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 808, Wed Apr 18 15:43:09 2001 UTC revision 1602, Tue Aug 10 16:58:44 2004 UTC
# Line 8  Line 8 
8  The form of an entry should be:  The form of an entry should be:
9    
10  Name:  Name:
11  Date:  Date: yyyy/mm/dd
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14    
15  ----------------------------------------------------------------------  ----------------------------------------------------------------------
16    Name: Matthias Blume (blume (at) tti - c (dot) org)
17    Date: 2004/10/04 12:00:00 CDT
18    Tag: Release_110_48
19    Description:
20    
21    New working version (110.48).  NEW BOOTFILES!
22    
23    ----------------------------------------------------------------------
24    Name: Allen Leung (leunga (at) reservoir (dot) com)
25    Date: 2004/08/09 12:21:00 EST
26    Tag: leunga-110_47-dijsktra
27    Description:
28    
29    Bug fix from Carl Hauser:
30    
31    single_source_shortest_paths in dijkstra.sml was observed to get wrong
32    answers (by comparing to single_source_shortest_paths in bellman-ford.sml).
33    
34    The problem is that following the expression A.update(dist,s,Num.zero)
35    it is necessary to update the priority queue using Q.decreaseWeight(Q,s).
36    
37    ----------------------------------------------------------------------
38    Name: Matthias Blume (blume (at) tti - c (dot) org)
39    Date: 2004/08/06 18:10:00 CDT
40    Tag: blume-20040806-cmdline
41    Description:
42    
43    Fiddled with handling of command-line options:
44    
45      * sml now quits after processing the command line
46        if -H, -S, -h<n>, or -s<n> appears as the last
47        command-line argument
48      * a new option -q terminates the session when encountered on
49        the command line; subsequent arguments will be ignored
50      * bug fixes: short (erroneous) arguments are no longer ignored
51        completely
52    
53    ----------------------------------------------------------------------
54    Name: Allen Leung (leunga (at) reservoir (dot) com)
55    Date: 2004/08/04 18:17:00 EST
56    Tag: leunga-110_47-ppc-ibm-asm
57    Description:
58    
59      - Added minimal IBM assembly syntax support for PowerPC.
60    
61      - Cygwin: manually changed the file cygwin.def.  Some exported symbols have
62        been altered in the runtime.  We need an automatic way to keep the file
63        in sync.
64    
65    ----------------------------------------------------------------------
66    Name: Matthias Blume (blume (at) tti - c (dot) org)
67    Date: 2004/08/04 14:00:00 CDT
68    Tag: Release_110_47
69    Description:
70    
71    New working version (110.47).  NEW BOOTFILES!
72    
73    ----------------------------------------------------------------------
74    Name: Matthias Blume (blume (at) tti - c (dot) org)
75    Date: 2004/08/03 14:25:00 CDT
76    Tag: blume-20040803-callingconv
77    Description:
78    
79    Added low-level support for choosing C calling conventions by
80    twiddling the type of rawccall.  (See
81    src/compiler/Semant/types/cproto.sml for details.)
82    
83    ----------------------------------------------------------------------
84    Name: Matthias Blume (blume (at) tti - c (dot) org)
85    Date: 2004/08/02 15:55:00 CDT
86    Tag: blume-20040802-backout
87    Description:
88    
89    Backed out of change to win32-filesys.c.  The earlier patch to
90    get_file_time caused CM to produce files with the wrong time stamp.
91    
92    ----------------------------------------------------------------------
93    Name: Matthias Blume (blume (at) tti - c (dot) org)
94    Date: 2004/08/02 14:45:00 CDT
95    Tag: blume-20040802-nlffi-win32
96    Description:
97    
98    Added NLFFI support for Win32, adapted from a patch provided by David
99    Hansel.  This is currently completely untested.  Also, the issue
100    concerning stdcall vs. ccall is still unresolved.
101    
102    ----------------------------------------------------------------------
103    Name: Matthias Blume (blume (at) tti - c (dot) org)
104    Date: 2004/07/30 17:55:00 CDT
105    Tag: blume-20040730-various
106    Description:
107    
108    Gearing up towards 110.47...
109    
110    - various minor bugfixes to ml-nlffigen
111    - a beginning of a manual for nlffi
112    
113    - eliminated 'export name=value' in config/install.sh as this does
114      not work with certain versions of /bin/sh
115      (Thanks to David King at Motorola for catching this.)
116    
117    - several bugfixes provided or suggested by David Hansel at Reactive Systems:
118      - added a test for tm==NULL to gmtime.c and localtime.c
119      - applied patch for incorrect GetFileTime under win32
120      - toSeconds -> toMilliseconds in Win32/win32-process.sml
121    
122    ----------------------------------------------------------------------
123    Name: Matthias Blume (blume (at) tti - c (dot) org)
124    Date: 2004/07/21 18:20:00 CDT
125    Tag: blume-20040721-nlffigen
126    Description:
127    
128    - Fixed minor issue in ml-nlffigen: Now generate structure T_foo
129      for a typedef to an incomplete type, but leave out the "typ" member.
130      (This is just for consistency.)
131    - Started to produce what is supposed to become better (i.e., comprehensive)
132      documentation of what ml-nlffigen does and produces.
133    
134    ----------------------------------------------------------------------
135    Name: Matthias Blume (blume (at) tti - c (dot) org)
136    Date: 2004/07/14 16:25:00 CDT
137    Tag: blume-20040714-union
138    Description:
139    
140    Added C_UNION to c-calls/c-types.sml and updated the machinery
141    (ml-nlffigen, cproto.sml) that conveys C function interface
142    information to the code generator.
143    
144    However, the actual architecture-specific implementation of function
145    arguments and results that are C unions is still not implemented.
146    
147    ----------------------------------------------------------------------
148    Name: Allen Leung (leunga (at) reservoir (dot) com)
149    Date: 2004/07/14 14:38:00 EST
150    Tag: leunga-110_46_1-ppc-lwzu
151    Description:
152    
153        Added these instructions to the PowerPC architecture:
154        LBZU(X), LHZU(X), LWZU(X),
155        STWU(X), STFDU, STFSU
156    
157        etc...
158    
159        Note: I haven't added their instruction encoding into the description.
160    
161    ----------------------------------------------------------------------
162    Name: Allen Leung (leunga (at) reservoir (dot) com)
163    Date: 2004/07/13 15:04:00 EST
164    Tag: leunga-110_46_1-ppc-lwarx
165    Description:
166    
167       Added the two instructions LWARX and STWCX to the PowerPC
168    instruction set.
169    
170       A (untested) rewrite of loop-structure.sml.  The old version
171    is completely broken.
172    
173    ----------------------------------------------------------------------
174    Name: Matthias Blume (blume (at) tti - c (dot) org)
175    Date: 2004/07/13 13:50:00 CDT
176    Tag: blume-20040713-nlffi
177    Description:
178    
179    - use paramAlloc to report c-calls with too many arguments
180      (for PPC version where parameter area is pre-allocated)
181    - added ccall_maxargspace to machspec (to implement the above)
182    - made "make" commend in CM's "make" tool configurable
183    - added option (default: on) for passing the name of the SML/NJ's "bin"
184      directory to "make"; the call looks like this:
185    
186         make <options> SMLNJ_BINDIR=<dir> <target>
187    
188      This can be used by the Makefile to, e.g., pick the "right" version
189      of ml-nlffigen.
190    - minor code tweaks
191    
192    ----------------------------------------------------------------------
193    Name: Matthias Blume (blume (at) tti - c (dot) org)
194    Date: 2004/07/12 22:50:00 CDT
195    Tag: blume-110_46_1-macosx-nlffi
196    Description:
197    
198    NLFFI under Mac OS X now working (sort of).  This is largely untested,
199    though.
200    
201    Note:  1. You have to make a new, clean build of the runtime system.
202           2. There are new BOOTFILES, you have to use them!
203              (Doing the bootstrap process yourself would be *very* painful!
204               If you absolutely have to do it, build the system under
205               a different architecture and then cross-compile.)
206    
207    Version bumped to 110.46.1 to account for runtime data format changes.
208    
209    ----------------------------------------------------------------------
210    Name: Matthias Blume (blume (at) tti - c (dot) org)
211    Date: 2004/06/18 14:30:00 CDT
212    Tag: blume-20040618-unix
213    Description:
214    
215    Changed the implementation of structure Unix so that the same stream
216    is returned every time one of the {text,bin}{In,Out}streamOf functions
217    is invoked on the same proc.  This is not what the spec currently
218    says -- although IMO it arguably should.  (See discussion below.)
219    
220    ----------------------------------------------------------------------
221    Name: Matthias Blume (blume (at) tti - c (dot) org)
222    Date: 2004/06/17 18:15:00 CDT
223    Tag: Release_110_46
224    Description:
225    
226    New working version (110.46).  NEW BOOTFILES!
227    
228    ----------------------------------------------------------------------
229    Name: Matthias Blume (blume (at) tti - c (dot) org)
230    Date: 2004/06/17 17:20:00 CDT
231    Tag: blume-20040617-timer-unix
232    Description:
233    
234    Changed the interface of structures Timer and Unix to match the most
235    recent Basis spec.
236    
237    In the case of Unix there still seems to be an open/weird issue:
238    
239       The {text,bin}{In,Out}streamOf functions are supposed to create
240       fresh streams whenever they are called -- as opposed to have them
241       return the same stream every time.  This design is supposed to
242       prevent space leaks caused by proc values hanging on to streams.
243    
244       The reap function, on the other hand, is supposed to close the
245       streams.  This cannot be done without having a handle on the
246       stream in proc after all...
247    
248    I took the liberty to implement the following stopgap solution:
249    
250      The proc value hangs on to the most recently created stream(s).
251      Reap closes those.  If either or both of the two streams hadn't
252      been created at all yet, then reap will close the corresponding
253      file descriptors directly.
254    
255    PS: I don't understand the original space leak argument anymore.  If
256    a proc hangs on to the imperative stream, then I/O operations on those
257    will advance the state of the cached stream and avoid the space leak.
