contract
884            eliminates a primop, then the associated Core function becomes
885            dead and goes away.  Intinf conversion primops that do not get
886            eliminated by CPS contract get rewritten into calls of their
887            core functions by a separate, new phase.
888    
889    interactive system:
890            - Control.Print.intinfDepth controls max length of intinf constants
891            being printed. (Analogous to Control.Print.stringDepth.)
892            - Cleanup in printutil and pputil: got rid of unused stuff and
893            duplicates; replaced some of the code with code that makes better
894            use of library functionality.
895    
896    CM:
897            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
898            are no longer silent.
899    
900    CKIT:
901            Fixed mismatched uses of Int32 and LargeInt.  I always decided
902            in favor of LargeInt -- which is now the same as IntInf.
903            CKIT-knowledgable people should check whether this is what's
904            intended and otherwise change things back to using Int32 or
905            FixedInt.
906    
907    Throughout the code:
908            Started using IntInf.int literals and built-in operations
909            (e.g., comparison with 0) where this seems appropriate.
910    
911    
912    ----------------------------------------------------------------------
913    Name: Dave MacQueen (dbm@cs.uchicago.edu)
914    Date: 2003/08/13 11:36:00 CDT
915    Tag: dbm-20030813-mcz-merge1
916    Description:
917    
918    Merging changes from the mcz-branch development branch into trunk.
919    These changes involve replacement of the emulated old prettyprinter
920    interface with direct use of the SML/NJ Lib PP library, and fixing
921    of a couple of bugs (895, 1186) relating to error messages.  A new
922    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
923    has been added.
924    
925    ----------------------------------------------------------------------
926    Name: Matthias Blume (blume (at) tti - c (dot) org)
927    Date: 2003/08/11 15:45:00 CDT
928    Tag: blume-20030811-windows
929    Description:
930    
931    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
932    
933        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
934    
935    This patch restores SML/NJ's ability to run under win32.  There are a
936    number of changes, including fixes for several bugs that had gone
937    unnoticed until now:
938    
939      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
940        trying to run the win32 version from within cygwin.)
941    
942      - There are a number of simple .bat scripts that substitute for their
943        corresponding Unix shell-scripts.  (See below.)
944    
945      - The internals of ml-build have been modified slightly.  The main
946        difference is that instead of calling ".link-sml" (or link-sml.bat)
947        using OS.Process.system, the ML process delegates this task back
948        to the script.  Otherwise problems arise in mixed environments such
949        as Cygwin where scripts look and work like Unix scripts, but
950        where OS.Process.system cannot run them.
951    
952      - In CM, the srcpath pickler used native pathname syntax -- which
953        is incorrect in the case of cross-compilation.  The new pickle format
954        is independent of platform-specific naming conventions.
955    
956      - Path configuration files (such as lib/pathconfig) can now choose
957        between native and standard syntax.  Placing a line of the form
958    
959            standard!
960    
961        into the file causes all subsequent paths to be interpreted using
962        CM standard pathname syntax (= Unix conventions); a line
963    
964            native!
965    
966        switches back to native style.  This was needed so that
967        path config files can be written portably, see src/system/pathconfig.
968    
969      - Runtime system:
970    
971          - win32-filesys.c:  get_file_time and set_file_time now
972            access modification time, not creation time.
973    
974          - I/O code made aware of new array representation.
975    
976          - Bug fixes in X86.prim.masm.
977    
978       - src/system/makeml made aware of win32. (For use under cygwin
979            and other Unix-environments for windows.)
980    
981       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
982         which caused BinIO.inputAll to fail consistently.
983    
984    .bat scripts:
985    
986       Windows .bat scripts assume that SMLNJ_HOME is defined.
987    
988       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
989         applications (sml, ml-yacc, ml-lex).
990       - ml-build.bat: analogous to ml-build.
991       - config\install.bat:  Analogous to config/install.sh.  This requires
992         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
993         (nmake etc. must be on the path, and vcvars32 must have been run.)
994         Moreover, sources for ml-lex and ml-yacc need to exist under src,
995         and the bootfile hierarchy must have been unpacked under
996         sml.boot.x86-win32.
997         The script is very primitive and does a poor job at error checking.
998         It only installs the base system, ml-lex, and ml-yacc.  No other
999         libraries are being installed (i.e., you get only those that
1000         are part of the compiler.)
1001       - link-sml.bat:  analogous to .link-sml, but not currently used
1002    
1003    Unrelated bug fixes:
1004    
1005      - ml-nlffigen now exports structures ST_* corresponding to incomplete
1006        types.
1007      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
1008        otherwise.)
1009    
1010    ----------------------------------------------------------------------
1011    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
1012    Date: 2003/06/17
1013    Tag: macqueen-20030617-bug895
1014    Description:
1015    
1016    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
1017    Tag will be used for new development branch (mcz-branch) for
1018    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
1019    
1020    ----------------------------------------------------------------------
1021    Name: Matthias Blume (blume (at) tti - c (dot) org)
1022    Date: 2003/05/27 16:55:00 CDT
1023    Tag: blume-20030527-polyeq
1024    Description:
1025    
1026    Tried to eliminated most cases of polymorphic equality.
1027    
1028    ----------------------------------------------------------------------
1029    Name: Matthias Blume (blume (at) tti - c (dot) org)
1030    Date: 2003/05/21 17:45:00 CDT
1031    Tag: blume-20030517-complete
1032    Description:
1033    
1034    Two changes:
1035    
1036      1. Added a flag for controlling whether non-exhaustive bindings will
1037         be treated as errors (default is false).
1038      2. Cleaned up the *entire* source tree so that CMB.make goes through
1039         without a single non-exhaustive match- or bind warning.
1040    
1041    ----------------------------------------------------------------------
1042    Name: Matthias Blume (blume (at) tti - c (dot) org)
1043    Date: 2003/05/17 10:20:00 CDT
1044    Tag: blume-20030517-absyn
1045    Description:
1046    
1047    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
1048    
1049       This mainly affects the quality of error messages.  However, some
1050       of the code is now more straightforward than before. (Treatment of
1051       the above four constructs in translate.sml is much simpler than
1052       the "macro-expansion" that was going on before.  Plus, the mach-
1053       compiler no longer gets invoked just to be able to compile an
1054       if-expression.)
1055    
1056    2. The ErrorMsg.Error exception is now caught and absorbed by the
1057       interactive loop.
1058    
1059    ----------------------------------------------------------------------
1060    Name: Allen Leung
1061    Date: 2003/05/16 13:05:00 CDT
1062    Tag: leunga-20030516-cygwin-runtime
1063    Description:
1064    
1065        Ported the runtime system to cygwin, which uses the unix
1066    x86-unix bin files.  Missing/buggy features:
1067    
1068       o getnetbyname, getnetbyaddr: these functions seem to be missing in
1069         the Cygwin library.
1070       o Ctrl-C handling may be flaky.
1071       o Windows system calls and Windows I/O are not supported.
1072    
1073       A new set of binfiles is located at:
1074    
1075         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
1076    
1077       This is only needed for bootstrapping the cygwin version of smlnj.
1078       Other x86 versions can use the existing binfiles.
1079    
1080    ----------------------------------------------------------------------
1081    Name: Matthias Blume
1082    Date: 2003/04/08 15:42:00 CDT
1083    Tag: blume-20030408-listpair
1084    Description:
1085    
1086    1. Added a target 'mlrisc' to installer.
1087    
1088    2. Added missing elements to structure ListPair.
1089    
1090    ----------------------------------------------------------------------
1091    Name: Allen Leung
1092    Date: 2003/01/07 10:40:00 EST
1093    Tag: leunga-20030107-int-rem
1094    Description:
1095    
1096    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
1097    The arguments to the SUBL instruction were swapped.
1098    
1099    ----------------------------------------------------------------------
1100    Name: Matthias Blume
1101    Date: 2002/12/12 16:25:00 EST
1102    Tag: blume-20021212-risc-ra
1103    Description:
1104    
1105    Fixed a serious bug in the rewrite code for FP spilling/reloading that
1106    sent the RA into an infinite loop when floating point registers get
1107    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
1108    110.37 and 110.38.)
1109    There was another set of potential problems related to the handling of
1110    MLRISC annotations (but those did not yet cause real problems, apparently).
1111    
1112    ----------------------------------------------------------------------
1113    Name: Matthias Blume
1114    Date: 2002/12/06 22:40:00 EST
1115    Tag: blume-20021206-cm-fileid
1116    Description:
1117    
1118    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
1119    This fixes the problem of CM getting confused by files that suddenly
1120    change their identity (e.g., by getting unlinked and recreated by some
1121    text editor such as vi).  There might be a better/cheaper/cleaner way
1122    of doing this, but for now this will have to do.
1123    
1124    ----------------------------------------------------------------------
1125    Name: Matthias Blume
1126    Date: 2002/10/28 09:50:00 EST
1127    Tag: blume-20021028-typecheck
1128    Description:
1129    
1130    Exported structure Typecheck from $smlnj/viscomp/core.cm.
1131    
1132    ----------------------------------------------------------------------
1133    Name: Matthias Blume
1134    Date: 2002/10/17 09:10:00 EDT
1135    Tag: Release_110_42
1136    Description:
1137    
1138    In good old tradition, there has been a slight hiccup so that we have
1139    to patch 110.42 after the fact.   The old release tag has been replaced
1140    (see below).
1141    
1142    The change solves a problem with two competing approaches the
1143    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
1144    each other's way.
1145    
1146    This change only affects the runtime system code and the installer script.
1147    (No new bootfiles.)
1148    
1149    ----------------------------------------------------------------------
1150    Name: Matthias Blume
1151    Date: 2002/10/16 12:00:00 EDT
1152    Tag: Release_110_42_removed
1153    Description:
1154    
1155    New working release.  New bootfiles.
1156    
1157    ----------------------------------------------------------------------
1158    Name: Matthias Blume
1159    Date: 2002/10/10 13:10:00 EDT
1160    Tag: blume-20021010-ppc-divs
1161    Description:
1162    
1163    The mltree operator DIVS must be implemented with an overflow check on
1164    the PPC because the hardware indicates divide-by-zero using "overflow" as
1165    well.
1166    
1167    ----------------------------------------------------------------------
1168    Name: Matthias Blume
1169    Date: 2002/07/23 11:45:00
1170    Tag: blume-20020723-smlnj-home
1171    Description:
1172    
1173    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
1174    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
1175    CM_PATHCONFIG is also set), the path configuration file is assumed
1176    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
1177    the entire tree to some other place and everything will "just work".
1178    
1179    (Companion commands such as ml-build and ml-makedepend also sense this
1180    variable.)
