Home My Page Projects Code Snippets Project Openings SML/NJ
Summary Activity Forums Tracker Lists Tasks Docs Surveys News SCM Files

SCM Repository

[smlnj] Diff of /sml/trunk/HISTORY
ViewVC logotype

Diff of /sml/trunk/HISTORY

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

revision 874, Mon Jul 9 20:01:40 2001 UTC revision 1599, Mon Aug 9 16:22:40 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: 2004/08/09 12:21:00 EST
18    Tag: leunga-110_47-dijsktra
19    Description:
20    
21    Bug fix from Carl Hauser:
22    
23    single_source_shortest_paths in dijkstra.sml was observed to get wrong
24    answers (by comparing to single_source_shortest_paths in bellman-ford.sml).
25    
26    The problem is that following the expression A.update(dist,s,Num.zero)
27    it is necessary to update the priority queue using Q.decreaseWeight(Q,s).
28    
29    ----------------------------------------------------------------------
30    Name: Matthias Blume (blume (at) tti - c (dot) org)
31    Date: 2004/08/06 18:10:00 CDT
32    Tag: blume-20040806-cmdline
33    Description:
34    
35    Fiddled with handling of command-line options:
36    
37      * sml now quits after processing the command line
38        if -H, -S, -h<n>, or -s<n> appears as the last
39        command-line argument
40      * a new option -q terminates the session when encountered on
41        the command line; subsequent arguments will be ignored
42      * bug fixes: short (erroneous) arguments are no longer ignored
43        completely
44    
45    ----------------------------------------------------------------------
46    Name: Allen Leung (leunga (at) reservoir (dot) com)
47    Date: 2004/08/04 18:17:00 EST
48    Tag: leunga-110_47-ppc-ibm-asm
49    Description:
50    
51      - Added minimal IBM assembly syntax support for PowerPC.
52    
53      - Cygwin: manually changed the file cygwin.def.  Some exported symbols have
54        been altered in the runtime.  We need an automatic way to keep the file
55        in sync.
56    
57    ----------------------------------------------------------------------
58    Name: Matthias Blume (blume (at) tti - c (dot) org)
59    Date: 2004/08/04 14:00:00 CDT
60    Tag: Release_110_47
61    Description:
62    
63    New working version (110.47).  NEW BOOTFILES!
64    
65    ----------------------------------------------------------------------
66    Name: Matthias Blume (blume (at) tti - c (dot) org)
67    Date: 2004/08/03 14:25:00 CDT
68    Tag: blume-20040803-callingconv
69    Description:
70    
71    Added low-level support for choosing C calling conventions by
72    twiddling the type of rawccall.  (See
73    src/compiler/Semant/types/cproto.sml for details.)
74    
75    ----------------------------------------------------------------------
76    Name: Matthias Blume (blume (at) tti - c (dot) org)
77    Date: 2004/08/02 15:55:00 CDT
78    Tag: blume-20040802-backout
79    Description:
80    
81    Backed out of change to win32-filesys.c.  The earlier patch to
82    get_file_time caused CM to produce files with the wrong time stamp.
83    
84    ----------------------------------------------------------------------
85    Name: Matthias Blume (blume (at) tti - c (dot) org)
86    Date: 2004/08/02 14:45:00 CDT
87    Tag: blume-20040802-nlffi-win32
88    Description:
89    
90    Added NLFFI support for Win32, adapted from a patch provided by David
91    Hansel.  This is currently completely untested.  Also, the issue
92    concerning stdcall vs. ccall is still unresolved.
93    
94    ----------------------------------------------------------------------
95    Name: Matthias Blume (blume (at) tti - c (dot) org)
96    Date: 2004/07/30 17:55:00 CDT
97    Tag: blume-20040730-various
98    Description:
99    
100    Gearing up towards 110.47...
101    
102    - various minor bugfixes to ml-nlffigen
103    - a beginning of a manual for nlffi
104    
105    - eliminated 'export name=value' in config/install.sh as this does
106      not work with certain versions of /bin/sh
107      (Thanks to David King at Motorola for catching this.)
108    
109    - several bugfixes provided or suggested by David Hansel at Reactive Systems:
110      - added a test for tm==NULL to gmtime.c and localtime.c
111      - applied patch for incorrect GetFileTime under win32
112      - toSeconds -> toMilliseconds in Win32/win32-process.sml
113    
114    ----------------------------------------------------------------------
115    Name: Matthias Blume (blume (at) tti - c (dot) org)
116    Date: 2004/07/21 18:20:00 CDT
117    Tag: blume-20040721-nlffigen
118    Description:
119    
120    - Fixed minor issue in ml-nlffigen: Now generate structure T_foo
121      for a typedef to an incomplete type, but leave out the "typ" member.
122      (This is just for consistency.)
123    - Started to produce what is supposed to become better (i.e., comprehensive)
124      documentation of what ml-nlffigen does and produces.
125    
126    ----------------------------------------------------------------------
127    Name: Matthias Blume (blume (at) tti - c (dot) org)
128    Date: 2004/07/14 16:25:00 CDT
129    Tag: blume-20040714-union
130    Description:
131    
132    Added C_UNION to c-calls/c-types.sml and updated the machinery
133    (ml-nlffigen, cproto.sml) that conveys C function interface
134    information to the code generator.
135    
136    However, the actual architecture-specific implementation of function
137    arguments and results that are C unions is still not implemented.
138    
139    ----------------------------------------------------------------------
140    Name: Allen Leung (leunga (at) reservoir (dot) com)
141    Date: 2004/07/14 14:38:00 EST
142    Tag: leunga-110_46_1-ppc-lwzu
143    Description:
144    
145        Added these instructions to the PowerPC architecture:
146        LBZU(X), LHZU(X), LWZU(X),
147        STWU(X), STFDU, STFSU
148    
149        etc...
150    
151        Note: I haven't added their instruction encoding into the description.
152    
153    ----------------------------------------------------------------------
154    Name: Allen Leung (leunga (at) reservoir (dot) com)
155    Date: 2004/07/13 15:04:00 EST
156    Tag: leunga-110_46_1-ppc-lwarx
157    Description:
158    
159       Added the two instructions LWARX and STWCX to the PowerPC
160    instruction set.
161    
162       A (untested) rewrite of loop-structure.sml.  The old version
163    is completely broken.
164    
165    ----------------------------------------------------------------------
166    Name: Matthias Blume (blume (at) tti - c (dot) org)
167    Date: 2004/07/13 13:50:00 CDT
168    Tag: blume-20040713-nlffi
169    Description:
170    
171    - use paramAlloc to report c-calls with too many arguments
172      (for PPC version where parameter area is pre-allocated)
173    - added ccall_maxargspace to machspec (to implement the above)
174    - made "make" commend in CM's "make" tool configurable
175    - added option (default: on) for passing the name of the SML/NJ's "bin"
176      directory to "make"; the call looks like this:
177    
178         make <options> SMLNJ_BINDIR=<dir> <target>
179    
180      This can be used by the Makefile to, e.g., pick the "right" version
181      of ml-nlffigen.
182    - minor code tweaks
183    
184    ----------------------------------------------------------------------
185    Name: Matthias Blume (blume (at) tti - c (dot) org)
186    Date: 2004/07/12 22:50:00 CDT
187    Tag: blume-110_46_1-macosx-nlffi
188    Description:
189    
190    NLFFI under Mac OS X now working (sort of).  This is largely untested,
191    though.
192    
193    Note:  1. You have to make a new, clean build of the runtime system.
194           2. There are new BOOTFILES, you have to use them!
195              (Doing the bootstrap process yourself would be *very* painful!
196               If you absolutely have to do it, build the system under
197               a different architecture and then cross-compile.)
198    
199    Version bumped to 110.46.1 to account for runtime data format changes.
200    
201    ----------------------------------------------------------------------
202    Name: Matthias Blume (blume (at) tti - c (dot) org)
203    Date: 2004/06/18 14:30:00 CDT
204    Tag: blume-20040618-unix
205    Description:
206    
207    Changed the implementation of structure Unix so that the same stream
208    is returned every time one of the {text,bin}{In,Out}streamOf functions
209    is invoked on the same proc.  This is not what the spec currently
210    says -- although IMO it arguably should.  (See discussion below.)
211    
212    ----------------------------------------------------------------------
213    Name: Matthias Blume (blume (at) tti - c (dot) org)
214    Date: 2004/06/17 18:15:00 CDT
215    Tag: Release_110_46
216    Description:
217    
218    New working version (110.46).  NEW BOOTFILES!
219    
220    ----------------------------------------------------------------------
221    Name: Matthias Blume (blume (at) tti - c (dot) org)
222    Date: 2004/06/17 17:20:00 CDT
223    Tag: blume-20040617-timer-unix
224    Description:
225    
226    Changed the interface of structures Timer and Unix to match the most
227    recent Basis spec.
228    
229    In the case of Unix there still seems to be an open/weird issue:
230    
231       The {text,bin}{In,Out}streamOf functions are supposed to create
232       fresh streams whenever they are called -- as opposed to have them
233       return the same stream every time.  This design is supposed to
234       prevent space leaks caused by proc values hanging on to streams.
235    
236       The reap function, on the other hand, is supposed to close the
237       streams.  This cannot be done without having a handle on the
238       stream in proc after all...
239    
240    I took the liberty to implement the following stopgap solution:
241    
242      The proc value hangs on to the most recently created stream(s).
243      Reap closes those.  If either or both of the two streams hadn't
244      been created at all yet, then reap will close the corresponding
245      file descriptors directly.
246    
247    PS: I don't understand the original space leak argument anymore.  If
248    a proc hangs on to the imperative stream, then I/O operations on those
249    will advance the state of the cached stream and avoid the space leak.
250    
251    ----------------------------------------------------------------------
252    Name: Matthias Blume (blume (at) tti - c (dot) org)
253    Date: 2004/05/28 16:45:00 CDT
254    Tag: blume-20040528-basis
255    Description:
256    
257    Added signature PACK_REAL and exported functor PrimIO.
258    
259    ----------------------------------------------------------------------
260    Name: Matthias Blume (blume (at) tti - c (dot) org)
261    Date: 2004/05/25 16:00:00 CDT
262    Tag: blume-20040525-group-owner
263    Description:
264    
265    CM now ignores (but still accepts) the "owner" information in group
266    descriptions.  The owner of a group is its next enclosing
267    library. Each group must have a unique owner.  (There is a virtual
268    "toplevel" library that owns groups which are not nested within a real
269    library.)  Previously, each group had to explicitly declare its owner,
270    and CM would check that such a declaration is correct.  The new scheme
271    is to have CM check that for each group there is precisely one owning
272    library.
273    
274    The advantage of the new scheme is that the programmer no longer needs
275    to maintain the somewhat annoying owner information.  The downside is
276    that CM cannot enforce the ownership rule across multiple runs of
277    CM.make.  Fortunately, enclosing the same group in two different
278    libraries A and B which are not part of the same program does not
279    cause real problems.
280    
281    ----------------------------------------------------------------------
282    Name: Matthias Blume (blume (at) tti - c (dot) org)
283    Date: 2004/05/20 16:00:00 CDT
284    Tag: blume-20040520-win32
285    Description:
286    
287    Made the win32 version work again.  (Strangely, a misplaced comma had
288    slipped into win32-process.c which prevented the runtime from being
289    compiled correctly.)
290    
291    Also, included a minor addition to ml-build.bat analogous to what was
292    done in blume-20040519-ml-build.
293    
294    ----------------------------------------------------------------------
295    Name: Matthias Blume (blume (at) tti - c (dot) org)
296    Date: 2004/05/19 22:10:00 CDT
297    Tag: blume-20040519-ml-build
298    Description:
299    
300    Arranged for ml-build to clean up after itself a little bit better.
301    The script generates a temporary SML source file and compiles it using
302    CM, so CM generates metadata (GUID, SKEL, objectfile) for it.  It now
303    gets rid of those at the end, so they don't accumulate under .cm.
304    
305    This required a minor change to install.sh because the name of the
306    metadata directory (default: .cm) is actually configurable at
307    installation time.
308    
309    ----------------------------------------------------------------------
310    Name: Matthias Blume (blume (at) tti - c (dot) org)
311    Date: 2004/05/18 15:50:00 CDT
312    Tag: blume-20040518-mkreader
313    Description:
314    
315    Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
316    implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
317    
318    ----------------------------------------------------------------------
319    Name: Matthias Blume (blume (at) tti - c (dot) org)
320    Date: 2004/05/11 14:35:00 CDT
321    Tag: blume-20040511-win32sock
322    Description:
323    
324    Added previously missing support for many socket-related functions
325    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
326    for the voluminous patch!
327    
328    (I have not tested this patch under win32 yet.)
329    
330    Here is David's e-mail:
331    
332    Hi,
333    
334    Attached to this email you find a diff against sml/nj 110.45
335    that will enable socket support under Windows.
336    
337    To apply the patch (using unix or cygwin)
338    1) gunzip runtime.diff.gz
339    2) "cd" into "src/runtime" in the source tree of a fresh
340       110.45 installation.
341    3) patch -p 1 < [your/path/to]runtime.diff
342    
343    The code compiles fine but has NOT yet been extensively tested.
344    I only ran a few tests for basic socket client functionality
345    (which worked fine).  Especially the functions that use ioctl
346    are not tested at all and might not work (see below).
347    
348    I implemented this since we want to move to a newer version of sml/nj
349    but need socket support in order to use it.  This is the first time I
350    even had a look at the sml/nj source,  so please review my changes
351    before making this part of the distribution!  Here are a few issues
352    that I think might be better for someone to solve who is more
353    familiar with the sml/nj source (and socket programming):
354    
355    - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
356      exception since I could not figure out what the windows equivalent
357      of these functions is
358    
359    - In sockets-osdep.h there are a some #include statements that are
360      only used in a few files that include sockets-osdep.h
361    
362    - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
363      does not process its return value since I don't know how to
364      report an error upwards.
365    
366    - It would probably be good to have a call to WSACleanup() when
367      the library is unloaded (if there is such a possibility).
368      Otherwise I think Windows will take care of this automatically
369      when the process finishes.
370    
371    - I used ioctlsocket() as a replacement for ioctl() but I have
372      no idea if that is actually the proper replacement on Windows.
373    
374    - All these issues are marked in the code by "FIXME" comments.
375    
376    We use sml/nj extensively in our products and are quite happy
377    with it.  I hope this contribution will help you.
378    
379    Keep up the good work!
380    
381    David
382    
383    ----------------------------------------------------------------------
384    Name: Matthias Blume (blume (at) tti - c (dot) org)
385    Date: 2004/05/11 14:20:00 CDT
386    Tag: blume-20040511-installml
387    Description:
388    
389    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
390    the patch.)
391    
392    ----------------------------------------------------------------------
393    Name: Matthias Blume (blume (at) tti - c (dot) org)
394    Date: 2004/05/11 14:05:00 CDT
395    Tag: blume-20040511-nlffi-netbsd
396    Description:
397    
398    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
399    the patch.)
400    
401    ----------------------------------------------------------------------
402    Name: Matthias Blume (blume (at) tti - c (dot) org)
403    Date: 2004/05/11 12:05:00 CDT
404    Tag: blume-20040511-exports
405    Description:
406    
407    As per request by Adam Chlipala <adam@hcoop.net>, extended various
408    export lists in compiler-related .cm-files.
409    
410    ----------------------------------------------------------------------
411    Name: Matthias Blume (blume (at) tti - c (dot) org)
412    Date: 2004/05/11 11:35:00 CDT
413    Tag: blume-20040511-allsource
414    Description:
415    
416    The installer now honors the "src-smlnj" target again, although its meaning
417    has changed from "all sources required for the compiler" to "all sources
418    the installer knows about".  In other words, if you enable "src-smlnj"
419    in the "targets" file, then the installer will pull in sources for
420    everything.  (Notice that this refers to source code only.  Compiled
421    code is still only installed for modules that were requested explicitly
422    or which are required for other modules that were requested explicitly.)
423    
424    ----------------------------------------------------------------------
425    Name: Matthias Blume (blume (at) tti - c (dot) org)
426    Date: 2004/04/23 17:40:00 CDT
427    Tag: blume-20040423-ieee-scan
428    Description:
429    
430    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
431    in the exponent calculation we get INF or ZERO (depending on the mantissa
432    and the sign of the exponent).