258    
259    ----------------------------------------------------------------------
260    Name: Matthias Blume (blume (at) tti - c (dot) org)
261    Date: 2004/05/28 16:45:00 CDT
262    Tag: blume-20040528-basis
263    Description:
264    
265    Added signature PACK_REAL and exported functor PrimIO.
266    
267    ----------------------------------------------------------------------
268    Name: Matthias Blume (blume (at) tti - c (dot) org)
269    Date: 2004/05/25 16:00:00 CDT
270    Tag: blume-20040525-group-owner
271    Description:
272    
273    CM now ignores (but still accepts) the "owner" information in group
274    descriptions.  The owner of a group is its next enclosing
275    library. Each group must have a unique owner.  (There is a virtual
276    "toplevel" library that owns groups which are not nested within a real
277    library.)  Previously, each group had to explicitly declare its owner,
278    and CM would check that such a declaration is correct.  The new scheme
279    is to have CM check that for each group there is precisely one owning
280    library.
281    
282    The advantage of the new scheme is that the programmer no longer needs
283    to maintain the somewhat annoying owner information.  The downside is
284    that CM cannot enforce the ownership rule across multiple runs of
285    CM.make.  Fortunately, enclosing the same group in two different
286    libraries A and B which are not part of the same program does not
287    cause real problems.
288    
289    ----------------------------------------------------------------------
290    Name: Matthias Blume (blume (at) tti - c (dot) org)
291    Date: 2004/05/20 16:00:00 CDT
292    Tag: blume-20040520-win32
293    Description:
294    
295    Made the win32 version work again.  (Strangely, a misplaced comma had
296    slipped into win32-process.c which prevented the runtime from being
297    compiled correctly.)
298    
299    Also, included a minor addition to ml-build.bat analogous to what was
300    done in blume-20040519-ml-build.
301    
302    ----------------------------------------------------------------------
303    Name: Matthias Blume (blume (at) tti - c (dot) org)
304    Date: 2004/05/19 22:10:00 CDT
305    Tag: blume-20040519-ml-build
306    Description:
307    
308    Arranged for ml-build to clean up after itself a little bit better.
309    The script generates a temporary SML source file and compiles it using
310    CM, so CM generates metadata (GUID, SKEL, objectfile) for it.  It now
311    gets rid of those at the end, so they don't accumulate under .cm.
312    
313    This required a minor change to install.sh because the name of the
314    metadata directory (default: .cm) is actually configurable at
315    installation time.
316    
317    ----------------------------------------------------------------------
318    Name: Matthias Blume (blume (at) tti - c (dot) org)
319    Date: 2004/05/18 15:50:00 CDT
320    Tag: blume-20040518-mkreader
321    Description:
322    
323    Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
324    implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
325    
326    ----------------------------------------------------------------------
327    Name: Matthias Blume (blume (at) tti - c (dot) org)
328    Date: 2004/05/11 14:35:00 CDT
329    Tag: blume-20040511-win32sock
330    Description:
331    
332    Added previously missing support for many socket-related functions
333    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
334    for the voluminous patch!
335    
336    (I have not tested this patch under win32 yet.)
337    
338    Here is David's e-mail:
339    
340    Hi,
341    
342    Attached to this email you find a diff against sml/nj 110.45
343    that will enable socket support under Windows.
344    
345    To apply the patch (using unix or cygwin)
346    1) gunzip runtime.diff.gz
347    2) "cd" into "src/runtime" in the source tree of a fresh
348       110.45 installation.
349    3) patch -p 1 < [your/path/to]runtime.diff
350    
351    The code compiles fine but has NOT yet been extensively tested.
352    I only ran a few tests for basic socket client functionality
353    (which worked fine).  Especially the functions that use ioctl
354    are not tested at all and might not work (see below).
355    
356    I implemented this since we want to move to a newer version of sml/nj
357    but need socket support in order to use it.  This is the first time I
358    even had a look at the sml/nj source,  so please review my changes
359    before making this part of the distribution!  Here are a few issues
360    that I think might be better for someone to solve who is more
361    familiar with the sml/nj source (and socket programming):
362    
363    - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
364      exception since I could not figure out what the windows equivalent
365      of these functions is
366    
367    - In sockets-osdep.h there are a some #include statements that are
368      only used in a few files that include sockets-osdep.h
369    
370    - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
371      does not process its return value since I don't know how to
372      report an error upwards.
373    
374    - It would probably be good to have a call to WSACleanup() when
375      the library is unloaded (if there is such a possibility).
376      Otherwise I think Windows will take care of this automatically
377      when the process finishes.
378    
379    - I used ioctlsocket() as a replacement for ioctl() but I have
380      no idea if that is actually the proper replacement on Windows.
381    
382    - All these issues are marked in the code by "FIXME" comments.
383    
384    We use sml/nj extensively in our products and are quite happy
385    with it.  I hope this contribution will help you.
386    
387    Keep up the good work!
388    
389    David
390    
391    ----------------------------------------------------------------------
392    Name: Matthias Blume (blume (at) tti - c (dot) org)
393    Date: 2004/05/11 14:20:00 CDT
394    Tag: blume-20040511-installml
395    Description:
396    
397    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
398    the patch.)
399    
400    ----------------------------------------------------------------------
401    Name: Matthias Blume (blume (at) tti - c (dot) org)
402    Date: 2004/05/11 14:05:00 CDT
403    Tag: blume-20040511-nlffi-netbsd
404    Description:
405    
406    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
407    the patch.)
408    
409    ----------------------------------------------------------------------
410    Name: Matthias Blume (blume (at) tti - c (dot) org)
411    Date: 2004/05/11 12:05:00 CDT
412    Tag: blume-20040511-exports
413    Description:
414    
415    As per request by Adam Chlipala <adam@hcoop.net>, extended various
416    export lists in compiler-related .cm-files.
417    
418    ----------------------------------------------------------------------
419    Name: Matthias Blume (blume (at) tti - c (dot) org)
420    Date: 2004/05/11 11:35:00 CDT
421    Tag: blume-20040511-allsource
422    Description:
423    
424    The installer now honors the "src-smlnj" target again, although its meaning
425    has changed from "all sources required for the compiler" to "all sources
426    the installer knows about".  In other words, if you enable "src-smlnj"
427    in the "targets" file, then the installer will pull in sources for
428    everything.  (Notice that this refers to source code only.  Compiled
429    code is still only installed for modules that were requested explicitly
430    or which are required for other modules that were requested explicitly.)
431    
432    ----------------------------------------------------------------------
433    Name: Matthias Blume (blume (at) tti - c (dot) org)
434    Date: 2004/04/23 17:40:00 CDT
435    Tag: blume-20040423-ieee-scan
436    Description:
437    
438    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
439    in the exponent calculation we get INF or ZERO (depending on the mantissa
440    and the sign of the exponent).
441    
442    ----------------------------------------------------------------------
443    Name: Matthias Blume (blume (at) tti - c (dot) org)
444    Date: 2004/04/23 10:40:00 CDT
445    Tag: blume-20040423-ml-build
446    Description:
447    
448    The ml-build script now terminates with a non-0 status when something
449    goes wrong.
450    
451    ----------------------------------------------------------------------
452    Name: Matthias Blume (blume (at) tti - c (dot) org)
453    Date: 2004/04/22 16:35:00 CDT
454    Tag: blume-20040422-Option
455    Description:
456    
457    Made exception Option to be the same as exception Option.Option
458    (as it should be).
459    
460    ----------------------------------------------------------------------
461    Name: Allen Leung (leunga (at) reservoir (dot) com)
462    Date: 2004/03/19 14:40:00 EST
463    Tag: leunga-20040319-cygwin-nlffi
464    Description:
465    
466    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
467    of SML/NJ.  The problem is that
468    
469       lib = dlopen(NULL, ...)
470       f   = dlsym(lib, "malloc");
471    
472    does not work on Windows unless we explicitly export symbols
473    such as 'malloc' during linking.  We fixed this by explicitly
474    exporting the required symbols with the magic gcc incantation:
475    
476        -Wl,--export-all cygwin.def
477    
478    where cygwin.def is a file containing all the symbols that
479    we wish to export.
480    
481    I suspect this is a Windows problem and we'll have to
482    do the same (somehow with windows compilers) when
483    we build the native win32 version with the system
484    calls LoadLibrary/GetProcAddress.
485    
486    ----------------------------------------------------------------------
487    Name: Matthias Blume (blume (at) tti - c (dot) org)
488    Date: 2004/03/04 16:35:00 CST
489    Tag: blume-20040304-intinf-fmt
490    Description:
491    
492    Fixed problem with IntInf.fmt (sign would show up on the right instead
493    of on the left for BIN, OCT, and HEX).
494    
495    ----------------------------------------------------------------------
496    Name: Matthias Blume (blume (at) tti - c (dot) org)
497    Date: 2004/03/04 11:25:00 CST
498    Tag: blume-20040304-symlinks
499    Description:
500    
501    Fixed problem with installer script (unix only) where bin/ml-yacc and
502    friends pointed (via symlinks) to absolute locations instead of just
503    .run-sml.  This was reported by Vesa A Norrman.
504    
505    ----------------------------------------------------------------------
506    Name: Matthias Blume (blume (at) tti - c (dot) org)
507    Date: 2004/02/13 14:50:00 CST
508    Tag: Release_110_45
509    Description:
510    
511    New working version (110.45). New bootfiles.
512    
513    ----------------------------------------------------------------------
514    Name: Matthias Blume (blume (at) tti - c (dot) org)
515    Date: 2004/01/26 15:15:15 CST
516    Tag: blume-20040126-toplevel
517    Description:
518    
519    Improved handling of exceptions at the interactive toplevel.
520    
521    ----------------------------------------------------------------------
522    Name: Matthias Blume (blume (at) tti - c (dot) org)
523    Date: 2004/01/26 11:25:00
524    Tag: blume-20040126-app
525    Description:
526    
527    Type of top-level "app" corrected.
528    Added code for setting vp_limitPtrMask to Win32-specific runtime.
529    
530    ----------------------------------------------------------------------
531    Name: Matthias Blume (blume (at) tti - c (dot) org)
532    Date: 2003/11/18 17:10 CST
533    Tag: blume-20031118-basis-fiddle
534    Description:
535    
536    - changed Timer interface to what might become the spec
537    - POSIX_FLAGS -> BIT_FLAGS according to spec
538    - some other minor discrepancies wrt. spec eliminated
539    
540    ----------------------------------------------------------------------
541    Name: Matthias Blume (blume (at) tti - c (dot) org)
542    Date: 2003/11/06 12:00:00 CST
543    Tag: Release_110_44
544    Description:
545    
546    New working version (110.44). New bootfiles.
547    
548    ----------------------------------------------------------------------
549    Name: Matthias Blume (blume (at) tti - c (dot) org)
550    Date: 2003/11/04 11:50:00 CST
551    Tag: blume-20031104-move-libraries
552    Description:
553    
554    Eliminated the "dont_move_libraries" directive in config/targets.
555    (The mechanism was broken and could not be fixed easily.  Moreover,
556    there does not seem to be any reason not to move all libraries into
557    lib during installation.  I originally implemented this directive as a
558    backward-compatibility feature when I first introduced the new CM.