1181    
1182    ----------------------------------------------------------------------
1183    Name: Matthias Blume
1184    Date: 2002/07/12 21:19:00 EDT
1185    Tag: blume-20020712-liveness
1186    Description:
1187    
1188    Exported two useful "step" functions from liveness module (MLRISC).
1189    
1190    ----------------------------------------------------------------------
1191    Name: Matthias Blume
1192    Date: 2002/07/05 16:00 EDT
1193    Tag: Release_110_41
1194    Description:
1195    
1196    New working release.  New bootfiles.
1197    
1198    ----------------------------------------------------------------------
1199    Name: Matthias Blume
1200    Date: 2002/07/05 10:25:00 EDT
1201    Tag: blume-20020705-btimp
1202    Description:
1203    
1204    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
1205    other clients can set up backtracing support.
1206    
1207    ----------------------------------------------------------------------
1208    Name: Matthias Blume
1209    Date: 2002/06/25 17:23:00 EDT
1210    Tag: blume-20020625-fpmax
1211    Description:
1212    
1213    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
1214    case.  (The sense of the isNaN test was reversed -- which made min and
1215    max always return their first argument.)
1216    
1217    ----------------------------------------------------------------------
1218    Name: Matthias Blume
1219    Date: 2002/06/11
1220    Tag: blume-20020611-unixpath
1221    Description:
1222    
1223    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
1224    
1225    ----------------------------------------------------------------------
1226    Name: Matthias Blume
1227    Date: 2002/06/10 16:35:00 EDT
1228    Tag: blume-20020610-ieeereal
1229    Description:
1230    
1231    I back-ported my implementation of IEEEReal.fromString from the
1232    idlbasis-devel branch so that we can test it.
1233    
1234    Another small change is that ppDec tries to give more information
1235    than just "<sig>" in the case of functors.  However, this code is
1236    broken in some mysterious way if the functor's body's signature
1237    has not been declared by ascription but gets inferred from the
1238    implementation.  This needs fixing...
1239    
1240    ----------------------------------------------------------------------
1241    Name: Matthias Blume
1242    Date: 2002/05/31
1243    Tag: blume-20020531-btrace-mode
1244    Description:
1245    
1246    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
1247    the wayside when I switched over to using Controls everywhere.)
1248    
1249    ----------------------------------------------------------------------
1250    Name: Lal George
1251    Date: 2002/05/23 12:21:40 EDT
1252    Tag: george-20020523-visual-labels
1253    Description:
1254    
1255    Labels are now displayed in the graphical output to make
1256    the fall-through and target blocks obvious.
1257    
1258    ----------------------------------------------------------------------
1259    Name: Matthias Blume
1260    Date: 2002/05/22 11:03:00 EDT
1261    Tag: blume-20020522-shrink
1262    Description:
1263    
1264    John tweaked yesterday's fix for 1131 to handle an out-of-memory
1265    situation that comes up when allocating huge arrays.
1266    
1267    ----------------------------------------------------------------------
1268    Name: Matthias Blume
1269    Date: 2002/05/21 16:00:00 EDT
1270    Tag: Release_110_40
1271    Description:
1272    
1273    New working release (110.40).  New bootfiles.
1274    
1275    [Also: John Reppy fixed GC bug 1131.]
1276    
1277    ----------------------------------------------------------------------
1278    Name: Matthias Blume
1279    Date: 2002/05/21 12:35:00 EDT
1280    Tag: blume-20020521-cmdoc
1281    Description:
1282    
1283    CM documentation update.
1284    
1285    ----------------------------------------------------------------------
1286    Name: Matthias Blume
1287    Date: 2002/05/21 10:55:00 EDT
1288    Tag: blume-20020521-misc
1289    Description:
1290    
1291    - John tweaked runtime to be silent on heap export (except when
1292      GC messages are on).
1293    - I added a few more things (cross-compiling versions of CMB) to
1294      config/preloads (as suggestions).
1295    
1296    ----------------------------------------------------------------------
1297    Name: Matthias Blume
1298    Date: 2002/05/20 22:25:00 EDT
1299    Tag: blume-20020520-controls
1300    Description:
1301    
1302    - Added ControlUtil structure to control-lib.cm.
1303    - Use it throughout.
1304    - Used Controls facility to define MLRISC controls (as opposed to
1305      registering MLRISC control ref cells with Controls after the fact)
1306    - Fixed messed-up controls priorities.
1307    
1308    * Removed again all the stuff from config/preloads that one wouldn't
1309      be able to preload at the time the initial heap image is built.
1310      (Many libraries, e.g., CML, do not exist yet at this time.  The
1311       only libraries that can be preloaded via config/preloads are those
1312       that come bundled with the bootfiles.)
1313    
1314    ----------------------------------------------------------------------
1315    Name: Matthias Blume
1316    Date: 2002/05/20 10:59:00 EDT
1317    Tag: blume-20020520-preloads
1318    Description:
1319    
1320    Added a lot of commented-out suggestions for things to be included
1321    in config/preloads.
1322    
1323    ----------------------------------------------------------------------
1324    Name: Allen Leung
1325    Date: 2002/05/18 14:20:00 EDT
1326    Tag: leunga-20020518-mdl
1327    Description:
1328    
1329    o Made the mdl tool stuff compile and run again.
1330    o I've disabled all the stuff that depends on RTL specifications; they
1331      are all badly broken anyway.
1332    
1333    ----------------------------------------------------------------------
1334    Name: Matthias Blume
1335    Date: 2002/05/17 16:49:00 EDT
1336    Tag: blume-20020517-controls
1337    Description:
1338    
1339    0. John Reppy made several modifications to the SML/NJ library.
1340       In particular, there is a shiny new controls-lib.cm.
1341    
1342    1. Pushed new controls interface through compiler so that everything
1343       compiles again.
1344    
1345    2. Added FormatComb and FORMAT_COMB to the CML version of the
1346       SML/NJ library (so that CML compiles again).
1347    
1348    3. Modified init scripts because XXX_DEFAULT environment variables
1349       are no longer with us.  (Boot-time initialization is now done
1350       using the same environment variables that are also used for
1351       startup-time initialization of controls.)
1352    
1353    ----------------------------------------------------------------------
1354    Name: Lal George
1355    Date: 2002/05/15 09:20:10 EDT
1356    Tag: george-20020515-pseudo-op-decls
1357    Description:
1358    
1359            All pseudo-ops emitted before the first segment declaration
1360    such as TEXT, DATA, and BSS directives are assumed to be global
1361    declarations and are emitted first in the assembly file. This is
1362    useful in a number of situations where one has pseudo-ops that are not
1363    specific to any segment, and also works around the constraint that one
1364    cannot have client pseudo-ops in the TEXT segment.
1365    
1366            Because no segment is associated with these declarations it is
1367    an error to allocate any space or objects before the first segment
1368    directive and an exception will be raised. However, we cannot make
1369    this check for client pseudo-ops.
1370    
1371            These top level declarations are a field in the CFG graph_info.
1372    In theory you can continue to add to this field after the CFG has been
1373    built -- provided you know what you are doing;-)
1374    
1375    ----------------------------------------------------------------------
1376    Name: Matthias Blume
1377    Date: 2002/05/13 16:40:00 EDT
1378    Tag: blume-20020513-pp-etc
1379    Description:
1380    
1381    A few minor bugfixes:
1382    
1383      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1384        (Bogus printouts for redefined bindings still occur.  Compiler
1385        bug should no longer occur now.  We need to redo the prettyprinter
1386        from scratch.)
1387    
1388      - CM pathname printer now also adds escape sequences for ( and )
1389    
1390      - commend and docu fixes for ml-nlffi
1391    
1392    ----------------------------------------------------------------------
1393    Name: Matthias Blume
1394    Date: 2002/05/10 16:40:00 EDT
1395    Tag: blume-20020510-erg-textio
1396    Description:
1397    
1398    Applied the following bugfix provided by Emden Gansner:
1399    
1400        Output is corrupted when outputSubstr is used rather than output.
1401    
1402        The problem occurs when a substring
1403    
1404            ss = (s, dataStart, dataLen)
1405    
1406        where dataStart > 0, fills a stream buffer with avail bytes left.
1407        avail bytes of s, starting at index dataStart, are copied into the
1408        buffer, the buffer is flushed, and then the remaining dataLen-avail
1409        bytes of ss are copied into the beginning of the buffer. Instead of
1410        starting this copy at index dataStart+avail in s, the current code
1411        starts the copy at index avail.
1412    
1413        Fix:
1414        In text-io-fn.sml, change line 695 from
1415             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1416        to
1417             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1418    
1419    ----------------------------------------------------------------------
1420    Name: Matthias Blume
1421    Date: 2002/04/12 13:55:00 EDT
1422    Tag: blume-20020412-assyntax
1423    Description:
1424    
1425    1. Grabbed newer assyntax.h from the XFree86 project.
1426    2. Fiddled with how to compile X86.prim.asm without warnings.
1427    3. (Very) Minor cleanup in CM.
1428    
1429    ----------------------------------------------------------------------
1430    Name: Matthias Blume
1431    Date: 2002/04/01 (no joke!) 17:07:00 EST
1432    Tag: blume-20020401-x86div
1433    Description:
1434    
1435    Added full support for div/mod/rem/quot on the x86, using the machine
1436    instruction's two results (without clumsily recomputing the remainder)
1437    directly where appropriate.
1438    
1439    Some more extensive power-of-two support was added to the x86 instruction
1440    selector (avoiding expensive divs, mods, and muls where they can be
1441    replaced with cheaper shifts and masks).  However, this sort of thing
1442    ought to be done earlier, e.g., within the CPS optimizer so that
1443    all architectures benefit from it.
1444    
1445    The compiler compiles to a fixed point, but changes might be somewhat
1446    fragile nevertheless.  Please, report any strange things that you might
1447    see wrt. div/mod/quot/rem...
1448    
1449    ----------------------------------------------------------------------
1450    Name: Matthias Blume
1451    Date: 2002/03/29 17:22:00
1452    Tag: blume-20020329-div
1453    Description:
1454    
1455    Fixed my broken div/mod logic.  Unfortunately, this means that the
1456    inline code for div/mod now has one more comparison than before.
1457    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1458    The problem was with quotient = 0, because that alone does not tell
1459    us which way the rounding went.  One then has to look at whether
1460    remainder and divisor have the same sign...  :(
1461    
1462    Anyway, I replaced the bootfiles with fresh ones...
1463    
1464    ----------------------------------------------------------------------
1465    Name: Matthias Blume
1466    Date: 2002/03/29 14:10:00 EST
1467    Tag: blume-20020329-inlprims
1468    Description:
1469    
1470    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1471    
1472    Primops have changed. This means that the bin/boot-file formats have
1473    changed as well.
1474    
1475    To make sure that there is no confusion, I made a new version.
1476    
1477    
1478    CHANGES:
1479    
1480    * removed REMT from mltree (remainder should never overflow).