433    
434    ----------------------------------------------------------------------
435    Name: Matthias Blume (blume (at) tti - c (dot) org)
436    Date: 2004/04/23 10:40:00 CDT
437    Tag: blume-20040423-ml-build
438    Description:
439    
440    The ml-build script now terminates with a non-0 status when something
441    goes wrong.
442    
443    ----------------------------------------------------------------------
444    Name: Matthias Blume (blume (at) tti - c (dot) org)
445    Date: 2004/04/22 16:35:00 CDT
446    Tag: blume-20040422-Option
447    Description:
448    
449    Made exception Option to be the same as exception Option.Option
450    (as it should be).
451    
452    ----------------------------------------------------------------------
453    Name: Allen Leung (leunga (at) reservoir (dot) com)
454    Date: 2004/03/19 14:40:00 EST
455    Tag: leunga-20040319-cygwin-nlffi
456    Description:
457    
458    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
459    of SML/NJ.  The problem is that
460    
461       lib = dlopen(NULL, ...)
462       f   = dlsym(lib, "malloc");
463    
464    does not work on Windows unless we explicitly export symbols
465    such as 'malloc' during linking.  We fixed this by explicitly
466    exporting the required symbols with the magic gcc incantation:
467    
468        -Wl,--export-all cygwin.def
469    
470    where cygwin.def is a file containing all the symbols that
471    we wish to export.
472    
473    I suspect this is a Windows problem and we'll have to
474    do the same (somehow with windows compilers) when
475    we build the native win32 version with the system
476    calls LoadLibrary/GetProcAddress.
477    
478    ----------------------------------------------------------------------
479    Name: Matthias Blume (blume (at) tti - c (dot) org)
480    Date: 2004/03/04 16:35:00 CST
481    Tag: blume-20040304-intinf-fmt
482    Description:
483    
484    Fixed problem with IntInf.fmt (sign would show up on the right instead
485    of on the left for BIN, OCT, and HEX).
486    
487    ----------------------------------------------------------------------
488    Name: Matthias Blume (blume (at) tti - c (dot) org)
489    Date: 2004/03/04 11:25:00 CST
490    Tag: blume-20040304-symlinks
491    Description:
492    
493    Fixed problem with installer script (unix only) where bin/ml-yacc and
494    friends pointed (via symlinks) to absolute locations instead of just
495    .run-sml.  This was reported by Vesa A Norrman.
496    
497    ----------------------------------------------------------------------
498    Name: Matthias Blume (blume (at) tti - c (dot) org)
499    Date: 2004/02/13 14:50:00 CST
500    Tag: Release_110_45
501    Description:
502    
503    New working version (110.45). New bootfiles.
504    
505    ----------------------------------------------------------------------
506    Name: Matthias Blume (blume (at) tti - c (dot) org)
507    Date: 2004/01/26 15:15:15 CST
508    Tag: blume-20040126-toplevel
509    Description:
510    
511    Improved handling of exceptions at the interactive toplevel.
512    
513    ----------------------------------------------------------------------
514    Name: Matthias Blume (blume (at) tti - c (dot) org)
515    Date: 2004/01/26 11:25:00
516    Tag: blume-20040126-app
517    Description:
518    
519    Type of top-level "app" corrected.
520    Added code for setting vp_limitPtrMask to Win32-specific runtime.
521    
522    ----------------------------------------------------------------------
523    Name: Matthias Blume (blume (at) tti - c (dot) org)
524    Date: 2003/11/18 17:10 CST
525    Tag: blume-20031118-basis-fiddle
526    Description:
527    
528    - changed Timer interface to what might become the spec
529    - POSIX_FLAGS -> BIT_FLAGS according to spec
530    - some other minor discrepancies wrt. spec eliminated
531    
532    ----------------------------------------------------------------------
533    Name: Matthias Blume (blume (at) tti - c (dot) org)
534    Date: 2003/11/06 12:00:00 CST
535    Tag: Release_110_44
536    Description:
537    
538    New working version (110.44). New bootfiles.
539    
540    ----------------------------------------------------------------------
541    Name: Matthias Blume (blume (at) tti - c (dot) org)
542    Date: 2003/11/04 11:50:00 CST
543    Tag: blume-20031104-move-libraries
544    Description:
545    
546    Eliminated the "dont_move_libraries" directive in config/targets.
547    (The mechanism was broken and could not be fixed easily.  Moreover,
548    there does not seem to be any reason not to move all libraries into
549    lib during installation.  I originally implemented this directive as a
550    backward-compatibility feature when I first introduced the new CM.
551    Now that things have been stable for a long time and going back to the
552    old CM is not an option, there is no reason to keep it around.)
553    
554    ----------------------------------------------------------------------
555    Name: Matthias Blume (blume (at) tti - c (dot) org)
556    Date: 2003/11/03 16:00:00 CST
557    Tag: blume-20031103-installdir
558    Description:
559    
560    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
561    Richards for pointing out the problem and providing the solution.)
562    
563    ----------------------------------------------------------------------
564    Name: Matthias Blume (blume (at) tti - c (dot) org)
565    Date: 2003/10/01 17:05:00 CDT
566    Tag: blume-20031001-lal-mlrisc
567    Description:
568    
569    MLRISC bug fix from Lal.
570    
571    ----------------------------------------------------------------------
572    Name: Matthias Blume (blume (at) tti - c (dot) org)
573    Date: 2003/09/30 16:10:00 CDT
574    Tag: blume-20030930-primio-bat
575    Description:
576    
577    1. Added openVector, nullRd, and nullWr to PRIM_IO.
578    2. Improved .bat files (for Win32 port) to make things work under Win95.
579       (thanks to Aaron S. Hawley for this one)
580    
581    ----------------------------------------------------------------------
582    Name: Matthias Blume (blume (at) tti - c (dot) org)
583    Date: 2003/09/26 16:05:00 CDT
584    Tag: blume-20030926-wrappriv
585    Description:
586    
587    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
588    
589    ----------------------------------------------------------------------
590    Name: Matthias Blume (blume (at) tti - c (dot) org)
591    Date: 2003/09/26 15:00:00 CDT
592    Tag: blume-20030926-110_43_3
593    Description:
594    
595    - additional cleanup
596    - version number bump, NEW BOOTFILES
597    
598    ----------------------------------------------------------------------
599    Name: Matthias Blume (blume (at) tti - c (dot) org)
600    Date: 2003/09/26 12:00:00 CDT
601    Tag: blume-20030926-ppautoload
602    Description:
603    
604    I modified the read-eval-print loop so that the autoloader gets
605    invoked whenever the prettyprinter tries to look up a symbol that
606    is not currently defined in the toplevel environment but which
607    appears in CM's autoload registry.  As a result, we see far fewer of
608    those ?.Foo.Bar.xxx names in the prettyprinter's output.
609    
610    In addition to this I tried to clean up some pieces of the Basis
611    implementation (e.g., Socket, Word8Array) in order to prevent other
612    instances of these ?.Foo.Bar.xxx names from being printed.
613    
614    The mechanism that picks names for types still needs some work, though.
615    (Right now it seems that if there is a type A.t which is defined to
616    be B.u, but B is unavailable at toplevel, then A.t gets printed as
617    "?.B.u" although the perhaps more sensible solution would be to use
618    "A.t" in this case.  In other words, the prettyprinter should follow
619    a chain of DEFtycs not farther than there are corresponding toplevel
620    names in the current environment.)
621    
622    ----------------------------------------------------------------------
623    Name: Matthias Blume (blume (at) tti - c (dot) org)
624    Date: 2003/09/24 16:31:00 CDT
625    Tag: blume-20030924-installer
626    Description:
627    
628    Another installer tweak:  All the ML code for the installer is now
629    compiled during CMB.make and put into a little library called
630    $smlnj/installer.cm.  The installation then simply invokes
631    
632       sml -m $smlnj/installer.cm
633    
634    and everything happens automagically.
635    
636    Win32:  ML code senses value of environment variable SMLNJ_HOME.
637    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
638            and BINDIR.
639    
640    The new scheme guarantees that the ML code responsible for the installation
641    is in sync with the APIs of the main system.  Also, the installer is
642    somewhat faster because the installer script is precompiled.
643    
644    ----------------------------------------------------------------------
645    Name: Matthias Blume (blume (at) tti - c (dot) org)
646    Date: 2003/09/24 15:35:00 CDT
647    Tag: blume-20030924-synsock
648    Description:
649    
650    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
651    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
652    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
653    implementations of non-blocking operations from CML's Socket
654    structure.
655    
656    ----------------------------------------------------------------------
657    Name: Matthias Blume (blume (at) tti - c (dot) org)
658    Date: 2003/09/24 15:10:05 CDT
659    Tag: blume-20030924-sockets
660    Description:
661    
662    1. Fixed SOCKET API and implementation to match Basis spec.
663       This required changing the internal representation of sockets to one
664       that remembers (for each socket file descriptor) whether it is currently
665       blocking or non-blocking.  This state is maintained lazily (i.e., a system
666       call is made only if the state actually needs to change).
667    
668    2. OS-specific details of sockets were moved into separate files, thus
669       making it possible to unify the bulk of the socket implementations
670       between Unix and Win32.
671    
672    3. CML's socket API changed accordingly.
673       (Note that we need to remove non-blocking functions from this API
674       since they are redundant in the case of CML!)
675    
676    4. CML's socket implementation now makes use of non-blocking functions
677       provided by Basis, thus removing all OS-dependent code from this part
678       of CML.
679    
680    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
681    
682    ----------------------------------------------------------------------
683    Name: Matthias Blume (blume (at) tti - c (dot) org)
684    Date: 2003/09/22 12:10:00 CDT
685    Tag: blume-20030922-110_43_2
686    Description:
687    
688    Made a new interim version and bootfiles for developer's bootstrapping
689    convenience.
690    
691       110.43.2 -- NEW BOOTFILES
692    
693    ----------------------------------------------------------------------
694    Name: Matthias Blume (blume (at) tti - c (dot) org)
695    Date: 2003/09/19 15:55:00 CDT
696    Tag: blume-20030919-cmdir
697    Description:
698    
699    1. new-install.sh -> install.sh
700    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
701    3. tweaked installer so that another name instead of .cm can be chosen
702       at install time (by setting the CM_DIR_ARC environment variable
703       during installation); once installation is complete, the name is
704       fixed
705    
706    ----------------------------------------------------------------------
707    Name: Matthias Blume (blume (at) tti - c (dot) org)
708    Date: 2003/09/18 16:00:00 CDT
709    Tag: blume-20030918-110_43_1
710    Description:
711    
712    Made a new interim version and bootfiles for developer's bootstrapping
713    convenience.
714    
715       110.43.1 -- NEW BOOTFILES
716    
717    ----------------------------------------------------------------------
718    Name: Matthias Blume (blume (at) tti - c (dot) org)
719    Date: 2003/09/18 15:20:00 CDT
720    Tag: blume-20030918-misc
721    Description:
722    
723    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
724       this seems to be controversial at the moment) and used those in
725       Posix.ProcEnv.times.
726    
727    2. Added Time.{from,to}Nanoseconds to Time.
728    
729    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
730       For example, fromLargeInt never needs to look at more than 3 "big
731       digits" to get its 53 bits of precision.
732    
733    ----------------------------------------------------------------------
734    Name: Matthias Blume (blume (at) tti - c (dot) org)
735    Date: 2003/09/17 16:30:00 CDT
736    Tag: blume-20030917-real32-slices
737    Description:
738    
739    Added an entry to the primitive environment
740    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
741    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
742    
743    Removed some of the "magic" constants in real64.sml and replaced them
744    with code that generates these values from their corresponding
745    integer counterparts.
746    
747    Made all(?) the slice-related changes to the Basis and made everything
748    compile again...
749    
750    ----------------------------------------------------------------------
751    Name: Matthias Blume (blume (at) tti - c (dot) org)
752    Date: 2003/09/15 17:45:00 CDT
753    Tag: blume-20030915-rbase
754    Description:
755    
756    Fixed bug in Real.fromLargeInt.
757    
758    ----------------------------------------------------------------------
759    Name: Matthias Blume (blume (at) tti - c (dot) org)
760    Date: 2003/09/13 18:11:00 CDT
761    Tag: blume-20030913-libinstall
762    Description:
763    
764    Minor bugfix in config/libinstall (set anchor with path to
765    standalone tool after installing it, otherwise libraries that
766    need ml-lex or ml-yacc won't compile the first time the installer
767    runs).
768    
769    ----------------------------------------------------------------------
770    Name: Matthias Blume (blume (at) tti - c (dot) org)
771    Date: 2003/09/12 11:45:00 CDT
772    Tag: blume-20030912-various
773    Description:
774    
775     - fixed bug in Real.toLargeInt
776     - fixed bug in Posix.ProcEnv.times
777     - changed inputLine functions to return an option
778     - minor installer improvements / bugfixes
779     - changed default @SMLalloc parameter for x86/celeron to 64k
780    
781    ----------------------------------------------------------------------
782    Name: Matthias Blume (blume (at) tti - c (dot) org)
783    Date: 2003/09/09 22:00:00 CDT
784    Tag: Release_110_43
785    Description:
786    
787        New working release 110.43.  New bootfiles.
788    
789    ----------------------------------------------------------------------
790    Name: Matthias Blume (blume (at) tti - c (dot) org)
791    Date: 2003/09/09 19:20:00 CDT
792    Tag: blume-20030909-installer
793    Description:
794    
795    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
796    Modified config/install.bat to take advantage of it.  Also modified
797    config/install.sh (and called it config/new-install.sh) to take advantage
798    of it on Unix systems.  (The SML code is (supposed to be) platform-
799    independent.)
800    
801    The installer can now install everything under Win32
802    as well as under *nix as long as it compiles.
803    
804    Other changes:
805    
806      - made CML compile again under Win32
807      - made eXene compile under Win32 (by providing a fake structure UnixSock
808        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
809      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
810        same as type int;  under Win32 it isn't
811      - fixed some slice-related problems in the win32-specific parts of CML
812      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
813        (under Win32, the volume name is case-insensitive, and the
814        OS.Path code compares volume names for equality)
815    
816    ----------------------------------------------------------------------
817    Name: Matthias Blume (blume (at) tti - c (dot) org)
818    Date: 2003/09/08 11:55:00 CDT
819    Tag: blume-20030908-fullpath
820    Description:
821    
822    Made Win32 version of OS.FileSys.fullPath return current directory
823    when given an empty string.  This is what the spec says, and incidentally,
824    CM depends on it.  (CM otherwise goes into an infinite loop in certain
825    cases when presented with the name of a non-existing .cm file.)
826    
827    ----------------------------------------------------------------------
828    Name: Matthias Blume (blume (at) tti - c (dot) org)
829    Date: 2003/09/04 16:30:00 CDT
830    Tag: blume-20030905-slices-etc
831    Description:
832    
833      1. Changed interface to vectors and arrays in Basis to match
834         (draft) Basis spec.
835      2. Added signatures and implementations of slices according to
836         Basis spec.
837      3. Edited source code throughout the system to make it compile again
838         under 1. and 2.  (In some cases code had to be added to have it
839         match the new signatures.)
840      4. MLRISC should be backward-compatible: the copies of the originals
841         of files that needed to change under 3. were retained, the .cm files
842         check the compiler version number and use old versions when
843         appropriate.
844      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
845         dirstream -> string option (in accordance with Basis spec).
846      6. When generating code that counts lines, ml-lex used function
847         CharVector.foldli, taking advantage of its old interface.
848         This has been replaced with the corresponding code from
849         CharVectorSlice.  (html-lex must be re-lexed!)
850      7. BitArray in smlnj-lib/Util has been extended/modified to match the
851         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
852      8. Removed temporary additions (fromInternal, toInternal) from the
853         (now obsolete) IntInf in smlnj-lib/Util.
854      9. Cleaned up structure Byte.
855     10. Added localOffset, scan, and fromString to Date (according to spec).
856         Cleaned/corrected implementation of Date.
857         (Still need to check for correctness; implement better canonicalizeDate.)
858     11. Added "scan" to signature IEEE_REAL.
859     12. Some improvements to IntInf [in particular: efficiency-hack for
860         mod and rem when second operand is 2 (for parity checks).]
861     13. Changed representation of type Time.time, using a single IntInf.int
862         value counting microseconds.  This considerably simplified the
863         implementation of structure Time.  We now support negative time
864         values; scan and fromString handle signs.
865     14. Functor PrimIO now takes two additional arguments (VectorSlice and
866         ArraySlice).
867    
868    ----------------------------------------------------------------------
869    Name: Matthias Blume (blume (at) tti - c (dot) org)
870    Date: 2003/08/28 17:00:00 CDT
871    Tag: blume-20030828-intinf
872    Description:
873    
874    This is a major update which comes with a version number bump
875    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
876    and an implementation of IntInf in the Basis.