559    Now that things have been stable for a long time and going back to the
560    old CM is not an option, there is no reason to keep it around.)
561    
562    ----------------------------------------------------------------------
563    Name: Matthias Blume (blume (at) tti - c (dot) org)
564    Date: 2003/11/03 16:00:00 CST
565    Tag: blume-20031103-installdir
566    Description:
567    
568    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
569    Richards for pointing out the problem and providing the solution.)
570    
571    ----------------------------------------------------------------------
572    Name: Matthias Blume (blume (at) tti - c (dot) org)
573    Date: 2003/10/01 17:05:00 CDT
574    Tag: blume-20031001-lal-mlrisc
575    Description:
576    
577    MLRISC bug fix from Lal.
578    
579    ----------------------------------------------------------------------
580    Name: Matthias Blume (blume (at) tti - c (dot) org)
581    Date: 2003/09/30 16:10:00 CDT
582    Tag: blume-20030930-primio-bat
583    Description:
584    
585    1. Added openVector, nullRd, and nullWr to PRIM_IO.
586    2. Improved .bat files (for Win32 port) to make things work under Win95.
587       (thanks to Aaron S. Hawley for this one)
588    
589    ----------------------------------------------------------------------
590    Name: Matthias Blume (blume (at) tti - c (dot) org)
591    Date: 2003/09/26 16:05:00 CDT
592    Tag: blume-20030926-wrappriv
593    Description:
594    
595    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
596    
597    ----------------------------------------------------------------------
598    Name: Matthias Blume (blume (at) tti - c (dot) org)
599    Date: 2003/09/26 15:00:00 CDT
600    Tag: blume-20030926-110_43_3
601    Description:
602    
603    - additional cleanup
604    - version number bump, NEW BOOTFILES
605    
606    ----------------------------------------------------------------------
607    Name: Matthias Blume (blume (at) tti - c (dot) org)
608    Date: 2003/09/26 12:00:00 CDT
609    Tag: blume-20030926-ppautoload
610    Description:
611    
612    I modified the read-eval-print loop so that the autoloader gets
613    invoked whenever the prettyprinter tries to look up a symbol that
614    is not currently defined in the toplevel environment but which
615    appears in CM's autoload registry.  As a result, we see far fewer of
616    those ?.Foo.Bar.xxx names in the prettyprinter's output.
617    
618    In addition to this I tried to clean up some pieces of the Basis
619    implementation (e.g., Socket, Word8Array) in order to prevent other
620    instances of these ?.Foo.Bar.xxx names from being printed.
621    
622    The mechanism that picks names for types still needs some work, though.
623    (Right now it seems that if there is a type A.t which is defined to
624    be B.u, but B is unavailable at toplevel, then A.t gets printed as
625    "?.B.u" although the perhaps more sensible solution would be to use
626    "A.t" in this case.  In other words, the prettyprinter should follow
627    a chain of DEFtycs not farther than there are corresponding toplevel
628    names in the current environment.)
629    
630    ----------------------------------------------------------------------
631    Name: Matthias Blume (blume (at) tti - c (dot) org)
632    Date: 2003/09/24 16:31:00 CDT
633    Tag: blume-20030924-installer
634    Description:
635    
636    Another installer tweak:  All the ML code for the installer is now
637    compiled during CMB.make and put into a little library called
638    $smlnj/installer.cm.  The installation then simply invokes
639    
640       sml -m $smlnj/installer.cm
641    
642    and everything happens automagically.
643    
644    Win32:  ML code senses value of environment variable SMLNJ_HOME.
645    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
646            and BINDIR.
647    
648    The new scheme guarantees that the ML code responsible for the installation
649    is in sync with the APIs of the main system.  Also, the installer is
650    somewhat faster because the installer script is precompiled.
651    
652    ----------------------------------------------------------------------
653    Name: Matthias Blume (blume (at) tti - c (dot) org)
654    Date: 2003/09/24 15:35:00 CDT
655    Tag: blume-20030924-synsock
656    Description:
657    
658    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
659    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
660    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
661    implementations of non-blocking operations from CML's Socket
662    structure.
663    
664    ----------------------------------------------------------------------
665    Name: Matthias Blume (blume (at) tti - c (dot) org)
666    Date: 2003/09/24 15:10:05 CDT
667    Tag: blume-20030924-sockets
668    Description:
669    
670    1. Fixed SOCKET API and implementation to match Basis spec.
671       This required changing the internal representation of sockets to one
672       that remembers (for each socket file descriptor) whether it is currently
673       blocking or non-blocking.  This state is maintained lazily (i.e., a system
674       call is made only if the state actually needs to change).
675    
676    2. OS-specific details of sockets were moved into separate files, thus
677       making it possible to unify the bulk of the socket implementations
678       between Unix and Win32.
679    
680    3. CML's socket API changed accordingly.
681       (Note that we need to remove non-blocking functions from this API
682       since they are redundant in the case of CML!)
683    
684    4. CML's socket implementation now makes use of non-blocking functions
685       provided by Basis, thus removing all OS-dependent code from this part
686       of CML.
687    
688    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
689    
690    ----------------------------------------------------------------------
691    Name: Matthias Blume (blume (at) tti - c (dot) org)
692    Date: 2003/09/22 12:10:00 CDT
693    Tag: blume-20030922-110_43_2
694    Description:
695    
696    Made a new interim version and bootfiles for developer's bootstrapping
697    convenience.
698    
699       110.43.2 -- NEW BOOTFILES
700    
701    ----------------------------------------------------------------------
702    Name: Matthias Blume (blume (at) tti - c (dot) org)
703    Date: 2003/09/19 15:55:00 CDT
704    Tag: blume-20030919-cmdir
705    Description:
706    
707    1. new-install.sh -> install.sh
708    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
709    3. tweaked installer so that another name instead of .cm can be chosen
710       at install time (by setting the CM_DIR_ARC environment variable
711       during installation); once installation is complete, the name is
712       fixed
713    
714    ----------------------------------------------------------------------
715    Name: Matthias Blume (blume (at) tti - c (dot) org)
716    Date: 2003/09/18 16:00:00 CDT
717    Tag: blume-20030918-110_43_1
718    Description:
719    
720    Made a new interim version and bootfiles for developer's bootstrapping
721    convenience.
722    
723       110.43.1 -- NEW BOOTFILES
724    
725    ----------------------------------------------------------------------
726    Name: Matthias Blume (blume (at) tti - c (dot) org)
727    Date: 2003/09/18 15:20:00 CDT
728    Tag: blume-20030918-misc
729    Description:
730    
731    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
732       this seems to be controversial at the moment) and used those in
733       Posix.ProcEnv.times.
734    
735    2. Added Time.{from,to}Nanoseconds to Time.
736    
737    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
738       For example, fromLargeInt never needs to look at more than 3 "big
739       digits" to get its 53 bits of precision.
740    
741    ----------------------------------------------------------------------
742    Name: Matthias Blume (blume (at) tti - c (dot) org)
743    Date: 2003/09/17 16:30:00 CDT
744    Tag: blume-20030917-real32-slices
745    Description:
746    
747    Added an entry to the primitive environment
748    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
749    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
750    
751    Removed some of the "magic" constants in real64.sml and replaced them
752    with code that generates these values from their corresponding
753    integer counterparts.
754    
755    Made all(?) the slice-related changes to the Basis and made everything
756    compile again...
757    
758    ----------------------------------------------------------------------
759    Name: Matthias Blume (blume (at) tti - c (dot) org)
760    Date: 2003/09/15 17:45:00 CDT
761    Tag: blume-20030915-rbase
762    Description:
763    
764    Fixed bug in Real.fromLargeInt.
765    
766    ----------------------------------------------------------------------
767    Name: Matthias Blume (blume (at) tti - c (dot) org)
768    Date: 2003/09/13 18:11:00 CDT
769    Tag: blume-20030913-libinstall
770    Description:
771    
772    Minor bugfix in config/libinstall (set anchor with path to
773    standalone tool after installing it, otherwise libraries that
774    need ml-lex or ml-yacc won't compile the first time the installer
775    runs).
776    
777    ----------------------------------------------------------------------
778    Name: Matthias Blume (blume (at) tti - c (dot) org)
779    Date: 2003/09/12 11:45:00 CDT
780    Tag: blume-20030912-various
781    Description:
782    
783     - fixed bug in Real.toLargeInt
784     - fixed bug in Posix.ProcEnv.times
785     - changed inputLine functions to return an option
786     - minor installer improvements / bugfixes
787     - changed default @SMLalloc parameter for x86/celeron to 64k
788    
789    ----------------------------------------------------------------------
790    Name: Matthias Blume (blume (at) tti - c (dot) org)
791    Date: 2003/09/09 22:00:00 CDT
792    Tag: Release_110_43
793    Description:
794    
795        New working release 110.43.  New bootfiles.
796    
797    ----------------------------------------------------------------------
798    Name: Matthias Blume (blume (at) tti - c (dot) org)
799    Date: 2003/09/09 19:20:00 CDT
800    Tag: blume-20030909-installer
801    Description:
802    
803    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
804    Modified config/install.bat to take advantage of it.  Also modified
805    config/install.sh (and called it config/new-install.sh) to take advantage
806    of it on Unix systems.  (The SML code is (supposed to be) platform-
807    independent.)