1481    
1482    * added primops to deal with divisions of all flavors to the frontend
1483    
1484    * handled these primops all the way through so they map to their respective
1485      MLRISC support
1486    
1487    * used these primops in the implementation of Int, Int32, Word, Word32
1488    
1489    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1490    
1491    * parameterized INLMIN, INLMAX, and INLABS by a numkind
1492    
1493    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1494      floating point
1495    
1496    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1497      and Real (but Real.abs maps to a separate floating-point-only primop)
1498    
1499    
1500    TODO items:
1501    
1502    * Hacked Alpha32 instruction selection, disabling the selection of REMx
1503      instructions because the machine instruction encoder cannot handle
1504      them.  (Hppa, PPC, and Sparc instruction selection did not handle
1505      REM in the first place, and REM is supported by the x86 machine coder.)
1506    
1507    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1508      selection phase.  (The two can be streamlined because the hardware
1509      delivers both quotient and remainder at the same time anyway.)
1510    
1511    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1512      (Currently the behavior is inconsistent both across architectures and
1513      wrt. the draft Basis spec.)
1514    
1515    * Word8 should eventually be handled natively, too.
1516    
1517    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1518      as if there currently is no execution path that could trigger it in
1519      SML/NJ.  (The assumptions underlying functions arith and promotable do not
1520      hold for things like multiplication and division.)
1521    
1522    ----------------------------------------------------------------------
1523    Name: Matthias Blume
1524    Date: 2002/03/27 16:27:00 EST
1525    Tag: blume-20020327-mlrisc-divisions
1526    Description:
1527    
1528    Added support for all four division operations (ML's div, mod, quot,
1529    and rem) to MLRISC.  In the course of doing so, I also rationalized
1530    the naming (no more annoying switch-around of DIV and QUOT), by
1531    parameterizing the operation by div_rounding_mode (which can be either
1532    DIV_TO_ZERO or DIV_TO_NEGINF).
1533    
1534    The generic MLTreeGen functor takes care of compiling all four
1535    operations down to only round-to-zero div.
1536    
1537    Missing pieces:
1538    
1539      * Doing something smarter than relying on MLTreeGen on architectures
1540        like, e.g., the x86 where hardware division delivers both quotient and
1541        remainder at the same time.  With this, the implementation of the
1542        round-to-neginf operations could be further streamlined.
1543    
1544      * Remove inlining support for div/mod/rem from the frontend and replace it
1545        with primops that get carried through to the backend.  Do this for all
1546        int and word types.
1547    
1548    ----------------------------------------------------------------------
1549    Name: Matthias Blume
1550    Date: 2002/03/25 17:25:00 EST
1551    Tag: blume-20020325-divmod
1552    Description:
1553    
1554    I improved (hopefully without breaking them) the implementation of Int.div,
1555    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1556    advantage of the following observations:
1557    
1558      Let  q = x quot y      r = x rem y
1559           d = x div  y      m = x mod y
1560    
1561    where "quot" is the round-to-zero version of integer division that
1562    hardware usually provides.  Then we have:
1563    
1564         r = x - q * y        where neither the * nor the - will overflow
1565         d = if q >= 0 orelse x = q * y then q else q - 1
1566                              where neither the * nor the - will overflow
1567         m = if q >= 0 orelse r = 0 then r else r + y
1568                              where the + will not overflow
1569    
1570    This results in substantial simplification of the generated code.
1571    The following table shows the number of CFG nodes and edges generated
1572    for
1573            fun f (x, y) = x OPER y
1574            (* with OPER \in div, mod, quot, rem *)
1575    
1576    
1577        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1578        --------------------------------------------------------
1579         div |         24 |         39 |         12 |         16
1580         mod |         41 |         71 |         12 |         16
1581        quot |          8 |         10 |          8 |         10
1582         rem |         10 |         14 |          8 |         10
1583    
1584    
1585    ----------------------------------------------------------------------
1586    Name: Matthias Blume
1587    Date: 2002/03/25 22:06:00 EST
1588    Tag: blume-20020325-cprotobug
1589    Description:
1590    
1591    Fixed a bug in cproto (c prototype decoder).
1592    
1593    ----------------------------------------------------------------------
1594    Name: Matthias Blume
1595    Date: 2002/03/25 16:00:00 EST
1596    Tag: blume-20020325-raw-primops
1597    Description:
1598    
1599    I did some cleanup to Allen's new primop code and
1600    replaced yesterday's bootfiles with new ones.
1601    (But they are stored in the same place.)
1602    
1603    ----------------------------------------------------------------------
1604    Name: Matthias Blume
1605    Date: 2002/03/24 22:40:00 EST
1606    Tag: blume-20020324-bootfiles
1607    Description:
1608    
1609    Made the bootfiles that Allen asked for.
1610    
1611    ----------------------------------------------------------------------
1612    Name: Allen Leung
1613    Date: 2002/03/23 15:50:00 EST
1614    Tag: leunga-20020323-flint-cps-rcc-primops
1615    Description:
1616    
1617      1. Changes to FLINT primops:
1618    
1619        (* make a call to a C-function;
1620         * The primop carries C function prototype information and specifies
1621         * which of its (ML-) arguments are floating point. C prototype
1622         * information is for use by the backend, ML information is for
1623         * use by the CPS converter. *)
1624      | RAW_CCALL of { c_proto: CTypes.c_proto,
1625                       ml_args: ccall_type list,
1626                       ml_res_opt: ccall_type option,
1627                       reentrant : bool
1628                     } option
1629       (* Allocate uninitialized storage on the heap.
1630        * The record is meant to hold short-lived C objects, i.e., they
1631        * are not ML pointers.  With the tag, the representation is
1632        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1633        *)
1634      | RAW_RECORD of {tag:bool,sz:int}
1635      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1636    
1637      2.  These CPS primops are now overloaded:
1638    
1639           rawload of {kind:numkind}
1640           rawstore of {kind:numkind}
1641    
1642          The one argument form is:
1643    
1644             rawload {kind} address
1645    
1646          The two argument form is:
1647    
1648             rawload {kind} [ml object, byte-offset]
1649    
1650      3. RAW_CCALL/RCC now takes two extra arguments:
1651    
1652         a. The first is whether the C call is reentrant, i.e., whether
1653            ML state should be saved and restored.
1654         b. The second argument is a string argument specifying the name of
1655            library and the C function.
1656    
1657         These things are currently not handled in the code generator, yet.
1658    
1659      4. In CProto,
1660    
1661         An encoding type of "bool" means "ml object" and is mapped into
1662         C prototype of PTR.  Note that "bool" is different than "string",
1663         even though "string" is also mapped into PTR, because "bool"
1664         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1665    
1666      5. Pickler/unpicker
1667    
1668         Changed to handle RAW_RECORD and newest RAW_CCALL
1669    
1670      6. MLRiscGen,
1671    
1672         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1673         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1674            in the file CodeGen/cpscompile/cps-c-calls.sml
1675    
1676      7. Added the conditional move operator
1677    
1678             condmove of branch
1679    
1680         to cps.  Generation of this is still buggy so it is currently
1681         disabled.
1682    
1683    ----------------------------------------------------------------------
1684    Name: Lal George
1685    Date: 2002/03/22 14:18:25 EST
1686    Tag: george-20020322-cps-branch-prob
1687    Description:
1688    
1689    Implemented the Ball-Larus branch prediction-heuristics, and
1690    incorporated graphical viewers for control flow graphs.
1691    
1692    Ball-Larus Heuristics:
1693    ---------------------
1694    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1695    
1696    By design it uses the Dempster-Shafer theory for combining
1697    probabilities.  For example, in the function:
1698    
1699        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1700    
1701    the ball-larus heuristics predicts that the n=0 is unlikely
1702    (OH-heuristic), and the 'then' branch is unlikely because of the
1703    RH-heuristic -- giving the 'then' branch an even lower combined
1704    probability using the Dempster-Shafer theory.
1705    
1706    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1707    probability of the 'then' branch because of the loop in the else
1708    branch.
1709    
1710    
1711    Graphical Viewing:
1712    ------------------
1713    I merely plugged in Allen's graphical viewers into the compiler. The
1714    additional code is not much. At the top level, saying:
1715    
1716            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1717    
1718    will display the graphical view of the control flow graph just before
1719    back-patching.  daVinci must be in your path for this to work. If
1720    daVinci is not available, then the default viewer can be changed
1721    using:
1722    
1723            Control.MLRISC.getString "viewer"
1724    
1725    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1726    course, these viewers must be in your path.
1727    
1728    The above will display the compilation unit at the level of clusters,
1729    many of which are small, boring, and un-interesting. Also setting:
1730    
1731            Control.MLRISC.getInt "cfg-graphical-view_size"
1732    
1733    will display clusters that are larger than the value set by the above.
1734    
1735    
1736    ----------------------------------------------------------------------
1737    Name: Matthias Blume
1738    Date: 2002/03/21 22:20:00 EST
1739    Tag: blume-20020321-kmp-bugfix
1740    Description:
1741    
1742    Changed the interface to the KMP routine in PreString and fixed
1743    a minor bug in one place where it was used.
1744    
1745    ----------------------------------------------------------------------
1746    Name: Allen Leung
1747    Date: 2002/03/21 20:30:00 EST
1748    Tag: leunga-20020321-cfg
1749    Description:
1750    
1751      Fixed a potential problem in cfg edge splitting.
1752    
1753    ----------------------------------------------------------------------
1754    Name: Allen Leung
1755    Date: 2002/03/21 17:15:00 EST
1756    Tag: leunga-20020321-x86-fp-cfg
1757    Description:
1758    
1759      1. Recoded the buggy parts of x86-fp.
1760    
1761         a. All the block reordering code has been removed.
1762            We now depend on the block placement phases to do this work.
1763    
1764         b. Critical edge splitting code has been simplified and moved into the
1765            CFG modules, as where they belong.
1766    
1767         Both of these were quite buggy and complex.  The code is now much, much
1768         simpler.
1769    
1770      2. X86 backend.
1771    
1772         a. Added instructions for 64-bit support.  Instruction selection for
1773            64-bit has not been committed, however, since that
1774            requires changes to MLTREE which haven't been approved by
1775            Lal and John.
1776    
1777         b. Added support for FUCOMI and FUCOMIP when generating code for
1778            PentiumPro and above.  We only generate these instructions in
1779            the fast-fp mode.
1780    
1781         c. Added cases for JP and JNP in X86FreqProps.
1782    
1783      3. CFG
1784    
1785         CFG now has a bunch of methods for edge splitting and merging.
1786    
1787      4. Machine description.
1788    
1789         John's simplification of MLTREE_BASIS.fcond broke a few machine
1790         description things:
1791    
1792         rtl-build.{sig,sml} and hppa.mdl fixed.
1793    
1794         NOTE: the machine description stuff in the repository is still broken.
1795               Again, I can't put my fixes in because that involves
1796               changes to MLTREE.
1797    
1798    ----------------------------------------------------------------------
1799    Name: Matthias Blume
1800    Date: 2002/03/20 15:55:00 EST
1801    Tag: blume-20020320-kmp
1802    Description:
1803    
1804    Implemented Knuth-Morris-Pratt string matching in PreString and used
1805    it for String.isSubstring, Substring.isSubstring, and
1806    Substring.position.
1807    
1808    (Might need some stress-testing.  Simple examples worked fine.)