877    
878    There are a fairly large number of related changes and updates throughout
879    the system:
880    
881    Basis:
882            - Implemented IntInf.
883            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
884            - Added some missing Real64 operations, most notably Real.toLargeInt.
885            - Added FixedInt as a synonym for Int32.
886    
887    compiler:
888            * Added support for a built-in intinf type.
889              - literals
890              - pattern matching
891              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
892              - overloading on literals and operations
893    
894            This required adding a primitive type intinf, some additional
895            primops, and implementations for several non-trivial intinf
896            operations in Core.  (The intinf type is completely abstract
897            to the compiler; all operations get delegated back to the Core.)
898    
899            * Intinf equality is handled by polyequal.  However, the compiler
900            does not print its usual warning in this case (since polyequal
901            is the right thing to do there).
902    
903            * Improved the organization of structure InlineT.
904    
905            * A word about conversion primops:
906            If conversions involving intinf do not cancel out during
907            CPS contract, then the compiler must insert calls to Core functions.
908            Since all core access must be resolved already during the FLINT
909            translate phase, it would be too late a the time of CPS contract
910            to add new Core calls.  For this reason, conversion primops
911            for intinf carry two arguments:  1. the numeric argument that
912            they are supposed to convert, and 2. the Core function that
913            can help with this conversion if necessary.  If CPS contract
914            eliminates a primop, then the associated Core function becomes
915            dead and goes away.  Intinf conversion primops that do not get
916            eliminated by CPS contract get rewritten into calls of their
917            core functions by a separate, new phase.
918    
919    interactive system:
920            - Control.Print.intinfDepth controls max length of intinf constants
921            being printed. (Analogous to Control.Print.stringDepth.)
922            - Cleanup in printutil and pputil: got rid of unused stuff and
923            duplicates; replaced some of the code with code that makes better
924            use of library functionality.
925    
926    CM:
927            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
928            are no longer silent.
929    
930    CKIT:
931            Fixed mismatched uses of Int32 and LargeInt.  I always decided
932            in favor of LargeInt -- which is now the same as IntInf.
933            CKIT-knowledgable people should check whether this is what's
934            intended and otherwise change things back to using Int32 or
935            FixedInt.
936    
937    Throughout the code:
938            Started using IntInf.int literals and built-in operations
939            (e.g., comparison with 0) where this seems appropriate.
940    
941    
942    ----------------------------------------------------------------------
943    Name: Dave MacQueen (dbm@cs.uchicago.edu)
944    Date: 2003/08/13 11:36:00 CDT
945    Tag: dbm-20030813-mcz-merge1
946    Description:
947    
948    Merging changes from the mcz-branch development branch into trunk.
949    These changes involve replacement of the emulated old prettyprinter
950    interface with direct use of the SML/NJ Lib PP library, and fixing
951    of a couple of bugs (895, 1186) relating to error messages.  A new
952    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
953    has been added.
954    
955    ----------------------------------------------------------------------
956    Name: Matthias Blume (blume (at) tti - c (dot) org)
957    Date: 2003/08/11 15:45:00 CDT
958    Tag: blume-20030811-windows
959    Description:
960    
961    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
962    
963        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
964    
965    This patch restores SML/NJ's ability to run under win32.  There are a
966    number of changes, including fixes for several bugs that had gone
967    unnoticed until now:
968    
969      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
970        trying to run the win32 version from within cygwin.)
971    
972      - There are a number of simple .bat scripts that substitute for their
973        corresponding Unix shell-scripts.  (See below.)
974    
975      - The internals of ml-build have been modified slightly.  The main
976        difference is that instead of calling ".link-sml" (or link-sml.bat)
977        using OS.Process.system, the ML process delegates this task back
978        to the script.  Otherwise problems arise in mixed environments such
979        as Cygwin where scripts look and work like Unix scripts, but
980        where OS.Process.system cannot run them.
981    
982      - In CM, the srcpath pickler used native pathname syntax -- which
983        is incorrect in the case of cross-compilation.  The new pickle format
984        is independent of platform-specific naming conventions.
985    
986      - Path configuration files (such as lib/pathconfig) can now choose
987        between native and standard syntax.  Placing a line of the form
988    
989            standard!
990    
991        into the file causes all subsequent paths to be interpreted using
992        CM standard pathname syntax (= Unix conventions); a line
993    
994            native!
995    
996        switches back to native style.  This was needed so that
997        path config files can be written portably, see src/system/pathconfig.
998    
999      - Runtime system:
1000    
1001          - win32-filesys.c:  get_file_time and set_file_time now
1002            access modification time, not creation time.
1003    
1004          - I/O code made aware of new array representation.
1005    
1006          - Bug fixes in X86.prim.masm.
1007    
1008       - src/system/makeml made aware of win32. (For use under cygwin
1009            and other Unix-environments for windows.)
1010    
1011       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
1012         which caused BinIO.inputAll to fail consistently.
1013    
1014    .bat scripts:
1015    
1016       Windows .bat scripts assume that SMLNJ_HOME is defined.
1017    
1018       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
1019         applications (sml, ml-yacc, ml-lex).
1020       - ml-build.bat: analogous to ml-build.
1021       - config\install.bat:  Analogous to config/install.sh.  This requires
1022         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
1023         (nmake etc. must be on the path, and vcvars32 must have been run.)
1024         Moreover, sources for ml-lex and ml-yacc need to exist under src,
1025         and the bootfile hierarchy must have been unpacked under
1026         sml.boot.x86-win32.
1027         The script is very primitive and does a poor job at error checking.
1028         It only installs the base system, ml-lex, and ml-yacc.  No other
1029         libraries are being installed (i.e., you get only those that
1030         are part of the compiler.)
1031       - link-sml.bat:  analogous to .link-sml, but not currently used
1032    
1033    Unrelated bug fixes:
1034    
1035      - ml-nlffigen now exports structures ST_* corresponding to incomplete
1036        types.
1037      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
1038        otherwise.)
1039    
1040    ----------------------------------------------------------------------
1041    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
1042    Date: 2003/06/17
1043    Tag: macqueen-20030617-bug895
1044    Description:
1045    
1046    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
1047    Tag will be used for new development branch (mcz-branch) for
1048    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
1049    
1050    ----------------------------------------------------------------------
1051    Name: Matthias Blume (blume (at) tti - c (dot) org)
1052    Date: 2003/05/27 16:55:00 CDT
1053    Tag: blume-20030527-polyeq
1054    Description:
1055    
1056    Tried to eliminated most cases of polymorphic equality.
1057    
1058    ----------------------------------------------------------------------
1059    Name: Matthias Blume (blume (at) tti - c (dot) org)
1060    Date: 2003/05/21 17:45:00 CDT
1061    Tag: blume-20030517-complete
1062    Description:
1063    
1064    Two changes:
1065    
1066      1. Added a flag for controlling whether non-exhaustive bindings will
1067         be treated as errors (default is false).
1068      2. Cleaned up the *entire* source tree so that CMB.make goes through
1069         without a single non-exhaustive match- or bind warning.
1070    
1071    ----------------------------------------------------------------------
1072    Name: Matthias Blume (blume (at) tti - c (dot) org)
1073    Date: 2003/05/17 10:20:00 CDT
1074    Tag: blume-20030517-absyn
1075    Description:
1076    
1077    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
1078    
1079       This mainly affects the quality of error messages.  However, some
1080       of the code is now more straightforward than before. (Treatment of
1081       the above four constructs in translate.sml is much simpler than
1082       the "macro-expansion" that was going on before.  Plus, the mach-
1083       compiler no longer gets invoked just to be able to compile an
1084       if-expression.)
1085    
1086    2. The ErrorMsg.Error exception is now caught and absorbed by the
1087       interactive loop.
1088    
1089    ----------------------------------------------------------------------
1090    Name: Allen Leung
1091    Date: 2003/05/16 13:05:00 CDT
1092    Tag: leunga-20030516-cygwin-runtime
1093    Description:
1094    
1095        Ported the runtime system to cygwin, which uses the unix
1096    x86-unix bin files.  Missing/buggy features:
1097    
1098       o getnetbyname, getnetbyaddr: these functions seem to be missing in
1099         the Cygwin library.
1100       o Ctrl-C handling may be flaky.
1101       o Windows system calls and Windows I/O are not supported.
1102    
1103       A new set of binfiles is located at:
1104    
1105         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
1106    
1107       This is only needed for bootstrapping the cygwin version of smlnj.
1108       Other x86 versions can use the existing binfiles.
1109    
1110    ----------------------------------------------------------------------
1111    Name: Matthias Blume
1112    Date: 2003/04/08 15:42:00 CDT
1113    Tag: blume-20030408-listpair
1114    Description:
1115    
1116    1. Added a target 'mlrisc' to installer.
1117    
1118    2. Added missing elements to structure ListPair.
1119    
1120    ----------------------------------------------------------------------
1121    Name: Allen Leung
1122    Date: 2003/01/07 10:40:00 EST
1123    Tag: leunga-20030107-int-rem
1124    Description:
1125    
1126    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
1127    The arguments to the SUBL instruction were swapped.
1128    
1129    ----------------------------------------------------------------------
1130    Name: Matthias Blume
1131    Date: 2002/12/12 16:25:00 EST
1132    Tag: blume-20021212-risc-ra
1133    Description:
1134    
1135    Fixed a serious bug in the rewrite code for FP spilling/reloading that
1136    sent the RA into an infinite loop when floating point registers get
1137    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
1138    110.37 and 110.38.)
1139    There was another set of potential problems related to the handling of
1140    MLRISC annotations (but those did not yet cause real problems, apparently).
1141    
1142    ----------------------------------------------------------------------
1143    Name: Matthias Blume
1144    Date: 2002/12/06 22:40:00 EST
1145    Tag: blume-20021206-cm-fileid
1146    Description:
1147    
1148    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
1149    This fixes the problem of CM getting confused by files that suddenly
1150    change their identity (e.g., by getting unlinked and recreated by some
1151    text editor such as vi).  There might be a better/cheaper/cleaner way
1152    of doing this, but for now this will have to do.
1153    
1154    ----------------------------------------------------------------------
1155    Name: Matthias Blume
1156    Date: 2002/10/28 09:50:00 EST
1157    Tag: blume-20021028-typecheck
1158    Description:
1159    
1160    Exported structure Typecheck from $smlnj/viscomp/core.cm.
1161    
1162    ----------------------------------------------------------------------
1163    Name: Matthias Blume
1164    Date: 2002/10/17 09:10:00 EDT
1165    Tag: Release_110_42
1166    Description:
1167    
1168    In good old tradition, there has been a slight hiccup so that we have
1169    to patch 110.42 after the fact.   The old release tag has been replaced
1170    (see below).
1171    
1172    The change solves a problem with two competing approaches the
1173    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
1174    each other's way.
1175    
1176    This change only affects the runtime system code and the installer script.
1177    (No new bootfiles.)
1178    
1179    ----------------------------------------------------------------------
1180    Name: Matthias Blume
1181    Date: 2002/10/16 12:00:00 EDT
1182    Tag: Release_110_42_removed
1183    Description:
1184    
1185    New working release.  New bootfiles.
1186    
1187    ----------------------------------------------------------------------
1188    Name: Matthias Blume
1189    Date: 2002/10/10 13:10:00 EDT
1190    Tag: blume-20021010-ppc-divs
1191    Description:
1192    
1193    The mltree operator DIVS must be implemented with an overflow check on
1194    the PPC because the hardware indicates divide-by-zero using "overflow" as
1195    well.
1196    
1197    ----------------------------------------------------------------------
1198    Name: Matthias Blume
1199    Date: 2002/07/23 11:45:00
1200    Tag: blume-20020723-smlnj-home
1201    Description:
1202    
1203    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
1204    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
1205    CM_PATHCONFIG is also set), the path configuration file is assumed
1206    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
1207    the entire tree to some other place and everything will "just work".
1208    
1209    (Companion commands such as ml-build and ml-makedepend also sense this
1210    variable.)
1211    
1212    ----------------------------------------------------------------------
1213    Name: Matthias Blume
1214    Date: 2002/07/12 21:19:00 EDT
1215    Tag: blume-20020712-liveness
1216    Description:
1217    
1218    Exported two useful "step" functions from liveness module (MLRISC).
1219    
1220    ----------------------------------------------------------------------
1221    Name: Matthias Blume
1222    Date: 2002/07/05 16:00 EDT
1223    Tag: Release_110_41
1224    Description:
1225    
1226    New working release.  New bootfiles.
1227    
1228    ----------------------------------------------------------------------
1229    Name: Matthias Blume
1230    Date: 2002/07/05 10:25:00 EDT
1231    Tag: blume-20020705-btimp
1232    Description:
1233    
1234    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
1235    other clients can set up backtracing support.
1236    
1237    ----------------------------------------------------------------------
1238    Name: Matthias Blume
1239    Date: 2002/06/25 17:23:00 EDT
1240    Tag: blume-20020625-fpmax
1241    Description:
1242    
1243    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
1244    case.  (The sense of the isNaN test was reversed -- which made min and
1245    max always return their first argument.)
1246    
1247    ----------------------------------------------------------------------
1248    Name: Matthias Blume
1249    Date: 2002/06/11
1250    Tag: blume-20020611-unixpath
1251    Description:
1252    
1253    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
1254    
1255    ----------------------------------------------------------------------
1256    Name: Matthias Blume
1257    Date: 2002/06/10 16:35:00 EDT
1258    Tag: blume-20020610-ieeereal
1259    Description:
1260    
1261    I back-ported my implementation of IEEEReal.fromString from the
1262    idlbasis-devel branch so that we can test it.
1263    
1264    Another small change is that ppDec tries to give more information
1265    than just "<sig>" in the case of functors.  However, this code is
1266    broken in some mysterious way if the functor's body's signature
1267    has not been declared by ascription but gets inferred from the
1268    implementation.  This needs fixing...
1269    
1270    ----------------------------------------------------------------------
1271    Name: Matthias Blume
1272    Date: 2002/05/31
1273    Tag: blume-20020531-btrace-mode
1274    Description:
1275    
1276    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
1277    the wayside when I switched over to using Controls everywhere.)
1278    
1279    ----------------------------------------------------------------------
1280    Name: Lal George
1281    Date: 2002/05/23 12:21:40 EDT
1282    Tag: george-20020523-visual-labels
1283    Description:
1284    
1285    Labels are now displayed in the graphical output to make
1286    the fall-through and target blocks obvious.
1287    
1288    ----------------------------------------------------------------------
1289    Name: Matthias Blume
1290    Date: 2002/05/22 11:03:00 EDT
1291    Tag: blume-20020522-shrink
1292    Description:
1293    
1294    John tweaked yesterday's fix for 1131 to handle an out-of-memory
1295    situation that comes up when allocating huge arrays.
1296    
1297    ----------------------------------------------------------------------
1298    Name: Matthias Blume
1299    Date: 2002/05/21 16:00:00 EDT
1300    Tag: Release_110_40
1301    Description:
1302    
1303    New working release (110.40).  New bootfiles.
1304    
1305    [Also: John Reppy fixed GC bug 1131.]
1306    
1307    ----------------------------------------------------------------------
1308    Name: Matthias Blume
1309    Date: 2002/05/21 12:35:00 EDT
1310    Tag: blume-20020521-cmdoc
1311    Description:
1312    
1313    CM documentation update.
1314    
1315    ----------------------------------------------------------------------
1316    Name: Matthias Blume
1317    Date: 2002/05/21 10:55:00 EDT
1318    Tag: blume-20020521-misc
1319    Description:
1320    
1321    - John tweaked runtime to be silent on heap export (except when
1322      GC messages are on).
1323    - I added a few more things (cross-compiling versions of CMB) to
1324      config/preloads (as suggestions).
1325    
1326    ----------------------------------------------------------------------
1327    Name: Matthias Blume
1328    Date: 2002/05/20 22:25:00 EDT
1329    Tag: blume-20020520-controls
1330    Description:
1331    
1332    - Added ControlUtil structure to control-lib.cm.
1333    - Use it throughout.
1334    - Used Controls facility to define MLRISC controls (as opposed to
1335      registering MLRISC control ref cells with Controls after the fact)
1336    - Fixed messed-up controls priorities.
1337    
1338    * Removed again all the stuff from config/preloads that one wouldn't
1339      be able to preload at the time the initial heap image is built.