808    
809    The installer can now install everything under Win32
810    as well as under *nix as long as it compiles.
811    
812    Other changes:
813    
814      - made CML compile again under Win32
815      - made eXene compile under Win32 (by providing a fake structure UnixSock
816        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
817      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
818        same as type int;  under Win32 it isn't
819      - fixed some slice-related problems in the win32-specific parts of CML
820      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
821        (under Win32, the volume name is case-insensitive, and the
822        OS.Path code compares volume names for equality)
823    
824    ----------------------------------------------------------------------
825    Name: Matthias Blume (blume (at) tti - c (dot) org)
826    Date: 2003/09/08 11:55:00 CDT
827    Tag: blume-20030908-fullpath
828    Description:
829    
830    Made Win32 version of OS.FileSys.fullPath return current directory
831    when given an empty string.  This is what the spec says, and incidentally,
832    CM depends on it.  (CM otherwise goes into an infinite loop in certain
833    cases when presented with the name of a non-existing .cm file.)
834    
835    ----------------------------------------------------------------------
836    Name: Matthias Blume (blume (at) tti - c (dot) org)
837    Date: 2003/09/04 16:30:00 CDT
838    Tag: blume-20030905-slices-etc
839    Description:
840    
841      1. Changed interface to vectors and arrays in Basis to match
842         (draft) Basis spec.
843      2. Added signatures and implementations of slices according to
844         Basis spec.
845      3. Edited source code throughout the system to make it compile again
846         under 1. and 2.  (In some cases code had to be added to have it
847         match the new signatures.)
848      4. MLRISC should be backward-compatible: the copies of the originals
849         of files that needed to change under 3. were retained, the .cm files
850         check the compiler version number and use old versions when
851         appropriate.
852      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
853         dirstream -> string option (in accordance with Basis spec).
854      6. When generating code that counts lines, ml-lex used function
855         CharVector.foldli, taking advantage of its old interface.
856         This has been replaced with the corresponding code from
857         CharVectorSlice.  (html-lex must be re-lexed!)
858      7. BitArray in smlnj-lib/Util has been extended/modified to match the
859         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
860      8. Removed temporary additions (fromInternal, toInternal) from the
861         (now obsolete) IntInf in smlnj-lib/Util.
862      9. Cleaned up structure Byte.
863     10. Added localOffset, scan, and fromString to Date (according to spec).
864         Cleaned/corrected implementation of Date.
865         (Still need to check for correctness; implement better canonicalizeDate.)
866     11. Added "scan" to signature IEEE_REAL.
867     12. Some improvements to IntInf [in particular: efficiency-hack for
868         mod and rem when second operand is 2 (for parity checks).]
869     13. Changed representation of type Time.time, using a single IntInf.int
870         value counting microseconds.  This considerably simplified the
871         implementation of structure Time.  We now support negative time
872         values; scan and fromString handle signs.
873     14. Functor PrimIO now takes two additional arguments (VectorSlice and
874         ArraySlice).
875    
876    ----------------------------------------------------------------------
877    Name: Matthias Blume (blume (at) tti - c (dot) org)
878    Date: 2003/08/28 17:00:00 CDT
879    Tag: blume-20030828-intinf
880    Description:
881    
882    This is a major update which comes with a version number bump
883    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
884    and an implementation of IntInf in the Basis.
885    
886    There are a fairly large number of related changes and updates throughout
887    the system:
888    
889    Basis:
890            - Implemented IntInf.
891            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
892            - Added some missing Real64 operations, most notably Real.toLargeInt.
893            - Added FixedInt as a synonym for Int32.
894    
895    compiler:
896            * Added support for a built-in intinf type.
897              - literals
898              - pattern matching
899              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
900              - overloading on literals and operations
901    
902            This required adding a primitive type intinf, some additional
903            primops, and implementations for several non-trivial intinf
904            operations in Core.  (The intinf type is completely abstract
905            to the compiler; all operations get delegated back to the Core.)
906    
907            * Intinf equality is handled by polyequal.  However, the compiler
908            does not print its usual warning in this case (since polyequal
909            is the right thing to do there).
910    
911            * Improved the organization of structure InlineT.
912    
913            * A word about conversion primops:
914            If conversions involving intinf do not cancel out during
915            CPS contract, then the compiler must insert calls to Core functions.
916            Since all core access must be resolved already during the FLINT
917            translate phase, it would be too late a the time of CPS contract
918            to add new Core calls.  For this reason, conversion primops
919            for intinf carry two arguments:  1. the numeric argument that
920            they are supposed to convert, and 2. the Core function that
921            can help with this conversion if necessary.  If CPS contract
922            eliminates a primop, then the associated Core function becomes
923            dead and goes away.  Intinf conversion primops that do not get
924            eliminated by CPS contract get rewritten into calls of their
925            core functions by a separate, new phase.
926    
927    interactive system:
928            - Control.Print.intinfDepth controls max length of intinf constants
929            being printed. (Analogous to Control.Print.stringDepth.)
930            - Cleanup in printutil and pputil: got rid of unused stuff and
931            duplicates; replaced some of the code with code that makes better
932            use of library functionality.
933    
934    CM:
935            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
936            are no longer silent.
937    
938    CKIT:
939            Fixed mismatched uses of Int32 and LargeInt.  I always decided
940            in favor of LargeInt -- which is now the same as IntInf.
941            CKIT-knowledgable people should check whether this is what's
942            intended and otherwise change things back to using Int32 or
943            FixedInt.
944    
945    Throughout the code:
946            Started using IntInf.int literals and built-in operations
947            (e.g., comparison with 0) where this seems appropriate.
948    
949    
950    ----------------------------------------------------------------------
951    Name: Dave MacQueen (dbm@cs.uchicago.edu)
952    Date: 2003/08/13 11:36:00 CDT
953    Tag: dbm-20030813-mcz-merge1
954    Description:
955    
956    Merging changes from the mcz-branch development branch into trunk.
957    These changes involve replacement of the emulated old prettyprinter
958    interface with direct use of the SML/NJ Lib PP library, and fixing
959    of a couple of bugs (895, 1186) relating to error messages.  A new
960    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
961    has been added.
962    
963    ----------------------------------------------------------------------
964    Name: Matthias Blume (blume (at) tti - c (dot) org)
965    Date: 2003/08/11 15:45:00 CDT
966    Tag: blume-20030811-windows
967    Description:
968    
969    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
970    
971        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
972    
973    This patch restores SML/NJ's ability to run under win32.  There are a
974    number of changes, including fixes for several bugs that had gone
975    unnoticed until now:
976    
977      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
978        trying to run the win32 version from within cygwin.)
979    
980      - There are a number of simple .bat scripts that substitute for their
981        corresponding Unix shell-scripts.  (See below.)
982    
983      - The internals of ml-build have been modified slightly.  The main
984        difference is that instead of calling ".link-sml" (or link-sml.bat)
985        using OS.Process.system, the ML process delegates this task back
986        to the script.  Otherwise problems arise in mixed environments such
987        as Cygwin where scripts look and work like Unix scripts, but
988        where OS.Process.system cannot run them.
989    
990      - In CM, the srcpath pickler used native pathname syntax -- which
991        is incorrect in the case of cross-compilation.  The new pickle format
992        is independent of platform-specific naming conventions.
993    
994      - Path configuration files (such as lib/pathconfig) can now choose
995        between native and standard syntax.  Placing a line of the form
996    
997            standard!
998    
999        into the file causes all subsequent paths to be interpreted using
1000        CM standard pathname syntax (= Unix conventions); a line
1001    
1002            native!
1003    
1004        switches back to native style.  This was needed so that
1005        path config files can be written portably, see src/system/pathconfig.
1006    
1007      - Runtime system:
1008    
1009          - win32-filesys.c:  get_file_time and set_file_time now
1010            access modification time, not creation time.
1011    
1012          - I/O code made aware of new array representation.
1013    
1014          - Bug fixes in X86.prim.masm.
1015    
1016       - src/system/makeml made aware of win32. (For use under cygwin
1017            and other Unix-environments for windows.)
1018    
1019       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
1020         which caused BinIO.inputAll to fail consistently.
1021    
1022    .bat scripts:
1023    
1024       Windows .bat scripts assume that SMLNJ_HOME is defined.
1025    
1026       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
1027         applications (sml, ml-yacc, ml-lex).
1028       - ml-build.bat: analogous to ml-build.
1029       - config\install.bat:  Analogous to config/install.sh.  This requires
1030         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
1031         (nmake etc. must be on the path, and vcvars32 must have been run.)
1032         Moreover, sources for ml-lex and ml-yacc need to exist under src,
1033         and the bootfile hierarchy must have been unpacked under
1034         sml.boot.x86-win32.
1035         The script is very primitive and does a poor job at error checking.
1036         It only installs the base system, ml-lex, and ml-yacc.  No other
1037         libraries are being installed (i.e., you get only those that
1038         are part of the compiler.)
1039       - link-sml.bat:  analogous to .link-sml, but not currently used
1040    
1041    Unrelated bug fixes:
1042    
1043      - ml-nlffigen now exports structures ST_* corresponding to incomplete
1044        types.
1045      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
1046        otherwise.)
1047    
1048    ----------------------------------------------------------------------
1049    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
1050    Date: 2003/06/17
1051    Tag: macqueen-20030617-bug895
1052    Description:
1053    
1054    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
1055    Tag will be used for new development branch (mcz-branch) for
1056    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
1057    
1058    ----------------------------------------------------------------------
1059    Name: Matthias Blume (blume (at) tti - c (dot) org)
1060    Date: 2003/05/27 16:55:00 CDT
1061    Tag: blume-20030527-polyeq
1062    Description:
1063    
1064    Tried to eliminated most cases of polymorphic equality.
1065    
1066    ----------------------------------------------------------------------
1067    Name: Matthias Blume (blume (at) tti - c (dot) org)
1068    Date: 2003/05/21 17:45:00 CDT
1069    Tag: blume-20030517-complete
1070    Description:
1071    
1072    Two changes:
1073    
1074      1. Added a flag for controlling whether non-exhaustive bindings will
1075         be treated as errors (default is false).