1809    
1810    ----------------------------------------------------------------------
1811    Name: Matthias Blume
1812    Date: 2002/03/19 16:37:00 EST
1813    Tag: blume-20020319-witnesses
1814    Description:
1815    
1816    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1817    
1818    This implements a generic mechanism for changing constness qualifiers
1819    anywhere within big C types without resorting to outright "casts".
1820    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1821    modify the constness at the outermost level.)
1822    The implementation of "convert" is based on the idea of "witness"
1823    values -- values that are not used by the operation but whose types
1824    "testify" to their applicability.  On the implementation side, "convert"
1825    is simply a projection (returning its second curried argument).  With
1826    cross-module inlining, it should not result in any machine code being
1827    generated.
1828    
1829    ----------------------------------------------------------------------
1830    Name: Matthias Blume
1831    Date: 2002/03/15 16:40:00 EST
1832    Tag: blume-20020315-basis
1833    Description:
1834    
1835    Provided (preliminary?) implementations for
1836    
1837      {String,Substring}.{concatWith,isSuffix,isSubstring}
1838    
1839    and
1840    
1841      Substring.full
1842    
1843    Those are in the Basis spec but they were missing in SML/NJ.
1844    
1845    ----------------------------------------------------------------------
1846    Name: Matthias Blume
1847    Date: 2002/03/14 21:30:00 EST
1848    Tag: blume-20020314-controls
1849    Description:
1850    
1851    Controls:
1852    ---------
1853    
1854    1. Factored out the recently-added Controls : CONTROLS stuff and put
1855       it into its own library $/controls-lib.cm.  The source tree for
1856       this is under src/smlnj-lib/Controls.
1857    
1858    2. Changed the names of types and functions in this interface, so they
1859       make a bit more "sense":
1860    
1861          module -> registry
1862          'a registry -> 'a group
1863    
1864    3. The interface now deals in ref cells only.  The getter/setter interface
1865       is (mostly) gone.
1866    
1867    4. Added a function that lets one register an already-existing ref cell.
1868    
1869    5. Made the corresponding modifications to the rest of the code so that
1870       everything compiles again.
1871    
1872    6. Changed the implementation of Controls.MLRISC back to something closer
1873       to the original.  In particular, this module (and therefore MLRISC)
1874       does not depend on Controls.  There now is some link-time code in
1875       int-sys.sml that registers the MLRISC controls with the Controls
1876       module.
1877    
1878    CM:
1879    ---
1880    
1881      * One can now specify the lambda-split aggressiveness in init.cmi.
1882    
1883    ----------------------------------------------------------------------
1884    Name: Allen Leung
1885    Date: 2002/03/13 17:30:00 EST
1886    Tag: leunga-20020313-x86-fp-unary
1887    Description:
1888    
1889    Bug fix for:
1890    
1891    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1892    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1893    > - fun f(x,(y,z)) = Real.~ y;
1894    > [autoloading]
1895    > [autoloading done]
1896    >       fchsl   (%eax), 184(%esp)
1897    > Error: MLRisc bug: X86MCEmitter.emitInstr
1898    >
1899    > uncaught exception Error
1900    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1901    
1902    The problem was that the code generator did not generate any fp registers
1903    in this case, and the ra didn't know that it needed to run the X86FP phase to
1904    translate the pseudo fp instruction.   This only happened with unary fp
1905    operators in certain situations.
1906    
1907    ----------------------------------------------------------------------
1908    Name: Matthias Blume
1909    Date: 2002/03/13 14:00:00 EST
1910    Tag: blume-20020313-overload-etc
1911    Description:
1912    
1913    1. Added _overload as a synonym for overload for backward compatibility.
1914       (Control.overloadKW must be true for either version to be accepted.)
1915    
1916    2. Fixed bug in install script that caused more things to be installed
1917       than what was requested in config/targets.
1918    
1919    3. Made CM aware of the (_)overload construct so that autoloading
1920       works.
1921    
1922    ----------------------------------------------------------------------
1923    Name: Matthias Blume
1924    Date: 2002/03/12 22:03:00 EST
1925    Tag: blume-20020312-url
1926    Description:
1927    
1928    Forgot to update BOOT and srcarchiveurl.
1929    
1930    ----------------------------------------------------------------------
1931    Name: Matthias Blume
1932    Date: 2002/03/12 17:30:00 EST
1933    Tag: blume-20020312-version110392
1934    Description:
1935    
1936    Yet another version number bump (because of small changes to the
1937    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1938    
1939    Changes:
1940    
1941      The new pid generation scheme described a few weeks ago was overly
1942      complicated.  I implemented a new mechanism that is simpler and
1943      provides a bit more "stability":  Once CM has seen a compilation
1944      unit, it keeps its identity constant (as long as you do not delete
1945      those crucial CM/GUID/* files).  This means that when you change
1946      an interface, compile, then go back to the old interface, and
1947      compile again, you arrive at the original pid.
1948    
1949      There now also is a mechanism that instructs CM to use the plain
1950      environment hash as a module's pid (effectively making its GUID
1951      the empty string).  For this, "noguid" must be specified as an
1952      option to the .sml file in question within its .cm file.
1953      This is most useful for code that is being generated by tools such
1954      as ml-nlffigen (because during development programmers tend to
1955      erase the tool's entire output directory tree including CM's cached
1956      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1957      revert to the old, broken behavior of SML/NJ, but in specific cases
1958      where there is no danger of interface confusion, its use is ok
1959      (I think).
1960    
1961      ml-nlffigen by default generates "noguid" annotations.  They can be
1962      turned off by specifying -guid in its command line.
1963    
1964    ----------------------------------------------------------------------
1965    Name: Lal George
1966    Date: 2002/03/12 12 14:42:36 EST
1967    Tag: george-20020312-frequency-computation
1968    Description:
1969    
1970    Integrated jump chaining and static block frequency into the
1971    compiler. More details and numbers later.
1972    
1973    ----------------------------------------------------------------------
1974    Name: Lal George
1975    Date: 2002/03/11 11 22:38:53 EST
1976    Tag: george-20020311-jump-chain-elim
1977    Description:
1978    
1979    Tested the jump chain elimination on all architectures (except the
1980    hppa).  This is on by default right now and is profitable for the
1981    alpha and x86, however, it may not be profitable for the sparc and ppc
1982    when compiling the compiler.
1983    
1984    The gc test will typically jump to a label at the end of the cluster,
1985    where there is another jump to an external cluster containing the actual
1986    code to invoke gc. This is to allow factoring of common gc invocation
1987    sequences. That is to say, we generate:
1988    
1989            f:
1990               testgc
1991               ja   L1      % jump if above to L1
1992    
1993            L1:
1994               jmp L2
1995    
1996    
1997    After jump chain elimination the 'ja L1' instructions is converted to
1998    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1999    up being implemented in their long form (if L2 is far away) using:
2000    
2001            jbe     L3      % jump if below or equal to L3
2002            jmp     L2
2003         L3:
2004            ...
2005    
2006    
2007    For large compilation units L2  may be far away.
2008    
2009    
2010    ----------------------------------------------------------------------
2011    Name: Matthias Blume
2012    Date: 2002/03/11 13:30:00 EST
2013    Tag: blume-20020311-mltreeeval
2014    Description:
2015    
2016    A functor parameter was missing.
2017    
2018    ----------------------------------------------------------------------
2019    Name: Allen Leung
2020    Date: 2002/03/11 10:30:00 EST
2021    Tag: leunga-20020311-runtime-string0
2022    Description:
2023    
2024       The representation of the empty string now points to a
2025    legal null terminated C string instead of unit.  It is now possible
2026    to convert an ML string into C string with InlineT.CharVector.getData.
2027    This compiles into one single machine instruction.
2028    
2029    ----------------------------------------------------------------------
2030    Name: Allen Leung
2031    Date: 2002/03/10 23:55:00 EST
2032    Tag: leunga-20020310-x86-call
2033    Description:
2034    
2035       Added machine generation for CALL instruction (relative displacement mode)
2036    
2037    ----------------------------------------------------------------------
2038    Name: Matthias Blume
2039    Date: 2002/03/08 16:05:00
2040    Tag: blume-20020308-entrypoints
2041    Description:
2042    
2043    Version number bumped to 110.39.1.  NEW BOOTFILES!
2044    
2045    Entrypoints: non-zero offset into a code object where execution should begin.
2046    
2047    - Added the notion of an entrypoint to CodeObj.
2048    - Added reading/writing of entrypoint info to Binfile.
2049    - Made runtime system bootloader aware of entrypoints.
2050    - Use the address of the label of the first function given to mlriscGen
2051      as the entrypoint.  This address is currently always 0, but it will
2052      not be 0 once we turn on block placement.
2053    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
2054      with entry points) from mlriscGen.
2055    
2056    ----------------------------------------------------------------------
2057    Name: Allen Leung
2058    Date: 2002/03/07 20:45:00 EST
2059    Tag: leunga-20020307-x86-cmov
2060    Description:
2061    
2062       Bug fixes for CMOVcc on x86.
2063    
2064       1. Added machine code generation for CMOVcc
2065       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
2066       3. CMOVcc cannot have an immediate operand as argument.
2067    
2068    ----------------------------------------------------------------------
2069    Name: Matthias Blume
2070    Date: 2002/03/07 16:15:00 EST
2071    Tag: blume-20020307-controls
2072    Description:
2073    
2074    This is a very large but mostly boring patch which makes (almost)
2075    every tuneable compiler knob (i.e., pretty much everything under
2076    Control.* plus a few other things) configurable via both the command
2077    line and environment variables in the style CM did its configuration
2078    until now.
2079    
2080    Try starting sml with '-h' (or, if you are brave, '-H')
2081    
2082    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
2083    implements the underlying generic mechanism.
2084    
2085    The interface to some of the existing such facilities has changed somewhat.
2086    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
2087    (The getFoo interface is still there for backward-compatibility, but its
2088    use is deprecated.)
2089    
2090    The ml-build script passes -Cxxx=yyy command-line arguments through so
2091    that one can now twiddle the compiler settings when using this "batch"
2092    compiler.
2093    
2094    TODO items:
2095    
2096    We should go through and throw out all controls that are no longer
2097    connected to anything.  Moreover, we should go through and provide
2098    meaningful (and correct!) documentation strings for those controls
2099    that still are connected.
2100    
2101    Currently, multiple calls to Controls.new are accepted (only the first
2102    has any effect).  Eventually we should make sure that every control
2103    is being made (via Controls.new) exactly once.  Future access can then
2104    be done using Controls.acc.
2105    
2106    Finally, it would probably be a good idea to use the getter-setter
2107    interface to controls rather than ref cells.  For the time being, both
2108    styles are provided by the Controls module, but getter-setter pairs are
2109    better if thread-safety is of any concern because they can be wrapped.