1340      (Many libraries, e.g., CML, do not exist yet at this time.  The
1341       only libraries that can be preloaded via config/preloads are those
1342       that come bundled with the bootfiles.)
1343    
1344    ----------------------------------------------------------------------
1345    Name: Matthias Blume
1346    Date: 2002/05/20 10:59:00 EDT
1347    Tag: blume-20020520-preloads
1348    Description:
1349    
1350    Added a lot of commented-out suggestions for things to be included
1351    in config/preloads.
1352    
1353    ----------------------------------------------------------------------
1354    Name: Allen Leung
1355    Date: 2002/05/18 14:20:00 EDT
1356    Tag: leunga-20020518-mdl
1357    Description:
1358    
1359    o Made the mdl tool stuff compile and run again.
1360    o I've disabled all the stuff that depends on RTL specifications; they
1361      are all badly broken anyway.
1362    
1363    ----------------------------------------------------------------------
1364    Name: Matthias Blume
1365    Date: 2002/05/17 16:49:00 EDT
1366    Tag: blume-20020517-controls
1367    Description:
1368    
1369    0. John Reppy made several modifications to the SML/NJ library.
1370       In particular, there is a shiny new controls-lib.cm.
1371    
1372    1. Pushed new controls interface through compiler so that everything
1373       compiles again.
1374    
1375    2. Added FormatComb and FORMAT_COMB to the CML version of the
1376       SML/NJ library (so that CML compiles again).
1377    
1378    3. Modified init scripts because XXX_DEFAULT environment variables
1379       are no longer with us.  (Boot-time initialization is now done
1380       using the same environment variables that are also used for
1381       startup-time initialization of controls.)
1382    
1383    ----------------------------------------------------------------------
1384    Name: Lal George
1385    Date: 2002/05/15 09:20:10 EDT
1386    Tag: george-20020515-pseudo-op-decls
1387    Description:
1388    
1389            All pseudo-ops emitted before the first segment declaration
1390    such as TEXT, DATA, and BSS directives are assumed to be global
1391    declarations and are emitted first in the assembly file. This is
1392    useful in a number of situations where one has pseudo-ops that are not
1393    specific to any segment, and also works around the constraint that one
1394    cannot have client pseudo-ops in the TEXT segment.
1395    
1396            Because no segment is associated with these declarations it is
1397    an error to allocate any space or objects before the first segment
1398    directive and an exception will be raised. However, we cannot make
1399    this check for client pseudo-ops.
1400    
1401            These top level declarations are a field in the CFG graph_info.
1402    In theory you can continue to add to this field after the CFG has been
1403    built -- provided you know what you are doing;-)
1404    
1405    ----------------------------------------------------------------------
1406    Name: Matthias Blume
1407    Date: 2002/05/13 16:40:00 EDT
1408    Tag: blume-20020513-pp-etc
1409    Description:
1410    
1411    A few minor bugfixes:
1412    
1413      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1414        (Bogus printouts for redefined bindings still occur.  Compiler
1415        bug should no longer occur now.  We need to redo the prettyprinter
1416        from scratch.)
1417    
1418      - CM pathname printer now also adds escape sequences for ( and )
1419    
1420      - commend and docu fixes for ml-nlffi
1421    
1422    ----------------------------------------------------------------------
1423    Name: Matthias Blume
1424    Date: 2002/05/10 16:40:00 EDT
1425    Tag: blume-20020510-erg-textio
1426    Description:
1427    
1428    Applied the following bugfix provided by Emden Gansner:
1429    
1430        Output is corrupted when outputSubstr is used rather than output.
1431    
1432        The problem occurs when a substring
1433    
1434            ss = (s, dataStart, dataLen)
1435    
1436        where dataStart > 0, fills a stream buffer with avail bytes left.
1437        avail bytes of s, starting at index dataStart, are copied into the
1438        buffer, the buffer is flushed, and then the remaining dataLen-avail
1439        bytes of ss are copied into the beginning of the buffer. Instead of
1440        starting this copy at index dataStart+avail in s, the current code
1441        starts the copy at index avail.
1442    
1443        Fix:
1444        In text-io-fn.sml, change line 695 from
1445             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1446        to
1447             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1448    
1449    ----------------------------------------------------------------------
1450    Name: Matthias Blume
1451    Date: 2002/04/12 13:55:00 EDT
1452    Tag: blume-20020412-assyntax
1453    Description:
1454    
1455    1. Grabbed newer assyntax.h from the XFree86 project.
1456    2. Fiddled with how to compile X86.prim.asm without warnings.
1457    3. (Very) Minor cleanup in CM.
1458    
1459    ----------------------------------------------------------------------
1460    Name: Matthias Blume
1461    Date: 2002/04/01 (no joke!) 17:07:00 EST
1462    Tag: blume-20020401-x86div
1463    Description:
1464    
1465    Added full support for div/mod/rem/quot on the x86, using the machine
1466    instruction's two results (without clumsily recomputing the remainder)
1467    directly where appropriate.
1468    
1469    Some more extensive power-of-two support was added to the x86 instruction
1470    selector (avoiding expensive divs, mods, and muls where they can be
1471    replaced with cheaper shifts and masks).  However, this sort of thing
1472    ought to be done earlier, e.g., within the CPS optimizer so that
1473    all architectures benefit from it.
1474    
1475    The compiler compiles to a fixed point, but changes might be somewhat
1476    fragile nevertheless.  Please, report any strange things that you might
1477    see wrt. div/mod/quot/rem...
1478    
1479    ----------------------------------------------------------------------
1480    Name: Matthias Blume
1481    Date: 2002/03/29 17:22:00
1482    Tag: blume-20020329-div
1483    Description:
1484    
1485    Fixed my broken div/mod logic.  Unfortunately, this means that the
1486    inline code for div/mod now has one more comparison than before.
1487    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1488    The problem was with quotient = 0, because that alone does not tell
1489    us which way the rounding went.  One then has to look at whether
1490    remainder and divisor have the same sign...  :(
1491    
1492    Anyway, I replaced the bootfiles with fresh ones...
1493    
1494    ----------------------------------------------------------------------
1495    Name: Matthias Blume
1496    Date: 2002/03/29 14:10:00 EST
1497    Tag: blume-20020329-inlprims
1498    Description:
1499    
1500    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1501    
1502    Primops have changed. This means that the bin/boot-file formats have
1503    changed as well.
1504    
1505    To make sure that there is no confusion, I made a new version.
1506    
1507    
1508    CHANGES:
1509    
1510    * removed REMT from mltree (remainder should never overflow).
1511    
1512    * added primops to deal with divisions of all flavors to the frontend
1513    
1514    * handled these primops all the way through so they map to their respective
1515      MLRISC support
1516    
1517    * used these primops in the implementation of Int, Int32, Word, Word32
1518    
1519    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1520    
1521    * parameterized INLMIN, INLMAX, and INLABS by a numkind
1522    
1523    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1524      floating point
1525    
1526    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1527      and Real (but Real.abs maps to a separate floating-point-only primop)
1528    
1529    
1530    TODO items:
1531    
1532    * Hacked Alpha32 instruction selection, disabling the selection of REMx
1533      instructions because the machine instruction encoder cannot handle
1534      them.  (Hppa, PPC, and Sparc instruction selection did not handle
1535      REM in the first place, and REM is supported by the x86 machine coder.)
1536    
1537    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1538      selection phase.  (The two can be streamlined because the hardware
1539      delivers both quotient and remainder at the same time anyway.)
1540    
1541    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1542      (Currently the behavior is inconsistent both across architectures and
1543      wrt. the draft Basis spec.)
1544    
1545    * Word8 should eventually be handled natively, too.
1546    
1547    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1548      as if there currently is no execution path that could trigger it in
1549      SML/NJ.  (The assumptions underlying functions arith and promotable do not
1550      hold for things like multiplication and division.)
1551    
1552    ----------------------------------------------------------------------
1553    Name: Matthias Blume
1554    Date: 2002/03/27 16:27:00 EST
1555    Tag: blume-20020327-mlrisc-divisions
1556    Description:
1557    
1558    Added support for all four division operations (ML's div, mod, quot,
1559    and rem) to MLRISC.  In the course of doing so, I also rationalized
1560    the naming (no more annoying switch-around of DIV and QUOT), by
1561    parameterizing the operation by div_rounding_mode (which can be either
1562    DIV_TO_ZERO or DIV_TO_NEGINF).
1563    
1564    The generic MLTreeGen functor takes care of compiling all four
1565    operations down to only round-to-zero div.
1566    
1567    Missing pieces:
1568    
1569      * Doing something smarter than relying on MLTreeGen on architectures
1570        like, e.g., the x86 where hardware division delivers both quotient and
1571        remainder at the same time.  With this, the implementation of the
1572        round-to-neginf operations could be further streamlined.
1573    
1574      * Remove inlining support for div/mod/rem from the frontend and replace it
1575        with primops that get carried through to the backend.  Do this for all
1576        int and word types.
1577    
1578    ----------------------------------------------------------------------
1579    Name: Matthias Blume
1580    Date: 2002/03/25 17:25:00 EST
1581    Tag: blume-20020325-divmod
1582    Description:
1583    
1584    I improved (hopefully without breaking them) the implementation of Int.div,
1585    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1586    advantage of the following observations:
1587    
1588      Let  q = x quot y      r = x rem y
1589           d = x div  y      m = x mod y
1590    
1591    where "quot" is the round-to-zero version of integer division that
1592    hardware usually provides.  Then we have:
1593    
1594         r = x - q * y        where neither the * nor the - will overflow
1595         d = if q >= 0 orelse x = q * y then q else q - 1
1596                              where neither the * nor the - will overflow
1597         m = if q >= 0 orelse r = 0 then r else r + y
1598                              where the + will not overflow
1599    
1600    This results in substantial simplification of the generated code.
1601    The following table shows the number of CFG nodes and edges generated
1602    for
1603            fun f (x, y) = x OPER y
1604            (* with OPER \in div, mod, quot, rem *)
1605    
1606    
1607        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1608        --------------------------------------------------------
1609         div |         24 |         39 |         12 |         16
1610         mod |         41 |         71 |         12 |         16
1611        quot |          8 |         10 |          8 |         10
1612         rem |         10 |         14 |          8 |         10
1613    
1614    
1615    ----------------------------------------------------------------------
1616    Name: Matthias Blume
1617    Date: 2002/03/25 22:06:00 EST
1618    Tag: blume-20020325-cprotobug
1619    Description:
1620    
1621    Fixed a bug in cproto (c prototype decoder).
1622    
1623    ----------------------------------------------------------------------
1624    Name: Matthias Blume
1625    Date: 2002/03/25 16:00:00 EST
1626    Tag: blume-20020325-raw-primops
1627    Description:
1628    
1629    I did some cleanup to Allen's new primop code and
1630    replaced yesterday's bootfiles with new ones.
1631    (But they are stored in the same place.)
1632    
1633    ----------------------------------------------------------------------
1634    Name: Matthias Blume
1635    Date: 2002/03/24 22:40:00 EST
1636    Tag: blume-20020324-bootfiles
1637    Description:
1638    
1639    Made the bootfiles that Allen asked for.
1640    
1641    ----------------------------------------------------------------------
1642    Name: Allen Leung
1643    Date: 2002/03/23 15:50:00 EST
1644    Tag: leunga-20020323-flint-cps-rcc-primops
1645    Description:
1646    
1647      1. Changes to FLINT primops:
1648    
1649        (* make a call to a C-function;
1650         * The primop carries C function prototype information and specifies
1651         * which of its (ML-) arguments are floating point. C prototype
1652         * information is for use by the backend, ML information is for
1653         * use by the CPS converter. *)
1654      | RAW_CCALL of { c_proto: CTypes.c_proto,
1655                       ml_args: ccall_type list,
1656                       ml_res_opt: ccall_type option,
1657                       reentrant : bool
1658                     } option
1659       (* Allocate uninitialized storage on the heap.
1660        * The record is meant to hold short-lived C objects, i.e., they
1661        * are not ML pointers.  With the tag, the representation is
1662        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1663        *)
1664      | RAW_RECORD of {tag:bool,sz:int}
1665      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1666    
1667      2.  These CPS primops are now overloaded:
1668    
1669           rawload of {kind:numkind}
1670           rawstore of {kind:numkind}
1671    
1672          The one argument form is:
1673    
1674             rawload {kind} address
1675    
1676          The two argument form is:
1677    
1678             rawload {kind} [ml object, byte-offset]
1679    
1680      3. RAW_CCALL/RCC now takes two extra arguments:
1681    
1682         a. The first is whether the C call is reentrant, i.e., whether
1683            ML state should be saved and restored.
1684         b. The second argument is a string argument specifying the name of
1685            library and the C function.
1686    
1687         These things are currently not handled in the code generator, yet.
1688    
1689      4. In CProto,
1690    
1691         An encoding type of "bool" means "ml object" and is mapped into
1692         C prototype of PTR.  Note that "bool" is different than "string",
1693         even though "string" is also mapped into PTR, because "bool"
1694         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1695    
1696      5. Pickler/unpicker
1697    
1698         Changed to handle RAW_RECORD and newest RAW_CCALL
1699    
1700      6. MLRiscGen,
1701    
1702         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1703         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1704            in the file CodeGen/cpscompile/cps-c-calls.sml
1705    
1706      7. Added the conditional move operator
1707    
1708             condmove of branch
1709    
1710         to cps.  Generation of this is still buggy so it is currently
1711         disabled.
1712    
1713    ----------------------------------------------------------------------
1714    Name: Lal George
1715    Date: 2002/03/22 14:18:25 EST
1716    Tag: george-20020322-cps-branch-prob
1717    Description:
1718    
1719    Implemented the Ball-Larus branch prediction-heuristics, and
1720    incorporated graphical viewers for control flow graphs.
1721    
1722    Ball-Larus Heuristics:
1723    ---------------------
1724    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1725    
1726    By design it uses the Dempster-Shafer theory for combining
1727    probabilities.  For example, in the function:
1728    
1729        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1730    
1731    the ball-larus heuristics predicts that the n=0 is unlikely
1732    (OH-heuristic), and the 'then' branch is unlikely because of the
1733    RH-heuristic -- giving the 'then' branch an even lower combined
1734    probability using the Dempster-Shafer theory.
1735    
1736    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1737    probability of the 'then' branch because of the loop in the else
1738    branch.
1739    
1740    
1741    Graphical Viewing:
1742    ------------------
1743    I merely plugged in Allen's graphical viewers into the compiler. The
1744    additional code is not much. At the top level, saying:
1745    
1746            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1747    
1748    will display the graphical view of the control flow graph just before
1749    back-patching.  daVinci must be in your path for this to work. If
1750    daVinci is not available, then the default viewer can be changed
1751    using:
1752    
1753            Control.MLRISC.getString "viewer"
1754    
1755    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1756    course, these viewers must be in your path.
1757    
1758    The above will display the compilation unit at the level of clusters,
1759    many of which are small, boring, and un-interesting. Also setting:
1760    
1761            Control.MLRISC.getInt "cfg-graphical-view_size"
1762    
1763    will display clusters that are larger than the value set by the above.
1764    
1765    
1766    ----------------------------------------------------------------------
1767    Name: Matthias Blume
1768    Date: 2002/03/21 22:20:00 EST
1769    Tag: blume-20020321-kmp-bugfix
1770    Description:
1771    
1772    Changed the interface to the KMP routine in PreString and fixed
1773    a minor bug in one place where it was used.
1774    
1775    ----------------------------------------------------------------------
1776    Name: Allen Leung
1777    Date: 2002/03/21 20:30:00 EST
1778    Tag: leunga-20020321-cfg
1779    Description:
1780    
1781      Fixed a potential problem in cfg edge splitting.
1782    
1783    ----------------------------------------------------------------------
1784    Name: Allen Leung
1785    Date: 2002/03/21 17:15:00 EST
1786    Tag: leunga-20020321-x86-fp-cfg
1787    Description:
1788    
1789      1. Recoded the buggy parts of x86-fp.
1790    
1791         a. All the block reordering code has been removed.
1792            We now depend on the block placement phases to do this work.
1793    
1794         b. Critical edge splitting code has been simplified and moved into the
1795            CFG modules, as where they belong.
1796    
1797         Both of these were quite buggy and complex.  The code is now much, much
1798         simpler.
1799    
1800      2. X86 backend.