1076      2. Cleaned up the *entire* source tree so that CMB.make goes through
1077         without a single non-exhaustive match- or bind warning.
1078    
1079    ----------------------------------------------------------------------
1080    Name: Matthias Blume (blume (at) tti - c (dot) org)
1081    Date: 2003/05/17 10:20:00 CDT
1082    Tag: blume-20030517-absyn
1083    Description:
1084    
1085    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
1086    
1087       This mainly affects the quality of error messages.  However, some
1088       of the code is now more straightforward than before. (Treatment of
1089       the above four constructs in translate.sml is much simpler than
1090       the "macro-expansion" that was going on before.  Plus, the mach-
1091       compiler no longer gets invoked just to be able to compile an
1092       if-expression.)
1093    
1094    2. The ErrorMsg.Error exception is now caught and absorbed by the
1095       interactive loop.
1096    
1097    ----------------------------------------------------------------------
1098    Name: Allen Leung
1099    Date: 2003/05/16 13:05:00 CDT
1100    Tag: leunga-20030516-cygwin-runtime
1101    Description:
1102    
1103        Ported the runtime system to cygwin, which uses the unix
1104    x86-unix bin files.  Missing/buggy features:
1105    
1106       o getnetbyname, getnetbyaddr: these functions seem to be missing in
1107         the Cygwin library.
1108       o Ctrl-C handling may be flaky.
1109       o Windows system calls and Windows I/O are not supported.
1110    
1111       A new set of binfiles is located at:
1112    
1113         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
1114    
1115       This is only needed for bootstrapping the cygwin version of smlnj.
1116       Other x86 versions can use the existing binfiles.
1117    
1118    ----------------------------------------------------------------------
1119    Name: Matthias Blume
1120    Date: 2003/04/08 15:42:00 CDT
1121    Tag: blume-20030408-listpair
1122    Description:
1123    
1124    1. Added a target 'mlrisc' to installer.
1125    
1126    2. Added missing elements to structure ListPair.
1127    
1128    ----------------------------------------------------------------------
1129    Name: Allen Leung
1130    Date: 2003/01/07 10:40:00 EST
1131    Tag: leunga-20030107-int-rem
1132    Description:
1133    
1134    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
1135    The arguments to the SUBL instruction were swapped.
1136    
1137    ----------------------------------------------------------------------
1138    Name: Matthias Blume
1139    Date: 2002/12/12 16:25:00 EST
1140    Tag: blume-20021212-risc-ra
1141    Description:
1142    
1143    Fixed a serious bug in the rewrite code for FP spilling/reloading that
1144    sent the RA into an infinite loop when floating point registers get
1145    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
1146    110.37 and 110.38.)
1147    There was another set of potential problems related to the handling of
1148    MLRISC annotations (but those did not yet cause real problems, apparently).
1149    
1150    ----------------------------------------------------------------------
1151    Name: Matthias Blume
1152    Date: 2002/12/06 22:40:00 EST
1153    Tag: blume-20021206-cm-fileid
1154    Description:
1155    
1156    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
1157    This fixes the problem of CM getting confused by files that suddenly
1158    change their identity (e.g., by getting unlinked and recreated by some
1159    text editor such as vi).  There might be a better/cheaper/cleaner way
1160    of doing this, but for now this will have to do.
1161    
1162    ----------------------------------------------------------------------
1163    Name: Matthias Blume
1164    Date: 2002/10/28 09:50:00 EST
1165    Tag: blume-20021028-typecheck
1166    Description:
1167    
1168    Exported structure Typecheck from $smlnj/viscomp/core.cm.
1169    
1170    ----------------------------------------------------------------------
1171    Name: Matthias Blume
1172    Date: 2002/10/17 09:10:00 EDT
1173    Tag: Release_110_42
1174    Description:
1175    
1176    In good old tradition, there has been a slight hiccup so that we have
1177    to patch 110.42 after the fact.   The old release tag has been replaced
1178    (see below).
1179    
1180    The change solves a problem with two competing approaches the
1181    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
1182    each other's way.
1183    
1184    This change only affects the runtime system code and the installer script.
1185    (No new bootfiles.)
1186    
1187    ----------------------------------------------------------------------
1188    Name: Matthias Blume
1189    Date: 2002/10/16 12:00:00 EDT
1190    Tag: Release_110_42_removed
1191    Description:
1192    
1193    New working release.  New bootfiles.
1194    
1195    ----------------------------------------------------------------------
1196    Name: Matthias Blume
1197    Date: 2002/10/10 13:10:00 EDT
1198    Tag: blume-20021010-ppc-divs
1199    Description:
1200    
1201    The mltree operator DIVS must be implemented with an overflow check on
1202    the PPC because the hardware indicates divide-by-zero using "overflow" as
1203    well.
1204    
1205    ----------------------------------------------------------------------
1206    Name: Matthias Blume
1207    Date: 2002/07/23 11:45:00
1208    Tag: blume-20020723-smlnj-home
1209    Description:
1210    
1211    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
1212    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
1213    CM_PATHCONFIG is also set), the path configuration file is assumed
1214    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
1215    the entire tree to some other place and everything will "just work".
1216    
1217    (Companion commands such as ml-build and ml-makedepend also sense this
1218    variable.)
1219    
1220    ----------------------------------------------------------------------
1221    Name: Matthias Blume
1222    Date: 2002/07/12 21:19:00 EDT
1223    Tag: blume-20020712-liveness
1224    Description:
1225    
1226    Exported two useful "step" functions from liveness module (MLRISC).
1227    
1228    ----------------------------------------------------------------------
1229    Name: Matthias Blume
1230    Date: 2002/07/05 16:00 EDT
1231    Tag: Release_110_41
1232    Description:
1233    
1234    New working release.  New bootfiles.
1235    
1236    ----------------------------------------------------------------------
1237    Name: Matthias Blume
1238    Date: 2002/07/05 10:25:00 EDT
1239    Tag: blume-20020705-btimp
1240    Description:
1241    
1242    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
1243    other clients can set up backtracing support.
1244    
1245    ----------------------------------------------------------------------
1246    Name: Matthias Blume
1247    Date: 2002/06/25 17:23:00 EDT
1248    Tag: blume-20020625-fpmax
1249    Description:
1250    
1251    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
1252    case.  (The sense of the isNaN test was reversed -- which made min and
1253    max always return their first argument.)
1254    
1255    ----------------------------------------------------------------------
1256    Name: Matthias Blume
1257    Date: 2002/06/11
1258    Tag: blume-20020611-unixpath
1259    Description:
1260    
1261    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
1262    
1263    ----------------------------------------------------------------------
1264    Name: Matthias Blume
1265    Date: 2002/06/10 16:35:00 EDT
1266    Tag: blume-20020610-ieeereal
1267    Description:
1268    
1269    I back-ported my implementation of IEEEReal.fromString from the
1270    idlbasis-devel branch so that we can test it.
1271    
1272    Another small change is that ppDec tries to give more information
1273    than just "<sig>" in the case of functors.  However, this code is
1274    broken in some mysterious way if the functor's body's signature
1275    has not been declared by ascription but gets inferred from the
1276    implementation.  This needs fixing...
1277    
1278    ----------------------------------------------------------------------
1279    Name: Matthias Blume
1280    Date: 2002/05/31
1281    Tag: blume-20020531-btrace-mode
1282    Description:
1283    
1284    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
1285    the wayside when I switched over to using Controls everywhere.)
1286    
1287    ----------------------------------------------------------------------
1288    Name: Lal George
1289    Date: 2002/05/23 12:21:40 EDT
1290    Tag: george-20020523-visual-labels
1291    Description:
1292    
1293    Labels are now displayed in the graphical output to make
1294    the fall-through and target blocks obvious.
1295    
1296    ----------------------------------------------------------------------
1297    Name: Matthias Blume
1298    Date: 2002/05/22 11:03:00 EDT
1299    Tag: blume-20020522-shrink
1300    Description:
1301    
1302    John tweaked yesterday's fix for 1131 to handle an out-of-memory
1303    situation that comes up when allocating huge arrays.
1304    
1305    ----------------------------------------------------------------------
1306    Name: Matthias Blume
1307    Date: 2002/05/21 16:00:00 EDT
1308    Tag: Release_110_40
1309    Description:
1310    
1311    New working release (110.40).  New bootfiles.
1312    
1313    [Also: John Reppy fixed GC bug 1131.]
1314    
1315    ----------------------------------------------------------------------
1316    Name: Matthias Blume
1317    Date: 2002/05/21 12:35:00 EDT
1318    Tag: blume-20020521-cmdoc
1319    Description:
1320    
1321    CM documentation update.
1322    
1323    ----------------------------------------------------------------------
1324    Name: Matthias Blume
1325    Date: 2002/05/21 10:55:00 EDT
1326    Tag: blume-20020521-misc
1327    Description:
1328    
1329    - John tweaked runtime to be silent on heap export (except when
1330      GC messages are on).
1331    - I added a few more things (cross-compiling versions of CMB) to
1332      config/preloads (as suggestions).
1333    
1334    ----------------------------------------------------------------------
1335    Name: Matthias Blume
1336    Date: 2002/05/20 22:25:00 EDT
1337    Tag: blume-20020520-controls
1338    Description:
1339    
1340    - Added ControlUtil structure to control-lib.cm.
1341    - Use it throughout.
1342    - Used Controls facility to define MLRISC controls (as opposed to
1343      registering MLRISC control ref cells with Controls after the fact)
1344    - Fixed messed-up controls priorities.
1345    
1346    * Removed again all the stuff from config/preloads that one wouldn't
1347      be able to preload at the time the initial heap image is built.
1348      (Many libraries, e.g., CML, do not exist yet at this time.  The
1349       only libraries that can be preloaded via config/preloads are those
1350       that come bundled with the bootfiles.)
1351    
1352    ----------------------------------------------------------------------
1353    Name: Matthias Blume
1354    Date: 2002/05/20 10:59:00 EDT
1355    Tag: blume-20020520-preloads
1356    Description:
1357    
1358    Added a lot of commented-out suggestions for things to be included
1359    in config/preloads.