2110    
2111    *****************************************
2112    
2113    One bug fix: The function blockPlacement in three of the MLRISC
2114    backpatch files used to be hard-wired to one of two possibilities at
2115    link time (according to the value of the placementFlag).  But (I
2116    think) it should rather sense the flag every time.
2117    
2118    *****************************************
2119    
2120    Other assorted changes (by other people who did not supply a HISTORY entry):
2121    
2122    1. the cross-module inliner now works much better (Monnier)
2123    2. representation of weights, frequencies, and probabilities in MLRISC
2124       changed in preparation of using those for weighted block placement
2125       (Reppy, George)
2126    
2127    ----------------------------------------------------------------------
2128    Name: Lal George
2129    Date: 2002/03/07 14:44:24 EST 2002
2130    Tag: george-20020307-weighted-block-placement
2131    
2132    Tested the weighted block placement optimization on all architectures
2133    (except the hppa) using AMPL to generate the block and edge frequencies.
2134    Changes were required in the machine properties to correctly
2135    categorize trap instructions. There is an MLRISC flag
2136    "weighted-block-placement" that can be used to enable weighted block
2137    placement, but this will be ineffective without block/edge
2138    frequencies (coming soon).
2139    
2140    
2141    ----------------------------------------------------------------------
2142    Name: Lal George
2143    Date: 2002/03/05 17:24:48 EST
2144    Tag: george-20020305-linkage-cluster
2145    
2146    In order to support the block placement optimization, a new cluster
2147    is generated as the very first cluster (called the linkage cluster).
2148    It contains a single jump to the 'real' entry point for the compilation
2149    unit. Block placement has no effect on the linkage cluster itself, but
2150    all the other clusters  have full freedom in the manner in which they
2151    reorder blocks or functions.
2152    
2153    On the x86 the typical linkage code that is generated is:
2154       ----------------------
2155            .align 2
2156       L0:
2157            addl    $L1-L0, 72(%esp)
2158            jmp     L1
2159    
2160    
2161            .align  2
2162       L1:
2163       ----------------------
2164    
2165    72(%esp) is the memory location for the stdlink register. This
2166    must contain the address of the CPS function being called. In the
2167    above example, it contains the address of  L0; before
2168    calling L1 (the real entry point for the compilation unit), it
2169    must contain the address for L1, and hence
2170    
2171            addl $L1-L0, 72(%esp)
2172    
2173    I have tested this on all architectures except the hppa.The increase
2174    in code size is of course negligible
2175    
2176    ----------------------------------------------------------------------
2177    Name: Allen Leung
2178    Date: 2002/03/03 13:20:00 EST
2179    Tag: leunga-20020303-mlrisc-tools
2180    
2181      Added #[ ... ] expressions to mlrisc tools
2182    
2183    ----------------------------------------------------------------------
2184    Name: Matthias Blume
2185    Date: 2002/02/27 12:29:00 EST
2186    Tag: blume-20020227-cdebug
2187    Description:
2188    
2189    - made types in structure C and C_Debug to be equal
2190    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
2191    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
2192    
2193    ----------------------------------------------------------------------
2194    Name: Matthias Blume
2195    Date: 2002/02/26 12:00:00 EST
2196    Tag: blume-20020226-ffi
2197    Description:
2198    
2199    1. Fixed a minor bug in CM's "noweb" tool:
2200       If numbering is turned off, then truly don't number (i.e., do not
2201       supply the -L option to noweb).  The previous behavior was to supply
2202       -L'' -- which caused noweb to use the "default" line numbering scheme.
2203       Thanks to Chris Richards for pointing this out (and supplying the fix).
2204    
2205    2. Once again, I reworked some aspects of the FFI:
2206    
2207       A. The incomplete/complete type business:
2208    
2209       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
2210         gone!
2211       - ML types representing an incomplete type are now *equal* to
2212         ML types representing their corresponding complete types (just like
2213         in C).  This is still safe because ml-nlffigen will not generate
2214         RTTI for incomplete types, nor will it generate functions that
2215         require access to such RTTI.   But when ML code generated from both
2216         incomplete and complete versions of the C type meet, the ML types
2217         are trivially interoperable.
2218    
2219         NOTE:  These changes restore the full generality of the translation
2220         (which was previously lost when I eliminated functorization)!
2221    
2222       B. Enum types:
2223    
2224       - Structure C now has a type constructor "enum" that is similar to
2225         how the "su" constructor works.  However, "enum" is not a phantom
2226         type because each "T enum" has values (and is isomorphic to
2227         MLRep.Signed.int).
2228       - There are generic access operations for enum objects (using
2229         MLRep.Signed.int).
2230       - ml-nlffigen will generate a structure E_foo for each "enum foo".
2231         * The structure contains the definition of type "mlrep" (the ML-side
2232         representation type of the enum).  Normally, mlrep is the same
2233         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
2234         then mlrep will be defined as a datatype -- thus facilitating
2235         pattern matching on mlrep values.
2236         ("-ec" will be suppressed if there are duplicate values in an
2237          enumeration.)
2238         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
2239         will be generated for each C enum constant xxx.
2240         * Conversion functions m2i and i2m convert between mlrep and
2241         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
2242         * Coversion functions c and ml convert between mlrep and "tag enum".
2243         * Access functions (get/set) fetch and store mlrep values.
2244       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
2245         enumerations are merged into one single enumeration represented by
2246         structure E_'.
2247    
2248    ----------------------------------------------------------------------
2249    Name: Allen Leung
2250    Date: 2002/02/25 04:45:00 EST
2251    Tag: leunga-20020225-cps-spill
2252    
2253    This is a new implementation of the CPS spill phase.
2254    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
2255    In case of problems, replace it with the old file spill.sml
2256    
2257    The current compiler runs into some serious performance problems when
2258    constructing a large record.  This can happen when we try to compile a
2259    structure with many items.  Even a very simple structure like the following
2260    makes the compiler slow down.
2261    
2262        structure Foo = struct
2263           val x_1 = 0w1 : Word32.int
2264           val x_2 = 0w2 : Word32.int
2265           val x_3 = 0w3 : Word32.int
2266           ...
2267           val x_N = 0wN : Word32.int
2268        end
2269    
2270    The following table shows the compile time, from N=1000 to N=4000,
2271    with the old compiler:
2272    
2273    N
2274    1000   CPS 100 spill                           0.04u  0.00s  0.00g
2275           MLRISC ra                               0.06u  0.00s  0.05g
2276              (spills = 0 reloads = 0)
2277           TOTAL                                   0.63u  0.07s  0.21g
2278    
2279    1100   CPS 100 spill                           8.25u  0.32s  0.64g
2280           MLRISC ra                               5.68u  0.59s  3.93g
2281              (spills = 0 reloads = 0)
2282           TOTAL                                   14.71u  0.99s  4.81g
2283    
2284    1500   CPS 100 spill                           58.55u  2.34s  1.74g
2285           MLRISC ra                               5.54u  0.65s  3.91g
2286              (spills = 543 reloads = 1082)
2287           TOTAL                                   65.40u  3.13s  6.00g
2288    
2289    2000   CPS 100 spill                           126.69u  4.84s  3.08g
2290           MLRISC ra                               0.80u  0.10s  0.55g
2291              (spills = 42 reloads = 84)
2292           TOTAL                                   129.42u  5.10s  4.13g
2293    
2294    3000   CPS 100 spill                           675.59u  19.03s  11.64g
2295           MLRISC ra                               2.69u  0.27s  1.38g
2296              (spills = 62 reloads = 124)
2297           TOTAL                                   682.48u  19.61s  13.99g
2298    
2299    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
2300           MLRISC ra                               4.96u  0.27s  2.72g
2301              (spills = 85 reloads = 170)
2302           TOTAL                                   2375.26u  57.21s  48.00g
2303    
2304    As you can see the old cps spill module suffers from some serious
2305    performance problem.  But since I cannot decipher the old code fully,
2306    instead of patching the problems up, I'm reimplementing it
2307    with a different algorithm.  The new code is more modular,
2308    smaller when compiled, and substantially faster
2309    (O(n log n) time and O(n) space).  Timing of the new spill module:
2310    
2311    4000  CPS 100 spill                           0.02u  0.00s  0.00g
2312          MLRISC ra                               0.25u  0.02s  0.15g
2313             (spills=1 reloads=3)
2314          TOTAL                                   7.74u  0.34s  1.62g
2315    
2316    Implementation details:
2317    
2318    As far as I can tell, the purpose of the CPS spill module is to make sure the
2319    number of live variables at any program point (the bandwidth)
2320    does not exceed a certain limit, which is determined by the
2321    size of the spill area.
2322    
2323    When the bandwidth is too large, we decrease the register pressure by
2324    packing live variables into spill records.  How we achieve this is
2325    completely different than what we did in the old code.
2326    
2327    First, there is something about the MLRiscGen code generator
2328    that we should be aware of:
2329    
2330    o MLRiscGen performs code motion!
2331    
2332       In particular, it will move floating point computations and
2333       address computations involving only the heap pointer to
2334       their use sites (if there is only a single use).
2335       What this means is that if we have a CPS record construction
2336       statement
2337    
2338           RECORD(k,vl,w,e)
2339    
2340       we should never count the new record address w as live if w
2341       has only one use (which is often the case).
2342    
2343       We should do something similar to floating point, but the transformation
2344       there is much more complex, so I won't deal with that.
2345    
2346    Secondly, there are now two new cps primops at our disposal:
2347    
2348     1. rawrecord of record_kind option
2349        This pure operator allocates some uninitialized storage from the heap.
2350        There are two forms:
2351    
2352         rawrecord NONE [INT n]  allocates a tagless record of length n
2353         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2354                                     and initializes the tag.
2355    
2356     2. rawupdate of cty
2357          rawupdate cty (v,i,x)
2358          Assigns to x to the ith component of record v.
2359          The storelist is not updated.
2360    
2361    We use these new primops for both spilling and increment record construction.
2362    
2363     1. Spilling.
2364    
2365        This is implemented with a linear scan algorithm (but generalized
2366        to trees).  The algorithm will create a single spill record at the
2367        beginning of the cps function and use rawupdate to spill to it,
2368        and SELECT or SELp to reload from it.  So both spills and reloads
2369        are fine-grain operations.  In contrast, in the old algorithm
2370        "spills" have to be bundled together in records.
2371    
2372        Ideally, we should sink the spill record construction to where
2373        it is needed.  We can even split the spill record into multiple ones
2374        at the places where they are needed.  But CPS is not a good
2375        representation for global code motion, so I'll keep it simple and
2376        am not attempting this.
2377    
2378     2. Incremental record construction (aka record splitting).
2379    
2380        Long records with many component values which are simulatenously live
2381        (recall that single use record addresses are not considered to
2382         be live) are constructed with rawrecord and rawupdate.
2383        We allocate space on the heap with rawrecord first, then gradually
2384        fill it in with rawupdate.  This is the technique suggested to me
2385        by Matthias.