1801    
1802         a. Added instructions for 64-bit support.  Instruction selection for
1803            64-bit has not been committed, however, since that
1804            requires changes to MLTREE which haven't been approved by
1805            Lal and John.
1806    
1807         b. Added support for FUCOMI and FUCOMIP when generating code for
1808            PentiumPro and above.  We only generate these instructions in
1809            the fast-fp mode.
1810    
1811         c. Added cases for JP and JNP in X86FreqProps.
1812    
1813      3. CFG
1814    
1815         CFG now has a bunch of methods for edge splitting and merging.
1816    
1817      4. Machine description.
1818    
1819         John's simplification of MLTREE_BASIS.fcond broke a few machine
1820         description things:
1821    
1822         rtl-build.{sig,sml} and hppa.mdl fixed.
1823    
1824         NOTE: the machine description stuff in the repository is still broken.
1825               Again, I can't put my fixes in because that involves
1826               changes to MLTREE.
1827    
1828    ----------------------------------------------------------------------
1829    Name: Matthias Blume
1830    Date: 2002/03/20 15:55:00 EST
1831    Tag: blume-20020320-kmp
1832    Description:
1833    
1834    Implemented Knuth-Morris-Pratt string matching in PreString and used
1835    it for String.isSubstring, Substring.isSubstring, and
1836    Substring.position.
1837    
1838    (Might need some stress-testing.  Simple examples worked fine.)
1839    
1840    ----------------------------------------------------------------------
1841    Name: Matthias Blume
1842    Date: 2002/03/19 16:37:00 EST
1843    Tag: blume-20020319-witnesses
1844    Description:
1845    
1846    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1847    
1848    This implements a generic mechanism for changing constness qualifiers
1849    anywhere within big C types without resorting to outright "casts".
1850    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1851    modify the constness at the outermost level.)
1852    The implementation of "convert" is based on the idea of "witness"
1853    values -- values that are not used by the operation but whose types
1854    "testify" to their applicability.  On the implementation side, "convert"
1855    is simply a projection (returning its second curried argument).  With
1856    cross-module inlining, it should not result in any machine code being
1857    generated.
1858    
1859    ----------------------------------------------------------------------
1860    Name: Matthias Blume
1861    Date: 2002/03/15 16:40:00 EST
1862    Tag: blume-20020315-basis
1863    Description:
1864    
1865    Provided (preliminary?) implementations for
1866    
1867      {String,Substring}.{concatWith,isSuffix,isSubstring}
1868    
1869    and
1870    
1871      Substring.full
1872    
1873    Those are in the Basis spec but they were missing in SML/NJ.
1874    
1875    ----------------------------------------------------------------------
1876    Name: Matthias Blume
1877    Date: 2002/03/14 21:30:00 EST
1878    Tag: blume-20020314-controls
1879    Description:
1880    
1881    Controls:
1882    ---------
1883    
1884    1. Factored out the recently-added Controls : CONTROLS stuff and put
1885       it into its own library $/controls-lib.cm.  The source tree for
1886       this is under src/smlnj-lib/Controls.
1887    
1888    2. Changed the names of types and functions in this interface, so they
1889       make a bit more "sense":
1890    
1891          module -> registry
1892          'a registry -> 'a group
1893    
1894    3. The interface now deals in ref cells only.  The getter/setter interface
1895       is (mostly) gone.
1896    
1897    4. Added a function that lets one register an already-existing ref cell.
1898    
1899    5. Made the corresponding modifications to the rest of the code so that
1900       everything compiles again.
1901    
1902    6. Changed the implementation of Controls.MLRISC back to something closer
1903       to the original.  In particular, this module (and therefore MLRISC)
1904       does not depend on Controls.  There now is some link-time code in
1905       int-sys.sml that registers the MLRISC controls with the Controls
1906       module.
1907    
1908    CM:
1909    ---
1910    
1911      * One can now specify the lambda-split aggressiveness in init.cmi.
1912    
1913    ----------------------------------------------------------------------
1914    Name: Allen Leung
1915    Date: 2002/03/13 17:30:00 EST
1916    Tag: leunga-20020313-x86-fp-unary
1917    Description:
1918    
1919    Bug fix for:
1920    
1921    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1922    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1923    > - fun f(x,(y,z)) = Real.~ y;
1924    > [autoloading]
1925    > [autoloading done]
1926    >       fchsl   (%eax), 184(%esp)
1927    > Error: MLRisc bug: X86MCEmitter.emitInstr
1928    >
1929    > uncaught exception Error
1930    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1931    
1932    The problem was that the code generator did not generate any fp registers
1933    in this case, and the ra didn't know that it needed to run the X86FP phase to
1934    translate the pseudo fp instruction.   This only happened with unary fp
1935    operators in certain situations.
1936    
1937    ----------------------------------------------------------------------
1938    Name: Matthias Blume
1939    Date: 2002/03/13 14:00:00 EST
1940    Tag: blume-20020313-overload-etc
1941    Description:
1942    
1943    1. Added _overload as a synonym for overload for backward compatibility.
1944       (Control.overloadKW must be true for either version to be accepted.)
1945    
1946    2. Fixed bug in install script that caused more things to be installed
1947       than what was requested in config/targets.
1948    
1949    3. Made CM aware of the (_)overload construct so that autoloading
1950       works.
1951    
1952    ----------------------------------------------------------------------
1953    Name: Matthias Blume
1954    Date: 2002/03/12 22:03:00 EST
1955    Tag: blume-20020312-url
1956    Description:
1957    
1958    Forgot to update BOOT and srcarchiveurl.
1959    
1960    ----------------------------------------------------------------------
1961    Name: Matthias Blume
1962    Date: 2002/03/12 17:30:00 EST
1963    Tag: blume-20020312-version110392
1964    Description:
1965    
1966    Yet another version number bump (because of small changes to the
1967    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1968    
1969    Changes:
1970    
1971      The new pid generation scheme described a few weeks ago was overly
1972      complicated.  I implemented a new mechanism that is simpler and
1973      provides a bit more "stability":  Once CM has seen a compilation
1974      unit, it keeps its identity constant (as long as you do not delete
1975      those crucial CM/GUID/* files).  This means that when you change
1976      an interface, compile, then go back to the old interface, and
1977      compile again, you arrive at the original pid.
1978    
1979      There now also is a mechanism that instructs CM to use the plain
1980      environment hash as a module's pid (effectively making its GUID
1981      the empty string).  For this, "noguid" must be specified as an
1982      option to the .sml file in question within its .cm file.
1983      This is most useful for code that is being generated by tools such
1984      as ml-nlffigen (because during development programmers tend to
1985      erase the tool's entire output directory tree including CM's cached
1986      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1987      revert to the old, broken behavior of SML/NJ, but in specific cases
1988      where there is no danger of interface confusion, its use is ok
1989      (I think).
1990    
1991      ml-nlffigen by default generates "noguid" annotations.  They can be
1992      turned off by specifying -guid in its command line.
1993    
1994    ----------------------------------------------------------------------
1995    Name: Lal George
1996    Date: 2002/03/12 12 14:42:36 EST
1997    Tag: george-20020312-frequency-computation
1998    Description:
1999    
2000    Integrated jump chaining and static block frequency into the
2001    compiler. More details and numbers later.
2002    
2003    ----------------------------------------------------------------------
2004    Name: Lal George
2005    Date: 2002/03/11 11 22:38:53 EST
2006    Tag: george-20020311-jump-chain-elim
2007    Description:
2008    
2009    Tested the jump chain elimination on all architectures (except the
2010    hppa).  This is on by default right now and is profitable for the
2011    alpha and x86, however, it may not be profitable for the sparc and ppc
2012    when compiling the compiler.
2013    
2014    The gc test will typically jump to a label at the end of the cluster,
2015    where there is another jump to an external cluster containing the actual
2016    code to invoke gc. This is to allow factoring of common gc invocation
2017    sequences. That is to say, we generate:
2018    
2019            f:
2020               testgc
2021               ja   L1      % jump if above to L1
2022    
2023            L1:
2024               jmp L2
2025    
2026    
2027    After jump chain elimination the 'ja L1' instructions is converted to
2028    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
2029    up being implemented in their long form (if L2 is far away) using:
2030    
2031            jbe     L3      % jump if below or equal to L3
2032            jmp     L2
2033         L3:
2034            ...
2035    
2036    
2037    For large compilation units L2  may be far away.
2038    
2039    
2040    ----------------------------------------------------------------------
2041    Name: Matthias Blume
2042    Date: 2002/03/11 13:30:00 EST
2043    Tag: blume-20020311-mltreeeval
2044    Description:
2045    
2046    A functor parameter was missing.
2047    
2048    ----------------------------------------------------------------------
2049    Name: Allen Leung
2050    Date: 2002/03/11 10:30:00 EST
2051    Tag: leunga-20020311-runtime-string0
2052    Description:
2053    
2054       The representation of the empty string now points to a
2055    legal null terminated C string instead of unit.  It is now possible
2056    to convert an ML string into C string with InlineT.CharVector.getData.
2057    This compiles into one single machine instruction.
2058    
2059    ----------------------------------------------------------------------
2060    Name: Allen Leung
2061    Date: 2002/03/10 23:55:00 EST
2062    Tag: leunga-20020310-x86-call
2063    Description:
2064    
2065       Added machine generation for CALL instruction (relative displacement mode)
2066    
2067    ----------------------------------------------------------------------
2068    Name: Matthias Blume
2069    Date: 2002/03/08 16:05:00
2070    Tag: blume-20020308-entrypoints
2071    Description:
2072    
2073    Version number bumped to 110.39.1.  NEW BOOTFILES!
2074    
2075    Entrypoints: non-zero offset into a code object where execution should begin.
2076    
2077    - Added the notion of an entrypoint to CodeObj.
2078    - Added reading/writing of entrypoint info to Binfile.
2079    - Made runtime system bootloader aware of entrypoints.
2080    - Use the address of the label of the first function given to mlriscGen
2081      as the entrypoint.  This address is currently always 0, but it will
2082      not be 0 once we turn on block placement.
2083    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
2084      with entry points) from mlriscGen.
2085    
2086    ----------------------------------------------------------------------
2087    Name: Allen Leung
2088    Date: 2002/03/07 20:45:00 EST
2089    Tag: leunga-20020307-x86-cmov
2090    Description:
2091    
2092       Bug fixes for CMOVcc on x86.
2093    
2094       1. Added machine code generation for CMOVcc
2095       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
2096       3. CMOVcc cannot have an immediate operand as argument.
2097    
2098    ----------------------------------------------------------------------
2099    Name: Matthias Blume
2100    Date: 2002/03/07 16:15:00 EST
2101    Tag: blume-20020307-controls
2102    Description:
2103    
2104    This is a very large but mostly boring patch which makes (almost)
2105    every tuneable compiler knob (i.e., pretty much everything under
2106    Control.* plus a few other things) configurable via both the command
2107    line and environment variables in the style CM did its configuration
2108    until now.
2109    
2110    Try starting sml with '-h' (or, if you are brave, '-H')
2111    
2112    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
2113    implements the underlying generic mechanism.
2114    
2115    The interface to some of the existing such facilities has changed somewhat.
2116    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
2117    (The getFoo interface is still there for backward-compatibility, but its
2118    use is deprecated.)
2119    
2120    The ml-build script passes -Cxxx=yyy command-line arguments through so
2121    that one can now twiddle the compiler settings when using this "batch"
2122    compiler.
2123    
2124    TODO items:
2125    
2126    We should go through and throw out all controls that are no longer
2127    connected to anything.  Moreover, we should go through and provide
2128    meaningful (and correct!) documentation strings for those controls
2129    that still are connected.
2130    
2131    Currently, multiple calls to Controls.new are accepted (only the first
2132    has any effect).  Eventually we should make sure that every control
2133    is being made (via Controls.new) exactly once.  Future access can then
2134    be done using Controls.acc.
2135    
2136    Finally, it would probably be a good idea to use the getter-setter
2137    interface to controls rather than ref cells.  For the time being, both
2138    styles are provided by the Controls module, but getter-setter pairs are
2139    better if thread-safety is of any concern because they can be wrapped.
2140    
2141    *****************************************
2142    
2143    One bug fix: The function blockPlacement in three of the MLRISC
2144    backpatch files used to be hard-wired to one of two possibilities at
2145    link time (according to the value of the placementFlag).  But (I
2146    think) it should rather sense the flag every time.
2147    
2148    *****************************************
2149    
2150    Other assorted changes (by other people who did not supply a HISTORY entry):
2151    
2152    1. the cross-module inliner now works much better (Monnier)
2153    2. representation of weights, frequencies, and probabilities in MLRISC
2154       changed in preparation of using those for weighted block placement
2155       (Reppy, George)
2156    
2157    ----------------------------------------------------------------------
2158    Name: Lal George
2159    Date: 2002/03/07 14:44:24 EST 2002
2160    Tag: george-20020307-weighted-block-placement
2161    
2162    Tested the weighted block placement optimization on all architectures
2163    (except the hppa) using AMPL to generate the block and edge frequencies.
2164    Changes were required in the machine properties to correctly
2165    categorize trap instructions. There is an MLRISC flag
2166    "weighted-block-placement" that can be used to enable weighted block
2167    placement, but this will be ineffective without block/edge
2168    frequencies (coming soon).
2169    
2170    
2171    ----------------------------------------------------------------------
2172    Name: Lal George
2173    Date: 2002/03/05 17:24:48 EST
2174    Tag: george-20020305-linkage-cluster
2175    
2176    In order to support the block placement optimization, a new cluster
2177    is generated as the very first cluster (called the linkage cluster).
2178    It contains a single jump to the 'real' entry point for the compilation
2179    unit. Block placement has no effect on the linkage cluster itself, but
2180    all the other clusters  have full freedom in the manner in which they
2181    reorder blocks or functions.
2182    
2183    On the x86 the typical linkage code that is generated is:
2184       ----------------------
2185            .align 2
2186       L0:
2187            addl    $L1-L0, 72(%esp)
2188            jmp     L1
2189    
2190    
2191            .align  2
2192       L1:
2193       ----------------------
2194    
2195    72(%esp) is the memory location for the stdlink register. This
2196    must contain the address of the CPS function being called. In the
2197    above example, it contains the address of  L0; before
2198    calling L1 (the real entry point for the compilation unit), it
2199    must contain the address for L1, and hence
2200    
2201            addl $L1-L0, 72(%esp)
2202    
2203    I have tested this on all architectures except the hppa.The increase
2204    in code size is of course negligible
2205    
2206    ----------------------------------------------------------------------
2207    Name: Allen Leung
2208    Date: 2002/03/03 13:20:00 EST
2209    Tag: leunga-20020303-mlrisc-tools
2210    
2211      Added #[ ... ] expressions to mlrisc tools
2212    
2213    ----------------------------------------------------------------------
2214    Name: Matthias Blume
2215    Date: 2002/02/27 12:29:00 EST
2216    Tag: blume-20020227-cdebug
2217    Description:
2218    
2219    - made types in structure C and C_Debug to be equal
2220    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
2221    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
2222    
2223    ----------------------------------------------------------------------
2224    Name: Matthias Blume
2225    Date: 2002/02/26 12:00:00 EST
2226    Tag: blume-20020226-ffi
2227    Description:
2228    
2229    1. Fixed a minor bug in CM's "noweb" tool:
2230       If numbering is turned off, then truly don't number (i.e., do not
2231       supply the -L option to noweb).  The previous behavior was to supply
2232       -L'' -- which caused noweb to use the "default" line numbering scheme.
2233       Thanks to Chris Richards for pointing this out (and supplying the fix).
2234    
2235    2. Once again, I reworked some aspects of the FFI:
2236    
2237       A. The incomplete/complete type business:
2238    
2239       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
2240         gone!
2241       - ML types representing an incomplete type are now *equal* to
2242         ML types representing their corresponding complete types (just like
2243         in C).  This is still safe because ml-nlffigen will not generate
2244         RTTI for incomplete types, nor will it generate functions that
2245         require access to such RTTI.   But when ML code generated from both
2246         incomplete and complete versions of the C type meet, the ML types
2247         are trivially interoperable.
2248    
2249         NOTE:  These changes restore the full generality of the translation
2250         (which was previously lost when I eliminated functorization)!
2251    
2252       B. Enum types:
2253    
2254       - Structure C now has a type constructor "enum" that is similar to
2255         how the "su" constructor works.  However, "enum" is not a phantom
2256         type because each "T enum" has values (and is isomorphic to
2257         MLRep.Signed.int).
2258       - There are generic access operations for enum objects (using
2259         MLRep.Signed.int).
2260       - ml-nlffigen will generate a structure E_foo for each "enum foo".