1360    
1361    ----------------------------------------------------------------------
1362    Name: Allen Leung
1363    Date: 2002/05/18 14:20:00 EDT
1364    Tag: leunga-20020518-mdl
1365    Description:
1366    
1367    o Made the mdl tool stuff compile and run again.
1368    o I've disabled all the stuff that depends on RTL specifications; they
1369      are all badly broken anyway.
1370    
1371    ----------------------------------------------------------------------
1372    Name: Matthias Blume
1373    Date: 2002/05/17 16:49:00 EDT
1374    Tag: blume-20020517-controls
1375    Description:
1376    
1377    0. John Reppy made several modifications to the SML/NJ library.
1378       In particular, there is a shiny new controls-lib.cm.
1379    
1380    1. Pushed new controls interface through compiler so that everything
1381       compiles again.
1382    
1383    2. Added FormatComb and FORMAT_COMB to the CML version of the
1384       SML/NJ library (so that CML compiles again).
1385    
1386    3. Modified init scripts because XXX_DEFAULT environment variables
1387       are no longer with us.  (Boot-time initialization is now done
1388       using the same environment variables that are also used for
1389       startup-time initialization of controls.)
1390    
1391    ----------------------------------------------------------------------
1392    Name: Lal George
1393    Date: 2002/05/15 09:20:10 EDT
1394    Tag: george-20020515-pseudo-op-decls
1395    Description:
1396    
1397            All pseudo-ops emitted before the first segment declaration
1398    such as TEXT, DATA, and BSS directives are assumed to be global
1399    declarations and are emitted first in the assembly file. This is
1400    useful in a number of situations where one has pseudo-ops that are not
1401    specific to any segment, and also works around the constraint that one
1402    cannot have client pseudo-ops in the TEXT segment.
1403    
1404            Because no segment is associated with these declarations it is
1405    an error to allocate any space or objects before the first segment
1406    directive and an exception will be raised. However, we cannot make
1407    this check for client pseudo-ops.
1408    
1409            These top level declarations are a field in the CFG graph_info.
1410    In theory you can continue to add to this field after the CFG has been
1411    built -- provided you know what you are doing;-)
1412    
1413    ----------------------------------------------------------------------
1414    Name: Matthias Blume
1415    Date: 2002/05/13 16:40:00 EDT
1416    Tag: blume-20020513-pp-etc
1417    Description:
1418    
1419    A few minor bugfixes:
1420    
1421      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1422        (Bogus printouts for redefined bindings still occur.  Compiler
1423        bug should no longer occur now.  We need to redo the prettyprinter
1424        from scratch.)
1425    
1426      - CM pathname printer now also adds escape sequences for ( and )
1427    
1428      - commend and docu fixes for ml-nlffi
1429    
1430    ----------------------------------------------------------------------
1431    Name: Matthias Blume
1432    Date: 2002/05/10 16:40:00 EDT
1433    Tag: blume-20020510-erg-textio
1434    Description:
1435    
1436    Applied the following bugfix provided by Emden Gansner:
1437    
1438        Output is corrupted when outputSubstr is used rather than output.
1439    
1440        The problem occurs when a substring
1441    
1442            ss = (s, dataStart, dataLen)
1443    
1444        where dataStart > 0, fills a stream buffer with avail bytes left.
1445        avail bytes of s, starting at index dataStart, are copied into the
1446        buffer, the buffer is flushed, and then the remaining dataLen-avail
1447        bytes of ss are copied into the beginning of the buffer. Instead of
1448        starting this copy at index dataStart+avail in s, the current code
1449        starts the copy at index avail.
1450    
1451        Fix:
1452        In text-io-fn.sml, change line 695 from
1453             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1454        to
1455             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1456    
1457    ----------------------------------------------------------------------
1458    Name: Matthias Blume
1459    Date: 2002/04/12 13:55:00 EDT
1460    Tag: blume-20020412-assyntax
1461    Description:
1462    
1463    1. Grabbed newer assyntax.h from the XFree86 project.
1464    2. Fiddled with how to compile X86.prim.asm without warnings.
1465    3. (Very) Minor cleanup in CM.
1466    
1467    ----------------------------------------------------------------------
1468    Name: Matthias Blume
1469    Date: 2002/04/01 (no joke!) 17:07:00 EST
1470    Tag: blume-20020401-x86div
1471    Description:
1472    
1473    Added full support for div/mod/rem/quot on the x86, using the machine
1474    instruction's two results (without clumsily recomputing the remainder)
1475    directly where appropriate.
1476    
1477    Some more extensive power-of-two support was added to the x86 instruction
1478    selector (avoiding expensive divs, mods, and muls where they can be
1479    replaced with cheaper shifts and masks).  However, this sort of thing
1480    ought to be done earlier, e.g., within the CPS optimizer so that
1481    all architectures benefit from it.
1482    
1483    The compiler compiles to a fixed point, but changes might be somewhat
1484    fragile nevertheless.  Please, report any strange things that you might
1485    see wrt. div/mod/quot/rem...
1486    
1487    ----------------------------------------------------------------------
1488    Name: Matthias Blume
1489    Date: 2002/03/29 17:22:00
1490    Tag: blume-20020329-div
1491    Description:
1492    
1493    Fixed my broken div/mod logic.  Unfortunately, this means that the
1494    inline code for div/mod now has one more comparison than before.
1495    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1496    The problem was with quotient = 0, because that alone does not tell
1497    us which way the rounding went.  One then has to look at whether
1498    remainder and divisor have the same sign...  :(
1499    
1500    Anyway, I replaced the bootfiles with fresh ones...
1501    
1502    ----------------------------------------------------------------------
1503    Name: Matthias Blume
1504    Date: 2002/03/29 14:10:00 EST
1505    Tag: blume-20020329-inlprims
1506    Description:
1507    
1508    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1509    
1510    Primops have changed. This means that the bin/boot-file formats have
1511    changed as well.
1512    
1513    To make sure that there is no confusion, I made a new version.
1514    
1515    
1516    CHANGES:
1517    
1518    * removed REMT from mltree (remainder should never overflow).
1519    
1520    * added primops to deal with divisions of all flavors to the frontend
1521    
1522    * handled these primops all the way through so they map to their respective
1523      MLRISC support
1524    
1525    * used these primops in the implementation of Int, Int32, Word, Word32
1526    
1527    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1528    
1529    * parameterized INLMIN, INLMAX, and INLABS by a numkind
1530    
1531    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1532      floating point
1533    
1534    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1535      and Real (but Real.abs maps to a separate floating-point-only primop)
1536    
1537    
1538    TODO items:
1539    
1540    * Hacked Alpha32 instruction selection, disabling the selection of REMx
1541      instructions because the machine instruction encoder cannot handle
1542      them.  (Hppa, PPC, and Sparc instruction selection did not handle
1543      REM in the first place, and REM is supported by the x86 machine coder.)
1544    
1545    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1546      selection phase.  (The two can be streamlined because the hardware
1547      delivers both quotient and remainder at the same time anyway.)
1548    
1549    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1550      (Currently the behavior is inconsistent both across architectures and
1551      wrt. the draft Basis spec.)
1552    
1553    * Word8 should eventually be handled natively, too.
1554    
1555    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1556      as if there currently is no execution path that could trigger it in
1557      SML/NJ.  (The assumptions underlying functions arith and promotable do not
1558      hold for things like multiplication and division.)
1559    
1560    ----------------------------------------------------------------------
1561    Name: Matthias Blume
1562    Date: 2002/03/27 16:27:00 EST
1563    Tag: blume-20020327-mlrisc-divisions
1564    Description:
1565    
1566    Added support for all four division operations (ML's div, mod, quot,
1567    and rem) to MLRISC.  In the course of doing so, I also rationalized
1568    the naming (no more annoying switch-around of DIV and QUOT), by
1569    parameterizing the operation by div_rounding_mode (which can be either
1570    DIV_TO_ZERO or DIV_TO_NEGINF).
1571    
1572    The generic MLTreeGen functor takes care of compiling all four
1573    operations down to only round-to-zero div.
1574    
1575    Missing pieces:
1576    
1577      * Doing something smarter than relying on MLTreeGen on architectures
1578        like, e.g., the x86 where hardware division delivers both quotient and
1579        remainder at the same time.  With this, the implementation of the
1580        round-to-neginf operations could be further streamlined.
1581    
1582      * Remove inlining support for div/mod/rem from the frontend and replace it
1583        with primops that get carried through to the backend.  Do this for all
1584        int and word types.
1585    
1586    ----------------------------------------------------------------------
1587    Name: Matthias Blume
1588    Date: 2002/03/25 17:25:00 EST
1589    Tag: blume-20020325-divmod
1590    Description:
1591    
1592    I improved (hopefully without breaking them) the implementation of Int.div,
1593    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1594    advantage of the following observations:
1595    
1596      Let  q = x quot y      r = x rem y
1597           d = x div  y      m = x mod y
1598    
1599    where "quot" is the round-to-zero version of integer division that
1600    hardware usually provides.  Then we have:
1601    
1602         r = x - q * y        where neither the * nor the - will overflow
1603         d = if q >= 0 orelse x = q * y then q else q - 1
1604                              where neither the * nor the - will overflow
1605         m = if q >= 0 orelse r = 0 then r else r + y
1606                              where the + will not overflow
1607    
1608    This results in substantial simplification of the generated code.
1609    The following table shows the number of CFG nodes and edges generated
1610    for
1611            fun f (x, y) = x OPER y
1612            (* with OPER \in div, mod, quot, rem *)
1613    
1614    
1615        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1616        --------------------------------------------------------
1617         div |         24 |         39 |         12 |         16
1618         mod |         41 |         71 |         12 |         16
1619        quot |          8 |         10 |          8 |         10
1620         rem |         10 |         14 |          8 |         10
1621    
1622    
1623    ----------------------------------------------------------------------
1624    Name: Matthias Blume
1625    Date: 2002/03/25 22:06:00 EST
1626    Tag: blume-20020325-cprotobug
1627    Description:
1628    
1629    Fixed a bug in cproto (c prototype decoder).
1630    
1631    ----------------------------------------------------------------------
1632    Name: Matthias Blume
1633    Date: 2002/03/25 16:00:00 EST
1634    Tag: blume-20020325-raw-primops
1635    Description:
1636    
1637    I did some cleanup to Allen's new primop code and
1638    replaced yesterday's bootfiles with new ones.