2386    
2387        Some restrictions on when this is applicable:
2388        1. It is not a VECTOR record.  The code generator currently does not handle
2389           this case. VECTOR record uses double indirection like arrays.
2390        2. All the record component values are defined in the same "basic block"
2391           as the record constructor.  This is to prevent speculative
2392           record construction.
2393    
2394    ----------------------------------------------------------------------
2395    Name: Allen Leung
2396    Date: 2002/02/22 01:02:00 EST
2397    Tag: leunga-20020222-mlrisc-tools
2398    
2399    Minor bug fixes in the parser and rewriter
2400    
2401    ----------------------------------------------------------------------
2402    Name: Allen Leung
2403    Date: 2002/02/21 20:20:00 EST
2404    Tag: leunga-20020221-peephole
2405    
2406    Regenerated the peephole files.  Some contained typos in the specification
2407    and some didn't compile because of pretty printing bugs in the old version
2408    of 'nowhere'.
2409    
2410    ----------------------------------------------------------------------
2411    Name: Allen Leung
2412    Date: 2002/02/19 20:20:00 EST
2413    Tag: leunga-20020219-mlrisc-tools
2414    Description:
2415    
2416       Minor bug fixes to the mlrisc-tools library:
2417    
2418       1.  Fixed up parsing colon suffixed keywords
2419       2.  Added the ability to shut the error messages up
2420       3.  Reimplemented the pretty printer and fixed up/improved
2421           the pretty printing of handle and -> types.
2422       4.  Fixed up generation of literal symbols in the nowhere tool.
2423       5.  Added some SML keywords to to sml.sty
2424    
2425    ----------------------------------------------------------------------
2426    Name: Matthias Blume
2427    Date: 2002/02/19 16:20:00 EST
2428    Tag: blume-20020219-cmffi
2429    Description:
2430    
2431    A wild mix of changes, some minor, some major:
2432    
2433    * All C FFI-related libraries are now anchored under $c:
2434        $/c.cm      --> $c/c.cm
2435        $/c-int.cm  --> $c/internals/c-int.cm
2436        $/memory.cm --> $c/memory/memory.cm
2437    
2438    * "make" tool (in CM) now treats its argument pathname slightly
2439      differently:
2440        1. If the native expansion is an absolute name, then before invoking
2441           the "make" command on it, CM will apply OS.Path.mkRelative
2442           (with relativeTo = OS.FileSys.getDir()) to it.
2443        2. The argument will be passed through to subsequent phases of CM
2444           processing without "going native".  In particular, if the argument
2445           was an anchored path, then "make" will not lose track of that anchor.
2446    
2447    * Compiler backends now "know" their respective C calling conventions
2448      instead of having to be told about it by ml-nlffigen.  This relieves
2449      ml-nlffigen from one of its burdens.
2450    
2451    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
2452    
2453    * Export C_DEBUG and C_Debug from $c/c.cm.
2454    
2455    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
2456      subtyping relationship between incomplete pointers and their complete
2457      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
2458      with the convention of instantiating 'o with ('t, 'c) obj whenever
2459      the pointer target type is complete.  In the incomplete case, 'o
2460      will be instantiated with some "'c iobj" -- a type obtained by
2461      using one of the functors PointerToIncompleteType or PointerToCompleteType.
2462    
2463      Operations that work on both incomplete and complete pointer types are
2464      typed as taking an 'o ptr while operations that require the target to
2465      be known are typed as taking some ('t, 'c) obj ptr.
2466    
2467      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
2468      where void is an eqtype without any values.  This makes it possible
2469      to work on voidptr values using functions meant to operate on light
2470      incomplete pointers.
2471    
2472    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
2473      been vastly simplified.
2474    
2475    ----------------------------------------------------------------------
2476    Name: Matthias Blume
2477    Date: 2002/02/19 10:48:00 EST
2478    Tag: blume-20020219-pqfix
2479    Description:
2480    
2481    Applied Chris Okasaki's bug fix for priority queues.
2482    
2483    ----------------------------------------------------------------------
2484    Name: Matthias Blume
2485    Date: 2002/02/15 17:05:00
2486    Tag: Release_110_39
2487    Description:
2488    
2489    Last-minute retagging is becoming a tradition... :-(
2490    
2491    This is the working release 110.39.
2492    
2493    ----------------------------------------------------------------------
2494    Name: Matthias Blume
2495    Date: 2002/02/15 16:00:00 EST
2496    Tag: Release_110_39-orig
2497    Description:
2498    
2499    Working release 110.39.  New bootfiles.
2500    
2501    (Update: There was a small bug in the installer so it wouldn't work
2502    with all shells.  So I retagged. -Matthias)
2503    
2504    ----------------------------------------------------------------------
2505    Name: Matthias Blume
2506    Date: 2002/02/15 14:17:00 EST
2507    Tag: blume-20020215-showbindings
2508    Description:
2509    
2510    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
2511    the latter can be useful for exploring what bindings are available at
2512    the interactive prompt.  (The first function returns only the list
2513    of symbols that are really bound, the second prints those but also the
2514    ones that CM's autoloading mechanism knows about.)
2515    
2516    ----------------------------------------------------------------------
2517    Name: Matthias Blume
2518    Date: 2002/02/15 12:08:00 EST
2519    Tag: blume-20020215-iptrs
2520    Description:
2521    
2522    Two improvements to ml-nlffigen:
2523    
2524      1. Write files only if they do not exist or if their current contents
2525         do not coincide with what's being written.  (That is, avoid messing
2526         with the time stamps unless absolutely necessary.)
2527    
2528      2. Implement a "repository" mechanism for generated files related
2529         to "incomplete pointer types".   See the README file for details.
2530    
2531    ----------------------------------------------------------------------
2532    Name: Matthias Blume
2533    Date: 2002/02/14 11:50:00 EST
2534    Tag: blume-20020214-quote
2535    Description:
2536    
2537    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
2538    because of the new and improved tag generation scheme.  (Thanks to Allen
2539    Leung for pointing it out.)
2540    
2541    ----------------------------------------------------------------------
2542    Name: Lal George
2543    Date: 2002/02/14 09:55:27 EST 2002
2544    Tag: george-20020214-isabelle-bug
2545    Description:
2546    
2547    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2548    of Isabelle on the x86.
2549    
2550    From Allen:
2551    -----------
2552     I've found the problem:
2553    
2554         in ra-core.sml, I use the counter "blocked" to keep track of the
2555         true number of elements in the freeze queue.  When the counter goes
2556         to zero, I skip examining the queue.  But I've messed up the
2557         bookkeeping in combine():
2558    
2559             else ();
2560             case !ucol of
2561               PSEUDO => (if !cntv > 0 then
2562                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2563                                        ^^^^^^^^^^^^^^^^^^^^^^^
2564                      moveu := mergeMoveList(!movev, !moveu)
2565                     )
2566                  else ();
2567    
2568         combine() is called to coalesce two nodes u and v.
2569         I think I was thinking that if the move counts of u and v are both
2570         greater than zero then after they are coalesced then one node is
2571         removed from the freeze queue.  Apparently I was thinking that
2572         both u and v are of low degree, but that's clearly not necessarily true.
2573    
2574    
2575    02/12/2002:
2576        Here's the patch.  HOL now compiles.
2577    
2578        I don't know how this impact on performance (compile
2579        time or runtime).  This bug caused the RA (especially on the x86)
2580        to go thru the potential spill phase when there are still nodes on the
2581        freeze queue.
2582    
2583    
2584    
2585    
2586    ----------------------------------------------------------------------
2587    Name: Matthias Blume
2588    Date: 2002/02/13 22:40:00 EST
2589    Tag: blume-20020213-fptr-rtti
2590    Description:
2591    
2592    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2593    updates.
2594    
2595    ----------------------------------------------------------------------
2596    Name: Matthias Blume
2597    Date: 2002/02/13 16:41:00 EST
2598    Tag: blume-20020213-cmlpq
2599    Description:
2600    
2601    Added new priority queue export symbols (which have just been added to
2602    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2603    not compile and the installer would choke.)
2604    
2605    ----------------------------------------------------------------------
2606    Name: Matthias Blume
2607    Date: 2002/02/13 16:15:00 EST
2608    Tag: blume-20020213-various
2609    Description:
2610    
2611    1. More tweaks to ml-nlffigen:
2612    
2613       - better internal datastructures (resulting in slight speedup)
2614       - "-match" option requires exact match
2615       - "localized" gensym counters (untagged structs/unions nested within
2616         other structs/unions or within typedefs get a fresh counter; their
2617         tag will be prefixed by a concatenation of their parents' tags)
2618       - bug fixes (related to calculation of transitive closure of types
2619         to be included in the output)
2620    
2621    2. Minor Basis updates:
2622    
2623       - added implementations for List.collate and Option.app
2624    
2625    ----------------------------------------------------------------------
2626    Name: Matthias Blume
2627    Date: 2002/02/11 15:55:00 EST
2628    Tag: blume-20020211-gensym
2629    Description:
2630    
2631    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2632    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2633    names (ML structure names that correspond to unnamed C structs, unions,
2634    and enums), so that separate runs of ml-nlffigen do not clash.
2635    
2636    ----------------------------------------------------------------------
2637    Name: Matthias Blume
2638    Date: 2002/02/11 12:05:00 EST
2639    Tag: blume-20020211-gensml
2640    Description:
2641    
2642    A quick fix for a problem with GenSML (in the pgraph-util library):
2643    Make generation of toplevel "local" optional.  (Strictly speaking,
2644    signature definitions within "local" are not legal SML.)
2645    
2646    Other than that: updates to INSTALL and cm/TODO.
2647    
2648    ----------------------------------------------------------------------
2649    Name: Matthias Blume
2650    Date: 2002/02/08 15:00:00 EST
2651    Tag: blume-20020208-uniquepid
2652    Description:
2653    
2654    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2655    
2656    1. The installer (config/install.sh) has gotten smarter:
2657    
2658         - Configuration options are a bit easier to specify now
2659           (in config/targets).
2660         - Bug in recognizing .tar.bz2 files fixed.
2661         - Installer automatically resolves dependencies between
2662           configuration options (e.g., if you ask for eXene, you will
2663           also get cml -- regardless whether you asked for it or not).
2664         - Installer can run in "quieter mode" by setting the environment
2665           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2666           "completely silent", though.
2667         - Build HashCons library as part of smlnj-lib.
2668    
2669    2. A new scheme for assigning persistent identifiers to compilation
2670       units (and, by extension, to types etc.) has been put into place.
2671       This fixes a long-standing bug where types and even dynamic values
2672       can get internally confused, thereby compromising type safety
2673       (abstraction) and dynamic correctness.  See
2674    
2675         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2676    
2677       for an example of how things could go wrong until now.
2678    
2679       The downside of the new scheme is that pids are not quite as
2680       persistent as they used to be: CM will generate a fresh pid
2681       for every compilation unit that it thinks it sees for the first
2682       time.  That means that if you compile starting from a clean, fresh
2683       source tree at two different times, you end up with different
2684       binaries.