2261         * The structure contains the definition of type "mlrep" (the ML-side
2262         representation type of the enum).  Normally, mlrep is the same
2263         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
2264         then mlrep will be defined as a datatype -- thus facilitating
2265         pattern matching on mlrep values.
2266         ("-ec" will be suppressed if there are duplicate values in an
2267          enumeration.)
2268         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
2269         will be generated for each C enum constant xxx.
2270         * Conversion functions m2i and i2m convert between mlrep and
2271         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
2272         * Coversion functions c and ml convert between mlrep and "tag enum".
2273         * Access functions (get/set) fetch and store mlrep values.
2274       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
2275         enumerations are merged into one single enumeration represented by
2276         structure E_'.
2277    
2278    ----------------------------------------------------------------------
2279    Name: Allen Leung
2280    Date: 2002/02/25 04:45:00 EST
2281    Tag: leunga-20020225-cps-spill
2282    
2283    This is a new implementation of the CPS spill phase.
2284    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
2285    In case of problems, replace it with the old file spill.sml
2286    
2287    The current compiler runs into some serious performance problems when
2288    constructing a large record.  This can happen when we try to compile a
2289    structure with many items.  Even a very simple structure like the following
2290    makes the compiler slow down.
2291    
2292        structure Foo = struct
2293           val x_1 = 0w1 : Word32.int
2294           val x_2 = 0w2 : Word32.int
2295           val x_3 = 0w3 : Word32.int
2296           ...
2297           val x_N = 0wN : Word32.int
2298        end
2299    
2300    The following table shows the compile time, from N=1000 to N=4000,
2301    with the old compiler:
2302    
2303    N
2304    1000   CPS 100 spill                           0.04u  0.00s  0.00g
2305           MLRISC ra                               0.06u  0.00s  0.05g
2306              (spills = 0 reloads = 0)
2307           TOTAL                                   0.63u  0.07s  0.21g
2308    
2309    1100   CPS 100 spill                           8.25u  0.32s  0.64g
2310           MLRISC ra                               5.68u  0.59s  3.93g
2311              (spills = 0 reloads = 0)
2312           TOTAL                                   14.71u  0.99s  4.81g
2313    
2314    1500   CPS 100 spill                           58.55u  2.34s  1.74g
2315           MLRISC ra                               5.54u  0.65s  3.91g
2316              (spills = 543 reloads = 1082)
2317           TOTAL                                   65.40u  3.13s  6.00g
2318    
2319    2000   CPS 100 spill                           126.69u  4.84s  3.08g
2320           MLRISC ra                               0.80u  0.10s  0.55g
2321              (spills = 42 reloads = 84)
2322           TOTAL                                   129.42u  5.10s  4.13g
2323    
2324    3000   CPS 100 spill                           675.59u  19.03s  11.64g
2325           MLRISC ra                               2.69u  0.27s  1.38g
2326              (spills = 62 reloads = 124)
2327           TOTAL                                   682.48u  19.61s  13.99g
2328    
2329    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
2330           MLRISC ra                               4.96u  0.27s  2.72g
2331              (spills = 85 reloads = 170)
2332           TOTAL                                   2375.26u  57.21s  48.00g
2333    
2334    As you can see the old cps spill module suffers from some serious
2335    performance problem.  But since I cannot decipher the old code fully,
2336    instead of patching the problems up, I'm reimplementing it
2337    with a different algorithm.  The new code is more modular,
2338    smaller when compiled, and substantially faster
2339    (O(n log n) time and O(n) space).  Timing of the new spill module:
2340    
2341    4000  CPS 100 spill                           0.02u  0.00s  0.00g
2342          MLRISC ra                               0.25u  0.02s  0.15g
2343             (spills=1 reloads=3)
2344          TOTAL                                   7.74u  0.34s  1.62g
2345    
2346    Implementation details:
2347    
2348    As far as I can tell, the purpose of the CPS spill module is to make sure the
2349    number of live variables at any program point (the bandwidth)
2350    does not exceed a certain limit, which is determined by the
2351    size of the spill area.
2352    
2353    When the bandwidth is too large, we decrease the register pressure by
2354    packing live variables into spill records.  How we achieve this is
2355    completely different than what we did in the old code.
2356    
2357    First, there is something about the MLRiscGen code generator
2358    that we should be aware of:
2359    
2360    o MLRiscGen performs code motion!
2361    
2362       In particular, it will move floating point computations and
2363       address computations involving only the heap pointer to
2364       their use sites (if there is only a single use).
2365       What this means is that if we have a CPS record construction
2366       statement
2367    
2368           RECORD(k,vl,w,e)
2369    
2370       we should never count the new record address w as live if w
2371       has only one use (which is often the case).
2372    
2373       We should do something similar to floating point, but the transformation
2374       there is much more complex, so I won't deal with that.
2375    
2376    Secondly, there are now two new cps primops at our disposal:
2377    
2378     1. rawrecord of record_kind option
2379        This pure operator allocates some uninitialized storage from the heap.
2380        There are two forms:
2381    
2382         rawrecord NONE [INT n]  allocates a tagless record of length n
2383         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2384                                     and initializes the tag.
2385    
2386     2. rawupdate of cty
2387          rawupdate cty (v,i,x)
2388          Assigns to x to the ith component of record v.
2389          The storelist is not updated.
2390    
2391    We use these new primops for both spilling and increment record construction.
2392    
2393     1. Spilling.
2394    
2395        This is implemented with a linear scan algorithm (but generalized
2396        to trees).  The algorithm will create a single spill record at the
2397        beginning of the cps function and use rawupdate to spill to it,
2398        and SELECT or SELp to reload from it.  So both spills and reloads
2399        are fine-grain operations.  In contrast, in the old algorithm
2400        "spills" have to be bundled together in records.
2401    
2402        Ideally, we should sink the spill record construction to where
2403        it is needed.  We can even split the spill record into multiple ones
2404        at the places where they are needed.  But CPS is not a good
2405        representation for global code motion, so I'll keep it simple and
2406        am not attempting this.
2407    
2408     2. Incremental record construction (aka record splitting).
2409    
2410        Long records with many component values which are simulatenously live
2411        (recall that single use record addresses are not considered to
2412         be live) are constructed with rawrecord and rawupdate.
2413        We allocate space on the heap with rawrecord first, then gradually
2414        fill it in with rawupdate.  This is the technique suggested to me
2415        by Matthias.
2416    
2417        Some restrictions on when this is applicable:
2418        1. It is not a VECTOR record.  The code generator currently does not handle
2419           this case. VECTOR record uses double indirection like arrays.
2420        2. All the record component values are defined in the same "basic block"
2421           as the record constructor.  This is to prevent speculative
2422           record construction.
2423    
2424    ----------------------------------------------------------------------
2425    Name: Allen Leung
2426    Date: 2002/02/22 01:02:00 EST
2427    Tag: leunga-20020222-mlrisc-tools
2428    
2429    Minor bug fixes in the parser and rewriter
2430    
2431    ----------------------------------------------------------------------
2432    Name: Allen Leung
2433    Date: 2002/02/21 20:20:00 EST
2434    Tag: leunga-20020221-peephole
2435    
2436    Regenerated the peephole files.  Some contained typos in the specification
2437    and some didn't compile because of pretty printing bugs in the old version
2438    of 'nowhere'.
2439    
2440    ----------------------------------------------------------------------
2441    Name: Allen Leung
2442    Date: 2002/02/19 20:20:00 EST
2443    Tag: leunga-20020219-mlrisc-tools
2444    Description:
2445    
2446       Minor bug fixes to the mlrisc-tools library:
2447    
2448       1.  Fixed up parsing colon suffixed keywords
2449       2.  Added the ability to shut the error messages up
2450       3.  Reimplemented the pretty printer and fixed up/improved
2451           the pretty printing of handle and -> types.
2452       4.  Fixed up generation of literal symbols in the nowhere tool.
2453       5.  Added some SML keywords to to sml.sty
2454    
2455    ----------------------------------------------------------------------
2456    Name: Matthias Blume
2457    Date: 2002/02/19 16:20:00 EST
2458    Tag: blume-20020219-cmffi
2459    Description:
2460    
2461    A wild mix of changes, some minor, some major:
2462    
2463    * All C FFI-related libraries are now anchored under $c:
2464        $/c.cm      --> $c/c.cm
2465        $/c-int.cm  --> $c/internals/c-int.cm
2466        $/memory.cm --> $c/memory/memory.cm
2467    
2468    * "make" tool (in CM) now treats its argument pathname slightly
2469      differently:
2470        1. If the native expansion is an absolute name, then before invoking
2471           the "make" command on it, CM will apply OS.Path.mkRelative
2472           (with relativeTo = OS.FileSys.getDir()) to it.
2473        2. The argument will be passed through to subsequent phases of CM
2474           processing without "going native".  In particular, if the argument
2475           was an anchored path, then "make" will not lose track of that anchor.
2476    
2477    * Compiler backends now "know" their respective C calling conventions
2478      instead of having to be told about it by ml-nlffigen.  This relieves
2479      ml-nlffigen from one of its burdens.
2480    
2481    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
2482    
2483    * Export C_DEBUG and C_Debug from $c/c.cm.
2484    
2485    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
2486      subtyping relationship between incomplete pointers and their complete
2487      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
2488      with the convention of instantiating 'o with ('t, 'c) obj whenever
2489      the pointer target type is complete.  In the incomplete case, 'o
2490      will be instantiated with some "'c iobj" -- a type obtained by
2491      using one of the functors PointerToIncompleteType or PointerToCompleteType.
2492    
2493      Operations that work on both incomplete and complete pointer types are
2494      typed as taking an 'o ptr while operations that require the target to
2495      be known are typed as taking some ('t, 'c) obj ptr.
2496    
2497      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
2498      where void is an eqtype without any values.  This makes it possible
2499      to work on voidptr values using functions meant to operate on light
2500      incomplete pointers.
2501    
2502    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
2503      been vastly simplified.
2504    
2505    ----------------------------------------------------------------------
2506    Name: Matthias Blume
2507    Date: 2002/02/19 10:48:00 EST
2508    Tag: blume-20020219-pqfix
2509    Description:
2510    
2511    Applied Chris Okasaki's bug fix for priority queues.
2512    
2513    ----------------------------------------------------------------------
2514    Name: Matthias Blume
2515    Date: 2002/02/15 17:05:00
2516    Tag: Release_110_39
2517    Description:
2518    
2519    Last-minute retagging is becoming a tradition... :-(
2520    
2521    This is the working release 110.39.
2522    
2523    ----------------------------------------------------------------------
2524    Name: Matthias Blume
2525    Date: 2002/02/15 16:00:00 EST
2526    Tag: Release_110_39-orig
2527    Description:
2528    
2529    Working release 110.39.  New bootfiles.
2530    
2531    (Update: There was a small bug in the installer so it wouldn't work
2532    with all shells.  So I retagged. -Matthias)
2533    
2534    ----------------------------------------------------------------------
2535    Name: Matthias Blume
2536    Date: 2002/02/15 14:17:00 EST
2537    Tag: blume-20020215-showbindings
2538    Description:
2539    
2540    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
2541    the latter can be useful for exploring what bindings are available at
2542    the interactive prompt.  (The first function returns only the list
2543    of symbols that are really bound, the second prints those but also the
2544    ones that CM's autoloading mechanism knows about.)
2545    
2546    ----------------------------------------------------------------------
2547    Name: Matthias Blume
2548    Date: 2002/02/15 12:08:00 EST
2549    Tag: blume-20020215-iptrs
2550    Description:
2551    
2552    Two improvements to ml-nlffigen:
2553    
2554      1. Write files only if they do not exist or if their current contents
2555         do not coincide with what's being written.  (That is, avoid messing
2556         with the time stamps unless absolutely necessary.)
2557    
2558      2. Implement a "repository" mechanism for generated files related
2559         to "incomplete pointer types".   See the README file for details.
2560    
2561    ----------------------------------------------------------------------
2562    Name: Matthias Blume
2563    Date: 2002/02/14 11:50:00 EST
2564    Tag: blume-20020214-quote
2565    Description:
2566    
2567    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
2568    because of the new and improved tag generation scheme.  (Thanks to Allen
2569    Leung for pointing it out.)
2570    
2571    ----------------------------------------------------------------------
2572    Name: Lal George
2573    Date: 2002/02/14 09:55:27 EST 2002
2574    Tag: george-20020214-isabelle-bug
2575    Description:
2576    
2577    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2578    of Isabelle on the x86.
2579    
2580    From Allen:
2581    -----------
2582     I've found the problem:
2583    
2584         in ra-core.sml, I use the counter "blocked" to keep track of the
2585         true number of elements in the freeze queue.  When the counter goes
2586         to zero, I skip examining the queue.  But I've messed up the
2587         bookkeeping in combine():
2588    
2589             else ();
2590             case !ucol of
2591               PSEUDO => (if !cntv > 0 then
2592                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2593                                        ^^^^^^^^^^^^^^^^^^^^^^^
2594                      moveu := mergeMoveList(!movev, !moveu)
2595                     )
2596                  else ();
2597    
2598         combine() is called to coalesce two nodes u and v.
2599         I think I was thinking that if the move counts of u and v are both
2600         greater than zero then after they are coalesced then one node is
2601         removed from the freeze queue.  Apparently I was thinking that
2602         both u and v are of low degree, but that's clearly not necessarily true.
2603    
2604    
2605    02/12/2002:
2606        Here's the patch.  HOL now compiles.
2607    
2608        I don't know how this impact on performance (compile
2609        time or runtime).  This bug caused the RA (especially on the x86)
2610        to go thru the potential spill phase when there are still nodes on the
2611        freeze queue.
2612    
2613    
2614    
2615    
2616    ----------------------------------------------------------------------
2617    Name: Matthias Blume
2618    Date: 2002/02/13 22:40:00 EST
2619    Tag: blume-20020213-fptr-rtti
2620    Description:
2621    
2622    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2623    updates.
2624    
2625    ----------------------------------------------------------------------
2626    Name: Matthias Blume
2627    Date: 2002/02/13 16:41:00 EST
2628    Tag: blume-20020213-cmlpq
2629    Description:
2630    
2631    Added new priority queue export symbols (which have just been added to
2632    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2633    not compile and the installer would choke.)
2634    
2635    ----------------------------------------------------------------------
2636    Name: Matthias Blume
2637    Date: 2002/02/13 16:15:00 EST
2638    Tag: blume-20020213-various
2639    Description:
2640    
2641    1. More tweaks to ml-nlffigen:
2642    
2643       - better internal datastructures (resulting in slight speedup)
2644       - "-match" option requires exact match
2645       - "localized" gensym counters (untagged structs/unions nested within
2646         other structs/unions or within typedefs get a fresh counter; their
2647         tag will be prefixed by a concatenation of their parents' tags)
2648       - bug fixes (related to calculation of transitive closure of types
2649         to be included in the output)
2650    
2651    2. Minor Basis updates:
2652    
2653       - added implementations for List.collate and Option.app
2654    
2655    ----------------------------------------------------------------------
2656    Name: Matthias Blume
2657    Date: 2002/02/11 15:55:00 EST
2658    Tag: blume-20020211-gensym
2659    Description:
2660    
2661    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2662    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2663    names (ML structure names that correspond to unnamed C structs, unions,
2664    and enums), so that separate runs of ml-nlffigen do not clash.
2665    
2666    ----------------------------------------------------------------------
2667    Name: Matthias Blume
2668    Date: 2002/02/11 12:05:00 EST
2669    Tag: blume-20020211-gensml
2670    Description:
2671    
2672    A quick fix for a problem with GenSML (in the pgraph-util library):
2673    Make generation of toplevel "local" optional.  (Strictly speaking,
2674    signature definitions within "local" are not legal SML.)
2675    
2676    Other than that: updates to INSTALL and cm/TODO.
2677    
2678    ----------------------------------------------------------------------
2679    Name: Matthias Blume
2680    Date: 2002/02/08 15:00:00 EST
2681    Tag: blume-20020208-uniquepid
2682    Description:
2683    
2684    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2685    
2686    1. The installer (config/install.sh) has gotten smarter:
2687    
2688         - Configuration options are a bit easier to specify now
2689           (in config/targets).
2690         - Bug in recognizing .tar.bz2 files fixed.
2691         - Installer automatically resolves dependencies between
2692           configuration options (e.g., if you ask for eXene, you will
2693           also get cml -- regardless whether you asked for it or not).