1639    (But they are stored in the same place.)
1640    
1641    ----------------------------------------------------------------------
1642    Name: Matthias Blume
1643    Date: 2002/03/24 22:40:00 EST
1644    Tag: blume-20020324-bootfiles
1645    Description:
1646    
1647    Made the bootfiles that Allen asked for.
1648    
1649    ----------------------------------------------------------------------
1650    Name: Allen Leung
1651    Date: 2002/03/23 15:50:00 EST
1652    Tag: leunga-20020323-flint-cps-rcc-primops
1653    Description:
1654    
1655      1. Changes to FLINT primops:
1656    
1657        (* make a call to a C-function;
1658         * The primop carries C function prototype information and specifies
1659         * which of its (ML-) arguments are floating point. C prototype
1660         * information is for use by the backend, ML information is for
1661         * use by the CPS converter. *)
1662      | RAW_CCALL of { c_proto: CTypes.c_proto,
1663                       ml_args: ccall_type list,
1664                       ml_res_opt: ccall_type option,
1665                       reentrant : bool
1666                     } option
1667       (* Allocate uninitialized storage on the heap.
1668        * The record is meant to hold short-lived C objects, i.e., they
1669        * are not ML pointers.  With the tag, the representation is
1670        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1671        *)
1672      | RAW_RECORD of {tag:bool,sz:int}
1673      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1674    
1675      2.  These CPS primops are now overloaded:
1676    
1677           rawload of {kind:numkind}
1678           rawstore of {kind:numkind}
1679    
1680          The one argument form is:
1681    
1682             rawload {kind} address
1683    
1684          The two argument form is:
1685    
1686             rawload {kind} [ml object, byte-offset]
1687    
1688      3. RAW_CCALL/RCC now takes two extra arguments:
1689    
1690         a. The first is whether the C call is reentrant, i.e., whether
1691            ML state should be saved and restored.
1692         b. The second argument is a string argument specifying the name of
1693            library and the C function.
1694    
1695         These things are currently not handled in the code generator, yet.
1696    
1697      4. In CProto,
1698    
1699         An encoding type of "bool" means "ml object" and is mapped into
1700         C prototype of PTR.  Note that "bool" is different than "string",
1701         even though "string" is also mapped into PTR, because "bool"
1702         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1703    
1704      5. Pickler/unpicker
1705    
1706         Changed to handle RAW_RECORD and newest RAW_CCALL
1707    
1708      6. MLRiscGen,
1709    
1710         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1711         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1712            in the file CodeGen/cpscompile/cps-c-calls.sml
1713    
1714      7. Added the conditional move operator
1715    
1716             condmove of branch
1717    
1718         to cps.  Generation of this is still buggy so it is currently
1719         disabled.
1720    
1721    ----------------------------------------------------------------------
1722    Name: Lal George
1723    Date: 2002/03/22 14:18:25 EST
1724    Tag: george-20020322-cps-branch-prob
1725    Description:
1726    
1727    Implemented the Ball-Larus branch prediction-heuristics, and
1728    incorporated graphical viewers for control flow graphs.
1729    
1730    Ball-Larus Heuristics:
1731    ---------------------
1732    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1733    
1734    By design it uses the Dempster-Shafer theory for combining
1735    probabilities.  For example, in the function:
1736    
1737        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1738    
1739    the ball-larus heuristics predicts that the n=0 is unlikely
1740    (OH-heuristic), and the 'then' branch is unlikely because of the
1741    RH-heuristic -- giving the 'then' branch an even lower combined
1742    probability using the Dempster-Shafer theory.
1743    
1744    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1745    probability of the 'then' branch because of the loop in the else
1746    branch.
1747    
1748    
1749    Graphical Viewing:
1750    ------------------
1751    I merely plugged in Allen's graphical viewers into the compiler. The
1752    additional code is not much. At the top level, saying:
1753    
1754            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1755    
1756    will display the graphical view of the control flow graph just before
1757    back-patching.  daVinci must be in your path for this to work. If
1758    daVinci is not available, then the default viewer can be changed
1759    using:
1760    
1761            Control.MLRISC.getString "viewer"
1762    
1763    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1764    course, these viewers must be in your path.
1765    
1766    The above will display the compilation unit at the level of clusters,
1767    many of which are small, boring, and un-interesting. Also setting:
1768    
1769            Control.MLRISC.getInt "cfg-graphical-view_size"
1770    
1771    will display clusters that are larger than the value set by the above.
1772    
1773    
1774    ----------------------------------------------------------------------
1775    Name: Matthias Blume
1776    Date: 2002/03/21 22:20:00 EST
1777    Tag: blume-20020321-kmp-bugfix
1778    Description:
1779    
1780    Changed the interface to the KMP routine in PreString and fixed
1781    a minor bug in one place where it was used.
1782    
1783    ----------------------------------------------------------------------
1784    Name: Allen Leung
1785    Date: 2002/03/21 20:30:00 EST
1786    Tag: leunga-20020321-cfg
1787    Description:
1788    
1789      Fixed a potential problem in cfg edge splitting.
1790    
1791    ----------------------------------------------------------------------
1792    Name: Allen Leung
1793    Date: 2002/03/21 17:15:00 EST
1794    Tag: leunga-20020321-x86-fp-cfg
1795    Description:
1796    
1797      1. Recoded the buggy parts of x86-fp.
1798    
1799         a. All the block reordering code has been removed.
1800            We now depend on the block placement phases to do this work.
1801    
1802         b. Critical edge splitting code has been simplified and moved into the
1803            CFG modules, as where they belong.
1804    
1805         Both of these were quite buggy and complex.  The code is now much, much
1806         simpler.
1807    
1808      2. X86 backend.
1809    
1810         a. Added instructions for 64-bit support.  Instruction selection for
1811            64-bit has not been committed, however, since that
1812            requires changes to MLTREE which haven't been approved by
1813            Lal and John.
1814    
1815         b. Added support for FUCOMI and FUCOMIP when generating code for
1816            PentiumPro and above.  We only generate these instructions in
1817            the fast-fp mode.
1818    
1819         c. Added cases for JP and JNP in X86FreqProps.
1820    
1821      3. CFG
1822    
1823         CFG now has a bunch of methods for edge splitting and merging.
1824    
1825      4. Machine description.
1826    
1827         John's simplification of MLTREE_BASIS.fcond broke a few machine
1828         description things:
1829    
1830         rtl-build.{sig,sml} and hppa.mdl fixed.
1831    
1832         NOTE: the machine description stuff in the repository is still broken.
1833               Again, I can't put my fixes in because that involves
1834               changes to MLTREE.
1835    
1836    ----------------------------------------------------------------------
1837    Name: Matthias Blume
1838    Date: 2002/03/20 15:55:00 EST
1839    Tag: blume-20020320-kmp
1840    Description:
1841    
1842    Implemented Knuth-Morris-Pratt string matching in PreString and used
1843    it for String.isSubstring, Substring.isSubstring, and
1844    Substring.position.
1845    
1846    (Might need some stress-testing.  Simple examples worked fine.)
1847    
1848    ----------------------------------------------------------------------
1849    Name: Matthias Blume
1850    Date: 2002/03/19 16:37:00 EST
1851    Tag: blume-20020319-witnesses
1852    Description:
1853    
1854    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1855    
1856    This implements a generic mechanism for changing constness qualifiers
1857    anywhere within big C types without resorting to outright "casts".
1858    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1859    modify the constness at the outermost level.)
1860    The implementation of "convert" is based on the idea of "witness"
1861    values -- values that are not used by the operation but whose types
1862    "testify" to their applicability.  On the implementation side, "convert"
1863    is simply a projection (returning its second curried argument).  With
1864    cross-module inlining, it should not result in any machine code being
1865    generated.
1866    
1867    ----------------------------------------------------------------------
1868    Name: Matthias Blume
1869    Date: 2002/03/15 16:40:00 EST
1870    Tag: blume-20020315-basis
1871    Description:
1872    
1873    Provided (preliminary?) implementations for
1874    
1875      {String,Substring}.{concatWith,isSuffix,isSubstring}
1876    
1877    and
1878    
1879      Substring.full
1880    
1881    Those are in the Basis spec but they were missing in SML/NJ.
1882    
1883    ----------------------------------------------------------------------
1884    Name: Matthias Blume
1885    Date: 2002/03/14 21:30:00 EST
1886    Tag: blume-20020314-controls
1887    Description:
1888    
1889    Controls:
1890    ---------
1891    
1892    1. Factored out the recently-added Controls : CONTROLS stuff and put
1893       it into its own library $/controls-lib.cm.  The source tree for
1894       this is under src/smlnj-lib/Controls.
1895    
1896    2. Changed the names of types and functions in this interface, so they
1897       make a bit more "sense":
1898    
1899          module -> registry
1900          'a registry -> 'a group
1901    
1902    3. The interface now deals in ref cells only.  The getter/setter interface
1903       is (mostly) gone.
1904    
1905    4. Added a function that lets one register an already-existing ref cell.
1906    
1907    5. Made the corresponding modifications to the rest of the code so that
1908       everything compiles again.
1909    
1910    6. Changed the implementation of Controls.MLRISC back to something closer
1911       to the original.  In particular, this module (and therefore MLRISC)
1912       does not depend on Controls.  There now is some link-time code in
1913       int-sys.sml that registers the MLRISC controls with the Controls
1914       module.
1915    
1916    CM:
1917    ---
1918    
1919      * One can now specify the lambda-split aggressiveness in init.cmi.
1920    
1921    ----------------------------------------------------------------------
1922    Name: Allen Leung
1923    Date: 2002/03/13 17:30:00 EST
1924    Tag: leunga-20020313-x86-fp-unary
1925    Description:
1926    
1927    Bug fix for:
1928    
1929    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1930    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1931    > - fun f(x,(y,z)) = Real.~ y;
1932    > [autoloading]
1933    > [autoloading done]
1934    >       fchsl   (%eax), 184(%esp)
1935    > Error: MLRisc bug: X86MCEmitter.emitInstr
1936    >
1937    > uncaught exception Error
1938    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1939    
1940    The problem was that the code generator did not generate any fp registers
1941    in this case, and the ra didn't know that it needed to run the X86FP phase to
1942    translate the pseudo fp instruction.   This only happened with unary fp
1943    operators in certain situations.