2685    
2686       Cutoff recompilation, however, has not been compromised because
2687       CM keeps pid information in special caches between runs.
2688    
2689    ----------------------------------------------------------------------
2690    Name: Lal George
2691    Date: 2002/02/07 15:34:13 EST 2002
2692    Tag: <none>
2693    Description:
2694    
2695    Compilers that generate assembly code may produce  global labels
2696    whose value is resolved at link time. The various peephole optimization
2697    modules did not take this in account.
2698    
2699    TODO. The Labels.addrOf function should really return an option
2700    type so that clients are forced to deal with this issue, rather
2701    than an exception being raised.
2702    
2703    ----------------------------------------------------------------------
2704    Name: Lal George
2705    Date: 2002/02/06 13:55:02 EST
2706    Tag: george-20020206-ra-breakup
2707    Description:
2708    
2709    1. A bug fix from Allen.
2710    
2711        A typo causes extra fstp %st(0)'s to be generated at compensation
2712        edges, which might cause stack underflow traps at runtime.  This
2713        occurs in fft where there are extraneous fstps right before the 'into'
2714        trap instruction (in this case they are harmless since none of the
2715        integers overflow.)
2716    
2717    2. Pulled out various utility modules that were embedded in the modules
2718       of the register allocator. I need these modules for other purposes, but
2719       they are not complete enough to put into a library (just yet).
2720    ----------------------------------------------------------------------
2721    Name: Matthias Blume
2722    Date: 2002/01/31 16:05:00 EST
2723    Tag: blume-20020131-sparc-ccalls
2724    Description:
2725    
2726    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2727       of extra stack space by mistake.  Fixed.
2728    
2729    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2730    
2731    ----------------------------------------------------------------------
2732    Name: Allen Leung
2733    Date: 2002/01/30
2734    Tag: leunga-20020130-nowhere-bug-fix
2735    Description:
2736    
2737       MLRISC bug fixes:
2738       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2739       2. MachineInt.fromString was negating its value.
2740    
2741    ----------------------------------------------------------------------
2742    Name: Matthias Blume
2743    Date: 2002/01/29
2744    Tag: blume-20020129-INSTALL
2745    Description:
2746    
2747    - Added somewhat detailed installation instructions (file INSTALL).
2748    - Fixed curl-detection bug in config/install.sh.
2749    - It is now possible to select the URL getter using the URLGETTER
2750      environment variable:
2751    
2752          not set / "unknown"      --> automatic detection (script tries wget,
2753                                       curl, and lynx)
2754          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2755                                       how to properly invoke them)
2756          other                    --> use $URLGETTER directly, it must take
2757                                       precisely two command-line arguments
2758                                       (source URL and destination file name)
2759    
2760    ----------------------------------------------------------------------
2761    Name: Matthias Blume
2762    Date: 2002/01/28
2763    Tag: blume-20020128-sparc-ccalls
2764    Description:
2765    
2766    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2767    - Make use of the allocParam argument in sparc-c-calls.
2768    
2769    ----------------------------------------------------------------------
2770    Name: Matthias Blume
2771    Date: 2002/01/28
2772    Tag: blume-20020128-allocParam
2773    Description:
2774    
2775    John Reppy:  Changes c-calls API to accept client-callback for
2776    allocating extra stack space.
2777    me: Corresponding changes to mlriscGen (using a dummy argument that
2778        does not change the current behavior).
2779    
2780    ----------------------------------------------------------------------
2781    Name: Matthias Blume
2782    Date: 2002/01/28 12:00:00
2783    Tag: Release_110_38
2784    Description:
2785    
2786    This time for real!!!
2787    
2788    ----------------------------------------------------------------------
2789    Name: Matthias Blume
2790    Date: 2002/01/28 10:56:00 EST
2791    Tag: blume-20020128-retraction
2792    Description:
2793    
2794    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
2795       with blume-Release_110_38-retracted.)
2796    
2797    1. Fixed a problem with incorrect rounding modes in real64.sml.
2798       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2799    
2800    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2801       and enums fixed.  The naming of corresponding ML identifiers should
2802       now be consistent again.
2803    
2804    ----------------------------------------------------------------------
2805    Name: Allen Leung
2806    Date: 2002/01/27
2807    Tag: leunga-20020127-nowhere
2808    Description:
2809    
2810       Added a target called nowhere in the configuration scripts.
2811       Enabling this will build the MLRISC 'nowhere' tool (for translating
2812       programs with where-clauses into legal SML code) during installation.
2813    
2814    ----------------------------------------------------------------------
2815    Name: Matthias Blume
2816    Date: 2002/01/25 21:27:00 EST
2817    Tag: blume-Release_110_38-retracted
2818    Description:
2819    
2820    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
2821    
2822    README will be added later.
2823    
2824    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
2825    (Release_110_38) removed.  Reason: Last-minute bug fixes.
2826    
2827    ----------------------------------------------------------------------
2828    Name: Matthias Blume
2829    Date: 2002/01/25
2830    Tag: blume-20020125-ffi
2831    Description:
2832    
2833    A large number of tweaks and improvements to ml-nlffi-lib and
2834    ml-nlffigen:
2835    
2836       - ML represenation types have been streamlined
2837       - getter and setter functions work with concrete values, not abstract
2838         ones where possible
2839       - ml-nlffigen command line more flexible (see README file there)
2840       - some bugs have been fixed (hopefully)
2841    
2842    ----------------------------------------------------------------------
2843    Name: Lal George
2844    Date: 2002/01/24
2845    Tag: george-20020124-risc-ra-interface
2846    Description:
2847    
2848       There is a dramatic simplification in the interface to the
2849       register allocator for RISC architectures as a result of making
2850       parallel copy instructions explicit.
2851    
2852    ----------------------------------------------------------------------
2853    Name: Matthias Blume
2854    Date: 2002/01/22
2855    Tag: blume-20020122-x86-ccalls
2856    Description:
2857    
2858    Bug fix for c-calls on x86 (having to do with how char- and
2859    short-arguments are being handled).
2860    
2861    ----------------------------------------------------------------------
2862    Name: Matthias Blume
2863    Date: 2002/01/21
2864    Tag: blume-20020121-ff
2865    Description:
2866    
2867    Another day of fiddling with the FFI...
2868    
2869    1. Bug fix/workaround:  CKIT does not complain about negative array
2870       dimensions, so ml-nlffigen has to guard itself against this possibility.
2871       (Otherwise a negative dimension would send it into an infinite loop.)
2872    
2873    2. Some of the abstract types (light objects, light pointers, most "base"
2874       types) in structure C are now eqtypes.
2875    
2876    3. Added constructors and test functions for NULL function pointers.
2877    
2878    ----------------------------------------------------------------------
2879    Name: Matthias Blume
2880    Date: 2002/01/18
2881    Tag: blume-20020118-ready-for-new-release
2882    Description:
2883    
2884    Made config/srcarchiveurl point to a new place.  (Will provide boot
2885    files shortly.)
2886    
2887    Maybe we christen this to be 110.38?
2888    
2889    ----------------------------------------------------------------------
2890    Name: Matthias Blume
2891    Date: 2002/01/18
2892    Tag: blume-20020118-more-ffifiddle
2893    Description:
2894    
2895    Today's FFI fiddling:
2896    
2897      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
2898        C.Set.* that use concrete (MLRep.*) arguments and results instead
2899        of abstract ones.
2900    
2901      - Provided word-style bit operations etc. for "int" representation
2902        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
2903        and Long.
2904    
2905    ----------------------------------------------------------------------
2906    Name: Matthias Blume
2907    Date: 2002/01/18
2908    Tag: blume-20020118-use-x86-fp
2909    Description:
2910    
2911    Now that x86-fast-fp seems to be working, I turned it back on again
2912    by default.  (Seems to work fine now, even with the FFI.)
2913    
2914    Other than that, I added some documentation about the FFI to
2915    src/ml-nlffigen/README and updated the FFI test examples in
2916    src/ml-nlffi-lib/Tests/*.
2917    
2918    ----------------------------------------------------------------------
2919    Name: Allen Leung
2920    Date: 2002/01/17
2921    Tag: leunga-20020117-x86-fast-fp-call
2922    Description:
2923    
2924       1. Fixed a problem with handling return fp values when x86's fast fp
2925          mode is turned on.
2926    
2927       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
2928          of %f32.
2929    
2930       3. Added a constructor INT32lit to the ast of MLRISC tools.
2931    
2932    ----------------------------------------------------------------------
2933    Name: Matthias Blume
2934    Date: 2002/01/16
2935    Tag: blume-20020116-ffifiddle
2936    Description:
2937    
2938    More fiddling with the FFI interface:
2939    
2940     - Make constness 'c instead of rw wherever possible.  This eliminates
2941       the need for certain explicit coercions.  (However, due to ML's
2942       value polymorphism, there will still be many cases where explicit
2943       coercions are necessary.  Phantom types are not the whole answer
2944       to modeling a subtyping relationship in ML.)
2945    
2946     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
2947    
2948     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
2949       clumsy workaround for varargs problem.)
2950    
2951    ----------------------------------------------------------------------
2952    Name: Lal George
2953    Date: 2002/01/15
2954    Tag: <none>
2955    Description:
2956    
2957    1. Since COPY instructions are no longer native to the architecture,
2958       a generic functor can be used to implement the expandCopies function.
2959    
2960    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
2961       TEXT segment.
2962    
2963    ----------------------------------------------------------------------
2964    Name: Matthias Blume
2965    Date: 2002/01/15
2966    Tag: blume-20020115-ffiupdates
2967    Description:
2968    
2969    1. Fix for bug resulting in single-precision float values being returned
2970       incorrectly from FFI calls.
2971    
2972    2. Small modifications to C FFI API:
2973    
2974        - memory-allocation routines return straight objects (no options)
2975          and raise an exception in out-of-memory situations
2976        - unsafe extensions to cast between function pointers and pointers
2977          from/to ints
2978        - added structure C_Debug as an alternative to structure C where
2979          pointer-dereferencing (|*| and |*!) always check for null-pointers
2980        - added open_lib' to DynLinkage;  open_lib' works like open_lib
2981          but also takes a (possibly empty) list of existing library handles
2982          that the current library depends on
2983    
2984    ----------------------------------------------------------------------
2985    Name: Matthias Blume
2986    Date: 2002/01/10
2987    Tag: blume-20020110-newffigen
2988    Description:
2989    
2990    1. Updates to portable graph code.
2991    
2992    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
2993       more scalable now so that even huge interfaces such as the one
2994       for GTK compile in finite time and space. :-)
2995       See src/ml-nlffigen/README for details on what's new.
2996    
2997    ----------------------------------------------------------------------
2998    Name: Lal George
2999    Date: 2001/01/09 14:31:35 EST 2002
3000    Tag: george-20011206-rm-native-copy
3001    Description:
3002    
3003            Removed the native COPY and FCOPY instructions
3004            from all the architectures and replaced it with the
3005            explicit COPY instruction from the previous commit.