2694         - Installer can run in "quieter mode" by setting the environment
2695           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2696           "completely silent", though.
2697         - Build HashCons library as part of smlnj-lib.
2698    
2699    2. A new scheme for assigning persistent identifiers to compilation
2700       units (and, by extension, to types etc.) has been put into place.
2701       This fixes a long-standing bug where types and even dynamic values
2702       can get internally confused, thereby compromising type safety
2703       (abstraction) and dynamic correctness.  See
2704    
2705         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2706    
2707       for an example of how things could go wrong until now.
2708    
2709       The downside of the new scheme is that pids are not quite as
2710       persistent as they used to be: CM will generate a fresh pid
2711       for every compilation unit that it thinks it sees for the first
2712       time.  That means that if you compile starting from a clean, fresh
2713       source tree at two different times, you end up with different
2714       binaries.
2715    
2716       Cutoff recompilation, however, has not been compromised because
2717       CM keeps pid information in special caches between runs.
2718    
2719    ----------------------------------------------------------------------
2720    Name: Lal George
2721    Date: 2002/02/07 15:34:13 EST 2002
2722    Tag: <none>
2723    Description:
2724    
2725    Compilers that generate assembly code may produce  global labels
2726    whose value is resolved at link time. The various peephole optimization
2727    modules did not take this in account.
2728    
2729    TODO. The Labels.addrOf function should really return an option
2730    type so that clients are forced to deal with this issue, rather
2731    than an exception being raised.
2732    
2733    ----------------------------------------------------------------------
2734    Name: Lal George
2735    Date: 2002/02/06 13:55:02 EST
2736    Tag: george-20020206-ra-breakup
2737    Description:
2738    
2739    1. A bug fix from Allen.
2740    
2741        A typo causes extra fstp %st(0)'s to be generated at compensation
2742        edges, which might cause stack underflow traps at runtime.  This
2743        occurs in fft where there are extraneous fstps right before the 'into'
2744        trap instruction (in this case they are harmless since none of the
2745        integers overflow.)
2746    
2747    2. Pulled out various utility modules that were embedded in the modules
2748       of the register allocator. I need these modules for other purposes, but
2749       they are not complete enough to put into a library (just yet).
2750    ----------------------------------------------------------------------
2751    Name: Matthias Blume
2752    Date: 2002/01/31 16:05:00 EST
2753    Tag: blume-20020131-sparc-ccalls
2754    Description:
2755    
2756    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2757       of extra stack space by mistake.  Fixed.
2758    
2759    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2760    
2761    ----------------------------------------------------------------------
2762    Name: Allen Leung
2763    Date: 2002/01/30
2764    Tag: leunga-20020130-nowhere-bug-fix
2765    Description:
2766    
2767       MLRISC bug fixes:
2768       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2769       2. MachineInt.fromString was negating its value.
2770    
2771    ----------------------------------------------------------------------
2772    Name: Matthias Blume
2773    Date: 2002/01/29
2774    Tag: blume-20020129-INSTALL
2775    Description:
2776    
2777    - Added somewhat detailed installation instructions (file INSTALL).
2778    - Fixed curl-detection bug in config/install.sh.
2779    - It is now possible to select the URL getter using the URLGETTER
2780      environment variable:
2781    
2782          not set / "unknown"      --> automatic detection (script tries wget,
2783                                       curl, and lynx)
2784          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2785                                       how to properly invoke them)
2786          other                    --> use $URLGETTER directly, it must take
2787                                       precisely two command-line arguments
2788                                       (source URL and destination file name)
2789    
2790    ----------------------------------------------------------------------
2791    Name: Matthias Blume
2792    Date: 2002/01/28
2793    Tag: blume-20020128-sparc-ccalls
2794    Description:
2795    
2796    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2797    - Make use of the allocParam argument in sparc-c-calls.
2798    
2799    ----------------------------------------------------------------------
2800    Name: Matthias Blume
2801    Date: 2002/01/28
2802    Tag: blume-20020128-allocParam
2803    Description:
2804    
2805    John Reppy:  Changes c-calls API to accept client-callback for
2806    allocating extra stack space.
2807    me: Corresponding changes to mlriscGen (using a dummy argument that
2808        does not change the current behavior).
2809    
2810    ----------------------------------------------------------------------
2811    Name: Matthias Blume
2812    Date: 2002/01/28 12:00:00
2813    Tag: Release_110_38
2814    Description:
2815    
2816    This time for real!!!
2817    
2818    ----------------------------------------------------------------------
2819    Name: Matthias Blume
2820    Date: 2002/01/28 10:56:00 EST
2821    Tag: blume-20020128-retraction
2822    Description:
2823    
2824    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
2825       with blume-Release_110_38-retracted.)
2826    
2827    1. Fixed a problem with incorrect rounding modes in real64.sml.
2828       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2829    
2830    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2831       and enums fixed.  The naming of corresponding ML identifiers should
2832       now be consistent again.
2833    
2834    ----------------------------------------------------------------------
2835    Name: Allen Leung
2836    Date: 2002/01/27
2837    Tag: leunga-20020127-nowhere
2838    Description:
2839    
2840       Added a target called nowhere in the configuration scripts.
2841       Enabling this will build the MLRISC 'nowhere' tool (for translating
2842       programs with where-clauses into legal SML code) during installation.
2843    
2844    ----------------------------------------------------------------------
2845    Name: Matthias Blume
2846    Date: 2002/01/25 21:27:00 EST
2847    Tag: blume-Release_110_38-retracted
2848    Description:
2849    
2850    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
2851    
2852    README will be added later.
2853    
2854    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
2855    (Release_110_38) removed.  Reason: Last-minute bug fixes.
2856    
2857    ----------------------------------------------------------------------
2858    Name: Matthias Blume
2859    Date: 2002/01/25
2860    Tag: blume-20020125-ffi
2861    Description:
2862    
2863    A large number of tweaks and improvements to ml-nlffi-lib and
2864    ml-nlffigen:
2865    
2866       - ML represenation types have been streamlined
2867       - getter and setter functions work with concrete values, not abstract
2868         ones where possible
2869       - ml-nlffigen command line more flexible (see README file there)
2870       - some bugs have been fixed (hopefully)
2871    
2872    ----------------------------------------------------------------------
2873    Name: Lal George
2874    Date: 2002/01/24
2875    Tag: george-20020124-risc-ra-interface
2876    Description:
2877    
2878       There is a dramatic simplification in the interface to the
2879       register allocator for RISC architectures as a result of making
2880       parallel copy instructions explicit.
2881    
2882    ----------------------------------------------------------------------
2883    Name: Matthias Blume
2884    Date: 2002/01/22
2885    Tag: blume-20020122-x86-ccalls
2886    Description:
2887    
2888    Bug fix for c-calls on x86 (having to do with how char- and
2889    short-arguments are being handled).
2890    
2891    ----------------------------------------------------------------------
2892    Name: Matthias Blume
2893    Date: 2002/01/21
2894    Tag: blume-20020121-ff
2895    Description:
2896    
2897    Another day of fiddling with the FFI...
2898    
2899    1. Bug fix/workaround:  CKIT does not complain about negative array
2900       dimensions, so ml-nlffigen has to guard itself against this possibility.
2901       (Otherwise a negative dimension would send it into an infinite loop.)
2902    
2903    2. Some of the abstract types (light objects, light pointers, most "base"
2904       types) in structure C are now eqtypes.
2905    
2906    3. Added constructors and test functions for NULL function pointers.
2907    
2908    ----------------------------------------------------------------------
2909    Name: Matthias Blume
2910    Date: 2002/01/18
2911    Tag: blume-20020118-ready-for-new-release
2912    Description:
2913    
2914    Made config/srcarchiveurl point to a new place.  (Will provide boot
2915    files shortly.)
2916    
2917    Maybe we christen this to be 110.38?
2918    
2919    ----------------------------------------------------------------------
2920    Name: Matthias Blume
2921    Date: 2002/01/18
2922    Tag: blume-20020118-more-ffifiddle
2923    Description:
2924    
2925    Today's FFI fiddling:
2926    
2927      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
2928        C.Set.* that use concrete (MLRep.*) arguments and results instead
2929        of abstract ones.
2930    
2931      - Provided word-style bit operations etc. for "int" representation
2932        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
2933        and Long.
2934    
2935    ----------------------------------------------------------------------
2936    Name: Matthias Blume
2937    Date: 2002/01/18
2938    Tag: blume-20020118-use-x86-fp
2939    Description:
2940    
2941    Now that x86-fast-fp seems to be working, I turned it back on again
2942    by default.  (Seems to work fine now, even with the FFI.)
2943    
2944    Other than that, I added some documentation about the FFI to
2945    src/ml-nlffigen/README and updated the FFI test examples in
2946    src/ml-nlffi-lib/Tests/*.
2947    
2948    ----------------------------------------------------------------------
2949    Name: Allen Leung
2950    Date: 2002/01/17
2951    Tag: leunga-20020117-x86-fast-fp-call
2952    Description:
2953    
2954       1. Fixed a problem with handling return fp values when x86's fast fp
2955          mode is turned on.
2956    
2957       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
2958          of %f32.
2959    
2960       3. Added a constructor INT32lit to the ast of MLRISC tools.
2961    
2962    ----------------------------------------------------------------------
2963    Name: Matthias Blume
2964    Date: 2002/01/16
2965    Tag: blume-20020116-ffifiddle
2966    Description:
2967    
2968    More fiddling with the FFI interface:
2969    
2970     - Make constness 'c instead of rw wherever possible.  This eliminates
2971       the need for certain explicit coercions.  (However, due to ML's
2972       value polymorphism, there will still be many cases where explicit
2973       coercions are necessary.  Phantom types are not the whole answer
2974       to modeling a subtyping relationship in ML.)
2975    
2976     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
2977    
2978     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
2979       clumsy workaround for varargs problem.)
2980    
2981    ----------------------------------------------------------------------
2982    Name: Lal George
2983    Date: 2002/01/15
2984    Tag: <none>
2985    Description:
2986    
2987    1. Since COPY instructions are no longer native to the architecture,
2988       a generic functor can be used to implement the expandCopies function.
2989    
2990    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
2991       TEXT segment.
2992    
2993    ----------------------------------------------------------------------
2994    Name: Matthias Blume
2995    Date: 2002/01/15
2996    Tag: blume-20020115-ffiupdates
2997    Description:
2998    
2999    1. Fix for bug resulting in single-precision float values being returned
3000       incorrectly from FFI calls.
3001    
3002    2. Small modifications to C FFI API:
3003    
3004        - memory-allocation routines return straight objects (no options)
3005          and raise an exception in out-of-memory situations
3006        - unsafe extensions to cast between function pointers and pointers
3007          from/to ints
3008        - added structure C_Debug as an alternative to structure C where
3009          pointer-dereferencing (|*| and |*!) always check for null-pointers
3010        - added open_lib' to DynLinkage;  open_lib' works like open_lib
3011          but also takes a (possibly empty) list of existing library handles
3012          that the current library depends on
3013    
3014    ----------------------------------------------------------------------
3015    Name: Matthias Blume
3016    Date: 2002/01/10
3017    Tag: blume-20020110-newffigen
3018    Description:
3019    
3020    1. Updates to portable graph code.
3021    
3022    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
3023       more scalable now so that even huge interfaces such as the one
3024       for GTK compile in finite time and space. :-)
3025       See src/ml-nlffigen/README for details on what's new.
3026    
3027    ----------------------------------------------------------------------
3028    Name: Lal George
3029    Date: 2001/01/09 14:31:35 EST 2002
3030    Tag: george-20011206-rm-native-copy
3031    Description:
3032    
3033            Removed the native COPY and FCOPY instructions
3034            from all the architectures and replaced it with the
3035            explicit COPY instruction from the previous commit.
3036    
3037            It is now possible to simplify many of the optimizations
3038            modules that manipulate copies. This has not been
3039            done in this change.
3040    
3041    ----------------------------------------------------------------------
3042    Name: Lal George
3043    Date: 2001/12/06 16:50:13 EST 2001
3044    Tag: george-20011206-mlrisc-instruction
3045    Description:
3046    
3047    Changed the representation of instructions from being fully abstract
3048    to being partially concrete. That is to say:
3049    
3050      from
3051            type instruction
3052    
3053      to
3054            type instr                              (* machine instruction *)
3055    
3056            datatype instruction =
3057                LIVE of {regs: C.cellset, spilled: C.cellset}
3058              | KILL of {regs: C.cellset, spilled: C.cellset}
3059              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
3060              | ANNOTATION of {i: instruction, a: Annotations.annotation}
3061              | INSTR of instr
3062    
3063    This makes the handling of certain special instructions that appear on
3064    all architectures easier and uniform.
3065    
3066    LIVE and KILL say that a list of registers are live or killed at the
3067    program point where they appear. No spill code is generated when an
3068    element of the 'regs' field is spilled, but the register is moved to
3069    the 'spilled' (which is present, more for debugging than anything else).
3070    
3071    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
3072    We used to generate:
3073    
3074            DEFFREG f1
3075            f1 := f2 + f3
3076            trapb
3077    
3078    but now generate:
3079    
3080            f1 := f2 + f3
3081            trapb
3082            LIVE {regs=[f1,f2,f3], spilled=[]}
3083    
3084    Furthermore, the DEFFREG (hack) required that all floating point instruction
3085    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
3086    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
3087    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
3088    this architecture).
3089    
3090    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
3091    all the architectures. This will result in further simplification of the
3092    register allocator that must be aware of them for coalescing purposes, and
3093    will also simplify certain aspects of the machine description that provides
3094    callbacks related to parallel copies.
3095    
3096    ANNOTATION should be obvious, and now INSTR represents the honest to God
3097    machine instruction set!
3098    
3099    The <arch>/instructions/<arch>Instr.sml files define certain utility
3100    functions for making porting easier -- essentially converting upper case
3101    to lower case. All machine instructions (of type instr) are in upper case,
3102    and the lower case form generates an MLRISC instruction. For example on
3103    the alpha we have:
3104    
3105      datatype instr =
3106         LDA of {r:cell, b:cell, d:operand}
3107       | ...
3108    
3109      val lda : {r:cell, b:cell, d:operand} -> instruction
3110        ...
3111    
3112    where lda is just (INSTR o LDA), etc.
3113    
3114    ----------------------------------------------------------------------
3115    Name: Matthias Blume
3116    Date: 2001/11/22 21:40:00 EST
3117    Tag: Release_110_37
3118    Description:
3119    
3120    Release 110.37.  This time for real.
3121    
3122    ----------------------------------------------------------------------
3123    Name: Matthias Blume
3124    Date: 2001/11/21 16:35:00 EST
3125    Tag: blume-20011121-foot-in-mouth
3126    Description:
3127    
3128    Removed the "Release_110_37" tag because of a serious bug.
3129    This will be re-tagged once the bug is fixed.
3130    
3131    ----------------------------------------------------------------------
3132    Name: Matthias Blume
3133    Date: 2001/11/21 16:14:00 EST
3134    Tag: blume-20011121-forgottenfile
3135    Description:
3136    
3137    Forgot to add a file.  (Just a .tex-file -- part of
3138    the CM manual source.)
3139    
3140    ----------------------------------------------------------------------
3141    Name: Matthias Blume
3142    Date: 2001/11/21 16:10:00 EST
3143    Tag: blume-20011121-invalid_110_37
3144    Description:
3145    
3146    Note: I removed the original tag "Release_110_37" from this commit
3147          because we found a serious bug in all non-x86 backends.
3148          - Matthias
3149    
3150    1. Modifications to the SML/NJ code generator and to the runtime system
3151       so that code object name strings are directly inserted into code
3152       objects at code generation time.  The only business the runtime system
3153       has with this is now to read the name strings on occasions.
3154       (The encoding of the name string has also changed somewhat.)
3155    
3156    2. CM now implements a simple "set calculus" for specifying export lists.
3157       In particular, it is now possible to refer to the export lists of
3158       other libraries/groups/sources and form unions as well as differences.
3159       See the latest CM manual for details.
3160    
3161    3. An separate notion of "proxy" libraries has again be eliminated from
3162       CM's model.  (Proxy libraries are now simply a special case of using
3163       the export list calculus.)
3164    
3165    4. Some of the existing libraries now take advantage of the new set
3166       calculus.
3167       (Notice that not all libraries have been converted because some
3168       of the existing .cm-files are supposed to be backward compatible
3169       with 110.0.x.)
3170    
3171    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
3172       "General.exnMessage"!  The former relies on a certain hook to be
3173       initialized, and that often does not happen in the stand-alone case.)