3006    
3007            It is now possible to simplify many of the optimizations
3008            modules that manipulate copies. This has not been
3009            done in this change.
3010    
3011    ----------------------------------------------------------------------
3012    Name: Lal George
3013    Date: 2001/12/06 16:50:13 EST 2001
3014    Tag: george-20011206-mlrisc-instruction
3015    Description:
3016    
3017    Changed the representation of instructions from being fully abstract
3018    to being partially concrete. That is to say:
3019    
3020      from
3021            type instruction
3022    
3023      to
3024            type instr                              (* machine instruction *)
3025    
3026            datatype instruction =
3027                LIVE of {regs: C.cellset, spilled: C.cellset}
3028              | KILL of {regs: C.cellset, spilled: C.cellset}
3029              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
3030              | ANNOTATION of {i: instruction, a: Annotations.annotation}
3031              | INSTR of instr
3032    
3033    This makes the handling of certain special instructions that appear on
3034    all architectures easier and uniform.
3035    
3036    LIVE and KILL say that a list of registers are live or killed at the
3037    program point where they appear. No spill code is generated when an
3038    element of the 'regs' field is spilled, but the register is moved to
3039    the 'spilled' (which is present, more for debugging than anything else).
3040    
3041    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
3042    We used to generate:
3043    
3044            DEFFREG f1
3045            f1 := f2 + f3
3046            trapb
3047    
3048    but now generate:
3049    
3050            f1 := f2 + f3
3051            trapb
3052            LIVE {regs=[f1,f2,f3], spilled=[]}
3053    
3054    Furthermore, the DEFFREG (hack) required that all floating point instruction
3055    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
3056    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
3057    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
3058    this architecture).
3059    
3060    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
3061    all the architectures. This will result in further simplification of the
3062    register allocator that must be aware of them for coalescing purposes, and
3063    will also simplify certain aspects of the machine description that provides
3064    callbacks related to parallel copies.
3065    
3066    ANNOTATION should be obvious, and now INSTR represents the honest to God
3067    machine instruction set!
3068    
3069    The <arch>/instructions/<arch>Instr.sml files define certain utility
3070    functions for making porting easier -- essentially converting upper case
3071    to lower case. All machine instructions (of type instr) are in upper case,
3072    and the lower case form generates an MLRISC instruction. For example on
3073    the alpha we have:
3074    
3075      datatype instr =
3076         LDA of {r:cell, b:cell, d:operand}
3077       | ...
3078    
3079      val lda : {r:cell, b:cell, d:operand} -> instruction
3080        ...
3081    
3082    where lda is just (INSTR o LDA), etc.
3083    
3084    ----------------------------------------------------------------------
3085    Name: Matthias Blume
3086    Date: 2001/11/22 21:40:00 EST
3087    Tag: Release_110_37
3088    Description:
3089    
3090    Release 110.37.  This time for real.
3091    
3092    ----------------------------------------------------------------------
3093    Name: Matthias Blume
3094    Date: 2001/11/21 16:35:00 EST
3095    Tag: blume-20011121-foot-in-mouth
3096    Description:
3097    
3098    Removed the "Release_110_37" tag because of a serious bug.
3099    This will be re-tagged once the bug is fixed.
3100    
3101    ----------------------------------------------------------------------
3102    Name: Matthias Blume
3103    Date: 2001/11/21 16:14:00 EST
3104    Tag: blume-20011121-forgottenfile
3105    Description:
3106    
3107    Forgot to add a file.  (Just a .tex-file -- part of
3108    the CM manual source.)
3109    
3110    ----------------------------------------------------------------------
3111    Name: Matthias Blume
3112    Date: 2001/11/21 16:10:00 EST
3113    Tag: blume-20011121-invalid_110_37
3114    Description:
3115    
3116    Note: I removed the original tag "Release_110_37" from this commit
3117          because we found a serious bug in all non-x86 backends.
3118          - Matthias
3119    
3120    1. Modifications to the SML/NJ code generator and to the runtime system
3121       so that code object name strings are directly inserted into code
3122       objects at code generation time.  The only business the runtime system
3123       has with this is now to read the name strings on occasions.
3124       (The encoding of the name string has also changed somewhat.)
3125    
3126    2. CM now implements a simple "set calculus" for specifying export lists.
3127       In particular, it is now possible to refer to the export lists of
3128       other libraries/groups/sources and form unions as well as differences.
3129       See the latest CM manual for details.
3130    
3131    3. An separate notion of "proxy" libraries has again be eliminated from
3132       CM's model.  (Proxy libraries are now simply a special case of using
3133       the export list calculus.)
3134    
3135    4. Some of the existing libraries now take advantage of the new set
3136       calculus.
3137       (Notice that not all libraries have been converted because some
3138       of the existing .cm-files are supposed to be backward compatible
3139       with 110.0.x.)
3140    
3141    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
3142       "General.exnMessage"!  The former relies on a certain hook to be
3143       initialized, and that often does not happen in the stand-alone case.)
3144    
3145    ----------------------------------------------------------------------
3146    Name: Lal George
3147    Date: 2001/11/21  13:56:18 EST
3148    Tag: george-2001121-pseudo-ops
3149    Description:
3150    
3151      Implemented a complete redesign of MLRISC pseudo-ops. Now there
3152      ought to never be any question of incompatabilities with
3153      pseudo-op syntax expected by host assemblers.
3154    
3155      For now, only modules supporting GAS syntax are implemented
3156      but more should follow, such as MASM, and vendor assembler
3157      syntax, e.g. IBM as, Sun as, etc.
3158    
3159    ----------------------------------------------------------------------
3160    Name: Matthias Blume
3161    Date: 2001/11/14 11:52:00 EST
3162    Tag: blume-20011114-srcname
3163    Description:
3164    
3165    1. Routed the name of the current source file to mlriscgen where it
3166       should be directly emitted into the code object.  (This last part
3167       is yet to be done.)
3168    
3169    2. Some cleanup of the pgraph code to make it match the proposal that
3170       I put out the other day.  (The proposal notwithstanding, things are
3171       still in flux here.)
3172    
3173    ----------------------------------------------------------------------
3174    Name: Lal George
3175    Date: 2001/11/14 09:44:04 EST
3176    Tag:
3177    Description:
3178    
3179      Fix for a backpatching bug reported by Allen.
3180    
3181      Because the boundary between short and long span-dependent
3182      instructions is +/- 128, there are an astounding number of
3183      span-dependent instructions whose size is over estimated.
3184    
3185      Allen came up with the idea of letting the size of span
3186      dependent instructions be non-monotonic, for a maxIter
3187      number of times, after which the size must be monotonically
3188      increasing.
3189    
3190      This table shows the number of span-dependent instructions
3191      whose size was over-estimated as a function of maxIter, for the
3192      file Parse/parse/ml.grm.sml:
3193    
3194         maxIter            # of instructions:
3195            10                      687
3196            20                      438
3197            30                      198
3198            40                        0
3199    
3200      In compiling the compiler, there is no significant difference in
3201      compilation speed between maxIter=10 and maxIter=40. Actually,
3202      my measurements showed that maxIter=40 was a tad faster than
3203      maxIter=10! Also 96% of the  files in the compiler reach a fix
3204      point within 13 iterations, so fixing maxIter at 40, while high,
3205      is okay.
3206    
3207    ----------------------------------------------------------------------
3208    Name: Matthias Blume
3209    Date: 2001/10/31 15:25:00 EST
3210    Tag: blume-20011031-pgraph
3211    Description:
3212    
3213    CKIT:
3214    * Changed the "Function" constructor of type Ast.ctype to carry optional
3215      argument identifiers.
3216    * Changed the return type of TypeUtil.getFunction accordingly.
3217    * Type equality ignores the argument names.
3218    * TypeUtil.composite tries to preserve argument names but gives up quickly
3219      if there is a mismatch.
3220    
3221    installation script:
3222    * attempts to use "curl" if available (unless "wget" is available as well)
3223    
3224    CM:
3225    * has an experimental implementation of "portable graphs" which I will
3226      soon propose as an implementation-independent library format
3227    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
3228    
3229    NLFFI-LIB:
3230    * some cleanup (all cosmetic)
3231    
3232    NLFFIGEN:
3233    * temporarily disabled the mechanism that suppresses ML output for
3234      C definitions whose identifiers start with an underscore character
3235    * generate val bindings for enum constants
3236    * user can request that only one style (light or heavy) is being used;
3237      default is to use both (command-line arguments: -heavy and -light)
3238    * fixed bug in handling of function types involving incomplete pointers
3239    * generate ML entry points that take record arguments (i.e., using
3240      named arguments) for C functions that have a prototype with named
3241      arguments
3242      (see changes to CKIT)
3243    
3244    ----------------------------------------------------------------------
3245    Name: Allen Leung
3246    Date: 2001/10/27 20:34:00 EDT
3247    Tag: leunga-20011027-x86-fast-fp-call
3248    Description:
3249    
3250       Fixed the bug described in blume-20010920-slowfp.
3251    
3252       The fix involves
3253          1. generating FCOPYs in FSTP in ia32-svid
3254          2. marking a CALL with the appropriate annotation
3255    
3256    ----------------------------------------------------------------------
3257    Name: Matthias Blume
3258    Date: 2001/10/16 11:32:00 EDT
3259    Tag: blume-20011016-netbsd
3260    Description:
3261    
3262    Underscore patch from Chris Richards (fixing problem with compiling
3263    runtime system under recent NetBSD).
3264    
3265    ----------------------------------------------------------------------
3266    Name: Allen Leung
3267    Date: 2001/10/12 17:18:32 EDT 2001
3268    Tag: leung-20011012-x86-printflowgraph
3269    Description:
3270    
3271    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
3272    
3273    ----------------------------------------------------------------------
3274    Name: Lal George
3275    Date: 2001/10/11 23:51:34 EDT
3276    Tag: george-20011011-too-many-instrs
3277    Description:
3278    
3279    The representation of a program point never expected to see more
3280    than 65536 instructions in a basic block!
3281    
3282    ----------------------------------------------------------------------
3283    Name: Lal George
3284    Date: 2001/10/09 09:41:37 EDT
3285    Tag: george-20011008-mlrisc-labels
3286    Description:
3287    
3288    Changed the machine description files to support printing of
3289    local and global labels in assembly code, based on host assembler
3290    conventions.
3291    
3292    ----------------------------------------------------------------------
3293  Name: Matthias Blume  Name: Matthias Blume
3294  Date: 2001/09/25 15:25:00 EDT  Date: 2001/09/25 15:25:00 EDT
3295  Tag: blume-20010925-exninfo  Tag: blume-20010925-exninfo

Legend:
Removed from v.939  
changed lines
  Added in v.1593

root@smlnj-gforge.cs.uchicago.edu
ViewVC Help
Powered by ViewVC 1.0.0