3174    
3175    ----------------------------------------------------------------------
3176    Name: Lal George
3177    Date: 2001/11/21  13:56:18 EST
3178    Tag: george-2001121-pseudo-ops
3179    Description:
3180    
3181      Implemented a complete redesign of MLRISC pseudo-ops. Now there
3182      ought to never be any question of incompatabilities with
3183      pseudo-op syntax expected by host assemblers.
3184    
3185      For now, only modules supporting GAS syntax are implemented
3186      but more should follow, such as MASM, and vendor assembler
3187      syntax, e.g. IBM as, Sun as, etc.
3188    
3189    ----------------------------------------------------------------------
3190    Name: Matthias Blume
3191    Date: 2001/11/14 11:52:00 EST
3192    Tag: blume-20011114-srcname
3193    Description:
3194    
3195    1. Routed the name of the current source file to mlriscgen where it
3196       should be directly emitted into the code object.  (This last part
3197       is yet to be done.)
3198    
3199    2. Some cleanup of the pgraph code to make it match the proposal that
3200       I put out the other day.  (The proposal notwithstanding, things are
3201       still in flux here.)
3202    
3203    ----------------------------------------------------------------------
3204    Name: Lal George
3205    Date: 2001/11/14 09:44:04 EST
3206    Tag:
3207    Description:
3208    
3209      Fix for a backpatching bug reported by Allen.
3210    
3211      Because the boundary between short and long span-dependent
3212      instructions is +/- 128, there are an astounding number of
3213      span-dependent instructions whose size is over estimated.
3214    
3215      Allen came up with the idea of letting the size of span
3216      dependent instructions be non-monotonic, for a maxIter
3217      number of times, after which the size must be monotonically
3218      increasing.
3219    
3220      This table shows the number of span-dependent instructions
3221      whose size was over-estimated as a function of maxIter, for the
3222      file Parse/parse/ml.grm.sml:
3223    
3224         maxIter            # of instructions:
3225            10                      687
3226            20                      438
3227            30                      198
3228            40                        0
3229    
3230      In compiling the compiler, there is no significant difference in
3231      compilation speed between maxIter=10 and maxIter=40. Actually,
3232      my measurements showed that maxIter=40 was a tad faster than
3233      maxIter=10! Also 96% of the  files in the compiler reach a fix
3234      point within 13 iterations, so fixing maxIter at 40, while high,
3235      is okay.
3236    
3237    ----------------------------------------------------------------------
3238    Name: Matthias Blume
3239    Date: 2001/10/31 15:25:00 EST
3240    Tag: blume-20011031-pgraph
3241    Description:
3242    
3243    CKIT:
3244    * Changed the "Function" constructor of type Ast.ctype to carry optional
3245      argument identifiers.
3246    * Changed the return type of TypeUtil.getFunction accordingly.
3247    * Type equality ignores the argument names.
3248    * TypeUtil.composite tries to preserve argument names but gives up quickly
3249      if there is a mismatch.
3250    
3251    installation script:
3252    * attempts to use "curl" if available (unless "wget" is available as well)
3253    
3254    CM:
3255    * has an experimental implementation of "portable graphs" which I will
3256      soon propose as an implementation-independent library format
3257    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
3258    
3259    NLFFI-LIB:
3260    * some cleanup (all cosmetic)
3261    
3262    NLFFIGEN:
3263    * temporarily disabled the mechanism that suppresses ML output for
3264      C definitions whose identifiers start with an underscore character
3265    * generate val bindings for enum constants
3266    * user can request that only one style (light or heavy) is being used;
3267      default is to use both (command-line arguments: -heavy and -light)
3268    * fixed bug in handling of function types involving incomplete pointers
3269    * generate ML entry points that take record arguments (i.e., using
3270      named arguments) for C functions that have a prototype with named
3271      arguments
3272      (see changes to CKIT)
3273    
3274    ----------------------------------------------------------------------
3275    Name: Allen Leung
3276    Date: 2001/10/27 20:34:00 EDT
3277    Tag: leunga-20011027-x86-fast-fp-call
3278    Description:
3279    
3280       Fixed the bug described in blume-20010920-slowfp.
3281    
3282       The fix involves
3283          1. generating FCOPYs in FSTP in ia32-svid
3284          2. marking a CALL with the appropriate annotation
3285    
3286    ----------------------------------------------------------------------
3287    Name: Matthias Blume
3288    Date: 2001/10/16 11:32:00 EDT
3289    Tag: blume-20011016-netbsd
3290    Description:
3291    
3292    Underscore patch from Chris Richards (fixing problem with compiling
3293    runtime system under recent NetBSD).
3294    
3295    ----------------------------------------------------------------------
3296    Name: Allen Leung
3297    Date: 2001/10/12 17:18:32 EDT 2001
3298    Tag: leung-20011012-x86-printflowgraph
3299    Description:
3300    
3301    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
3302    
3303    ----------------------------------------------------------------------
3304    Name: Lal George
3305    Date: 2001/10/11 23:51:34 EDT
3306    Tag: george-20011011-too-many-instrs
3307    Description:
3308    
3309    The representation of a program point never expected to see more
3310    than 65536 instructions in a basic block!
3311    
3312    ----------------------------------------------------------------------
3313    Name: Lal George
3314    Date: 2001/10/09 09:41:37 EDT
3315    Tag: george-20011008-mlrisc-labels
3316    Description:
3317    
3318    Changed the machine description files to support printing of
3319    local and global labels in assembly code, based on host assembler
3320    conventions.
3321    
3322    ----------------------------------------------------------------------
3323    Name: Matthias Blume
3324    Date: 2001/09/25 15:25:00 EDT
3325    Tag: blume-20010925-exninfo
3326    Description:
3327    
3328    I provided a non-hook implementation of exnName (at the toplevel) and
3329    made the "dummy" implementation of exnMessage (at the toplevel) more
3330    useful: if nothing gets "hooked in", then at least you are going to
3331    see the exception name and a message indicating why you don't see more.
3332    
3333    [For the time being, programs that need exnMessage and want to use
3334    ml-build should either use General.exnMessage (strongly recommended) or
3335    refer to structure General at some other point so that CM sees a
3336    static dependency.]
3337    
3338    [Similar remarks go for "print" and "use":  If you want to use their
3339    functionality in stand-alone programs generated by ml-build, then use
3340    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
3341    
3342    ----------------------------------------------------------------------
3343    Name: Matthias Blume
3344    Date: 2001/09/20 17:28:00 EDT
3345    Tag: blume-20010920-slowfp
3346    Description:
3347    
3348    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
3349    
3350    ----------------------------------------------------------------------
3351    Name: Matthias Blume
3352    Date: 2001/09/20 17:20:00 EDT
3353    Tag: blume-20010920-canonicalpaths
3354    Description:
3355    
3356    0. Updated the BOOT file (something that I forgot to do earlier).
3357    
3358    1. Small internal change to CM so that it avoids "/../" in filenames
3359       as much as possible (but only where it is safe).
3360    
3361    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
3362       that arguments that contain delimiters are passed through correctly.
3363       This change also means that all "special" arguments of the form
3364       @SMLxxx... must come first.
3365    
3366    3. Changed install script to put relative anchor names for tool commands
3367       into pathconfig.
3368    
3369    ----------------------------------------------------------------------
3370    Name: Matthias Blume
3371    Date: 2001/09/18 15:35:00 EDT
3372    Tag: blume-20010918-readme11036
3373    Description:
3374    
3375    Added README files.
3376    
3377    ----------------------------------------------------------------------
3378    Name: Matthias Blume
3379    Date: 2001/09/18 11:45:00 EDT
3380    Tag: Release_110_36 (retag)
3381    Description:
3382    
3383    Fixed mistake in config/preloads. Retagged as 110.36.
3384    
3385    ----------------------------------------------------------------------
3386    Name: Matthias Blume
3387    Date: 2001/09/18 09:40:00 EDT
3388    Tag: Release_110_36_orig (tag changed)
3389    Description:
3390    
3391    New version (110.36).  New bootfiles.
3392    
3393    ----------------------------------------------------------------------
3394    Name: Matthias Blume
3395    Date: 2001/09/14 16:15:00 EDT
3396    Tag: blume-20010914-x86fastfp
3397    Description:
3398    
3399    John committed some changes that Allen made, in particular a (hopefully)
3400    correctly working version of the x86-fp module.
3401    
3402    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
3403    flag to "true".  Everything seems to compile to a fixpoint ok, and
3404    "mandelbrot" speeds up by about 15%.
3405    
3406    ----------------------------------------------------------------------
3407    Name: Matthias Blume
3408    Date: 2001/09/13 11:20:00 EDT
3409    Tag: blume-20010913-minimal
3410    Description:
3411    
3412    1. Stefan Monnier's patch to fix a miscompilation problem that
3413       was brought to light by John Reppy's work on Moby.
3414    
3415    2. Implemented a minimal "structure Compiler" that contains just
3416       "version" and "architecture".  The minimal version will be
3417       available when the full version is not.  This is for backward-
3418       compatibility with code that wants to test Compiler.version.
3419    
3420    ----------------------------------------------------------------------
3421    Name: Matthias Blume
3422    Date: 2001/08/28 14:03:00 EDT
3423    Tag: blume-20010828-ml-lex
3424    Description:
3425    
3426    Fix for bug 1581, received from Neophytos Michael.
3427    
3428    ----------------------------------------------------------------------
3429    Name: Matthias Blume
3430    Date: 2001/08/27 11:20:00 EDT
3431    Tag: blume-20010827-readme11035
3432    Description:
3433    
3434    Fleshed out the README file for 110.35.
3435    
3436    ----------------------------------------------------------------------
3437    Name: Matthias Blume
3438    Date: 2001/08/24 17:10:00 EDT
3439    Tag: Release_110_35
3440    Description:
3441    
3442    New version number (110.35).  New bootfiles.
3443    
3444    ----------------------------------------------------------------------
3445    Name: Lal George
3446    Date: 2001/08/24 13:47:18 EDT 2001
3447    Tag: george-20010824-MLRISC-graphs
3448    Description:
3449    
3450     removed clusters from MLRISC completely and replaced with graphs.
3451    
3452    ----------------------------------------------------------------------
3453    Name: Matthias Blume
3454    Date: 2001/08/23 17:50:00 EDT
3455    Tag: blume-20010823-toplevel
3456    Description:
3457    
3458    - some reorganization of the code that implements various kinds of
3459      environments in the compiler (static, dynamic, symbolic, combined)
3460    - re-implemented the EnvRef module so that evalStream works properly
3461      (if the stream contains references to "use", "CM.make", etc.)
3462    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
3463    
3464    ----------------------------------------------------------------------
3465    Name: Matthias Blume
3466    Date: 2001/08/20 15:50 EDT
3467    Tag: blume20010820-slipup
3468    Description:
3469    
3470    I forgot to commit a few files.  Here they are...
3471    
3472    ----------------------------------------------------------------------
3473    Name: Matthias Blume
3474    Date: 2001/08/20 15:35:00 EDT
3475    Tag: blume-20010820-debugprof
3476    Description:
3477    
3478    !!!! NEW BOOTFILES !!!!
3479    
3480    This is another round of reorganizing the compiler sources.  This
3481    time the main goal was to factor out all the "instrumentation"
3482    passes (for profiling and backtracing) into their own library.
3483    The difficulty was to do it in such a way that it does not depend
3484    on elaborate.cm but only on elabdata.cm.
3485    
3486    Therefore there have been further changes to both elaborate.cm and
3487    elabdata.cm -- more "generic" things have been moved from the former
3488    to the latter.  As a result, I was forced to split the assignment
3489    of numbers indicating "primtyc"s into two portions: SML-generic and
3490    SML/NJ-specific.  Since it would have been awkward to maintain,
3491    I bit the bullet and actually _changed_ the mapping between these
3492    numbers and primtycs.  The bottom line of this is that you need
3493    a new set of bin- and bootfiles.
3494    
3495    I have built new bootfiles for all architectures, so doing a fresh
3496    checkout and config/install.sh should be all you need.
3497    
3498    The newly created library's name is
3499    
3500        $smlnj/viscomp/debugprof.cm
3501    
3502    and its sources live under
3503    
3504        src/compiler/DebugProf
3505    
3506    ----------------------------------------------------------------------
3507    Name: Matthias Blume
3508    Date: 2001/08/15 17:15:00 EDT
3509    Tag: blume-20010815-compreorg
3510    Description:
3511    
3512    This is a first cut at reorganizing the CM libraries that make up the
3513    core of the compiler.  The idea is to separate out pieces that could
3514    be used independently by tools, e.g., the parser, the typechecker, etc.
3515    
3516    The current status is a step in this direction, but it is not quite
3517    satisfactory yet.  Expect more changes in the future.
3518    
3519    Here is the current (new) organization...
3520    
3521        What used to be $smlnj/viscomp/core.cm is now divided into
3522        six CM libraries:
3523    
3524             $smlnj/viscomp/basics.cm
3525                           /parser.cm
3526                           /elabdata.cm
3527                           /elaborate.cm
3528                           /execute.cm
3529                           /core.cm
3530    
3531        The CM files for these libraries live under src/system/smlnj/viscomp.
3532        All these libraries are proxy libraries that contain precisely
3533        one CM library component.  Here are the locations of the components
3534        (all within the src/compiler tree):
3535    
3536             Basics/basics.cm
3537             Parse/parser.cm
3538             ElabData/elabdata.cm
3539             Elaborator/elaborate.cm
3540             Execution/execute.cm
3541             core.cm
3542    
3543         [This organization is the same that has been used already
3544         for a while for the architecture-specific parts of the visible
3545         compiler and for the old version of core.cm.]
3546    
3547         As you will notice, many source files have been moved from their
3548         respective original locations to a new home in one of the above
3549         subtrees.
3550    
3551         The division of labor between the new libraries is the following:
3552    
3553             basics.cm:
3554                - Simple, basic definitions that pertain to many (or all) of
3555                  the other libraries.
3556             parser.cm:
3557                - The SML parser, producing output of type Ast.dec.
3558                - The type family for Ast is also defined and exported here.
3559             elabdata.cm:
3560                - The datatypes that describe input and output of the elaborator.
3561                  This includes types, absyn, and static environments.
3562             elaborator.cm:
3563                - The SML/NJ type checker and elaborator.
3564                  This maps an Ast.dec (with a given static environment) to
3565                  an Absyn.dec (with a new static environment).
3566                - This libraries implements certain modules that used to be
3567                  structures as functors (to remove dependencies on FLINT).
3568             execute.cm:
3569                - Everything having to do with executing binary code objects.
3570                - Dynamic environments.
3571             core.cm:
3572                - SML/NJ-specific instantiations of the elaborator and MLRISC.
3573                - Top-level modules.
3574                - FLINT (this should eventually become its own library)
3575    
3576    Notes:
3577    
3578    I am not 100% happy with the way I separated the elaborator (and its
3579    data structures) from FLINT.  Two instances of the same problem:
3580    
3581        1. Data structures contain certain fields that carry FLINT-specific
3582           information.  I hacked around this using exn and the property list
3583           module from smlnj-lib.  But the fact that there are middle-end
3584           specific fields around at all is a bit annoying.
3585    
3586        2. The elaborator calculates certain FLINT-related information.  I tried
3587           to make this as abstract as I could using functorization, but, again,
3588           the fact that the elaborator has to perform calculations on behalf
3589           of the middle-end at all is not nice.
3590    
3591        3. Having to used exn and property lists is unfortunate because it
3592           weakens type checking.  The other alternative (parameterizing
3593           nearly *everything*) is not appealing, though.
3594    
3595    I removed the "rebinding =" warning hack because due to the new organization
3596    it was awkward to maintain it.  As a result, the compiler now issues some of
3597    these warnings when compiling init.cmi during bootstrap compilation. On
3598    the plus side, you also get a warning when you do, for example:
3599       val op = = Int32.+
3600    which was not the case up to now.
3601    
3602    I placed "assign" and "deref" into the _Core structure so that the
3603    code that deals with the "lazy" keyword can find them there.  This
3604    removes the need for having access to the primitive environment
3605    during elaboration.
3606    
3607    ----------------------------------------------------------------------
3608    Name: Matthias Blume
3609    Date: 2001/08/13
3610    Tag: blume-20010813-closures
3611    Description:
3612    
3613    This fix was sent to us by Zhong Shao.  It is supposed to improve the
3614    performance of certain loops by avoiding needless closure allocation.
3615    
3616    -------------------------------------------------------------------