1944    
1945    ----------------------------------------------------------------------
1946    Name: Matthias Blume
1947    Date: 2002/03/13 14:00:00 EST
1948    Tag: blume-20020313-overload-etc
1949    Description:
1950    
1951    1. Added _overload as a synonym for overload for backward compatibility.
1952       (Control.overloadKW must be true for either version to be accepted.)
1953    
1954    2. Fixed bug in install script that caused more things to be installed
1955       than what was requested in config/targets.
1956    
1957    3. Made CM aware of the (_)overload construct so that autoloading
1958       works.
1959    
1960    ----------------------------------------------------------------------
1961    Name: Matthias Blume
1962    Date: 2002/03/12 22:03:00 EST
1963    Tag: blume-20020312-url
1964    Description:
1965    
1966    Forgot to update BOOT and srcarchiveurl.
1967    
1968    ----------------------------------------------------------------------
1969    Name: Matthias Blume
1970    Date: 2002/03/12 17:30:00 EST
1971    Tag: blume-20020312-version110392
1972    Description:
1973    
1974    Yet another version number bump (because of small changes to the
1975    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1976    
1977    Changes:
1978    
1979      The new pid generation scheme described a few weeks ago was overly
1980      complicated.  I implemented a new mechanism that is simpler and
1981      provides a bit more "stability":  Once CM has seen a compilation
1982      unit, it keeps its identity constant (as long as you do not delete
1983      those crucial CM/GUID/* files).  This means that when you change
1984      an interface, compile, then go back to the old interface, and
1985      compile again, you arrive at the original pid.
1986    
1987      There now also is a mechanism that instructs CM to use the plain
1988      environment hash as a module's pid (effectively making its GUID
1989      the empty string).  For this, "noguid" must be specified as an
1990      option to the .sml file in question within its .cm file.
1991      This is most useful for code that is being generated by tools such
1992      as ml-nlffigen (because during development programmers tend to
1993      erase the tool's entire output directory tree including CM's cached
1994      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1995      revert to the old, broken behavior of SML/NJ, but in specific cases
1996      where there is no danger of interface confusion, its use is ok
1997      (I think).
1998    
1999      ml-nlffigen by default generates "noguid" annotations.  They can be
2000      turned off by specifying -guid in its command line.
2001    
2002    ----------------------------------------------------------------------
2003    Name: Lal George
2004    Date: 2002/03/12 12 14:42:36 EST
2005    Tag: george-20020312-frequency-computation
2006    Description:
2007    
2008    Integrated jump chaining and static block frequency into the
2009    compiler. More details and numbers later.
2010    
2011    ----------------------------------------------------------------------
2012    Name: Lal George
2013    Date: 2002/03/11 11 22:38:53 EST
2014    Tag: george-20020311-jump-chain-elim
2015    Description:
2016    
2017    Tested the jump chain elimination on all architectures (except the
2018    hppa).  This is on by default right now and is profitable for the
2019    alpha and x86, however, it may not be profitable for the sparc and ppc
2020    when compiling the compiler.
2021    
2022    The gc test will typically jump to a label at the end of the cluster,
2023    where there is another jump to an external cluster containing the actual
2024    code to invoke gc. This is to allow factoring of common gc invocation
2025    sequences. That is to say, we generate:
2026    
2027            f:
2028               testgc
2029               ja   L1      % jump if above to L1
2030    
2031            L1:
2032               jmp L2
2033    
2034    
2035    After jump chain elimination the 'ja L1' instructions is converted to
2036    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
2037    up being implemented in their long form (if L2 is far away) using:
2038    
2039            jbe     L3      % jump if below or equal to L3
2040            jmp     L2
2041         L3:
2042            ...
2043    
2044    
2045    For large compilation units L2  may be far away.
2046    
2047    
2048    ----------------------------------------------------------------------
2049    Name: Matthias Blume
2050    Date: 2002/03/11 13:30:00 EST
2051    Tag: blume-20020311-mltreeeval
2052    Description:
2053    
2054    A functor parameter was missing.
2055    
2056    ----------------------------------------------------------------------
2057    Name: Allen Leung
2058    Date: 2002/03/11 10:30:00 EST
2059    Tag: leunga-20020311-runtime-string0
2060    Description:
2061    
2062       The representation of the empty string now points to a
2063    legal null terminated C string instead of unit.  It is now possible
2064    to convert an ML string into C string with InlineT.CharVector.getData.
2065    This compiles into one single machine instruction.
2066    
2067    ----------------------------------------------------------------------
2068    Name: Allen Leung
2069    Date: 2002/03/10 23:55:00 EST
2070    Tag: leunga-20020310-x86-call
2071    Description:
2072    
2073       Added machine generation for CALL instruction (relative displacement mode)
2074    
2075    ----------------------------------------------------------------------
2076    Name: Matthias Blume
2077    Date: 2002/03/08 16:05:00
2078    Tag: blume-20020308-entrypoints
2079    Description:
2080    
2081    Version number bumped to 110.39.1.  NEW BOOTFILES!
2082    
2083    Entrypoints: non-zero offset into a code object where execution should begin.
2084    
2085    - Added the notion of an entrypoint to CodeObj.
2086    - Added reading/writing of entrypoint info to Binfile.
2087    - Made runtime system bootloader aware of entrypoints.
2088    - Use the address of the label of the first function given to mlriscGen
2089      as the entrypoint.  This address is currently always 0, but it will
2090      not be 0 once we turn on block placement.
2091    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
2092      with entry points) from mlriscGen.
2093    
2094    ----------------------------------------------------------------------
2095    Name: Allen Leung
2096    Date: 2002/03/07 20:45:00 EST
2097    Tag: leunga-20020307-x86-cmov
2098    Description:
2099    
2100       Bug fixes for CMOVcc on x86.
2101    
2102       1. Added machine code generation for CMOVcc
2103       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
2104       3. CMOVcc cannot have an immediate operand as argument.
2105    
2106    ----------------------------------------------------------------------
2107    Name: Matthias Blume
2108    Date: 2002/03/07 16:15:00 EST
2109    Tag: blume-20020307-controls
2110    Description:
2111    
2112    This is a very large but mostly boring patch which makes (almost)
2113    every tuneable compiler knob (i.e., pretty much everything under
2114    Control.* plus a few other things) configurable via both the command
2115    line and environment variables in the style CM did its configuration
2116    until now.
2117    
2118    Try starting sml with '-h' (or, if you are brave, '-H')
2119    
2120    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
2121    implements the underlying generic mechanism.
2122    
2123    The interface to some of the existing such facilities has changed somewhat.
2124    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
2125    (The getFoo interface is still there for backward-compatibility, but its
2126    use is deprecated.)
2127    
2128    The ml-build script passes -Cxxx=yyy command-line arguments through so
2129    that one can now twiddle the compiler settings when using this "batch"
2130    compiler.
2131    
2132    TODO items:
2133    
2134    We should go through and throw out all controls that are no longer
2135    connected to anything.  Moreover, we should go through and provide
2136    meaningful (and correct!) documentation strings for those controls
2137    that still are connected.
2138    
2139    Currently, multiple calls to Controls.new are accepted (only the first
2140    has any effect).  Eventually we should make sure that every control
2141    is being made (via Controls.new) exactly once.  Future access can then
2142    be done using Controls.acc.
2143    
2144    Finally, it would probably be a good idea to use the getter-setter
2145    interface to controls rather than ref cells.  For the time being, both
2146    styles are provided by the Controls module, but getter-setter pairs are
2147    better if thread-safety is of any concern because they can be wrapped.
2148    
2149    *****************************************
2150    
2151    One bug fix: The function blockPlacement in three of the MLRISC
2152    backpatch files used to be hard-wired to one of two possibilities at
2153    link time (according to the value of the placementFlag).  But (I
2154    think) it should rather sense the flag every time.
2155    
2156    *****************************************
2157    
2158    Other assorted changes (by other people who did not supply a HISTORY entry):
2159    
2160    1. the cross-module inliner now works much better (Monnier)
2161    2. representation of weights, frequencies, and probabilities in MLRISC
2162       changed in preparation of using those for weighted block placement
2163       (Reppy, George)
2164    
2165    ----------------------------------------------------------------------
2166    Name: Lal George
2167    Date: 2002/03/07 14:44:24 EST 2002
2168    Tag: george-20020307-weighted-block-placement
2169    
2170    Tested the weighted block placement optimization on all architectures
2171    (except the hppa) using AMPL to generate the block and edge frequencies.
2172    Changes were required in the machine properties to correctly
2173    categorize trap instructions. There is an MLRISC flag
2174    "weighted-block-placement" that can be used to enable weighted block
2175    placement, but this will be ineffective without block/edge
2176    frequencies (coming soon).
2177    
2178    
2179    ----------------------------------------------------------------------
2180    Name: Lal George
2181    Date: 2002/03/05 17:24:48 EST
2182    Tag: george-20020305-linkage-cluster
2183    
2184    In order to support the block placement optimization, a new cluster
2185    is generated as the very first cluster (called the linkage cluster).
2186    It contains a single jump to the 'real' entry point for the compilation
2187    unit. Block placement has no effect on the linkage cluster itself, but
2188    all the other clusters  have full freedom in the manner in which they
2189    reorder blocks or functions.
2190    
2191    On the x86 the typical linkage code that is generated is:
2192       ----------------------
2193            .align 2
2194       L0:
2195            addl    $L1-L0, 72(%esp)
2196            jmp     L1
2197    
2198    
2199            .align  2
2200       L1:
2201       ----------------------
2202    
2203    72(%esp) is the memory location for the stdlink register. This
2204    must contain the address of the CPS function being called. In the
2205    above example, it contains the address of  L0; before
2206    calling L1 (the real entry point for the compilation unit), it
2207    must contain the address for L1, and hence
2208    
2209            addl $L1-L0, 72(%esp)
2210    
2211    I have tested this on all architectures except the hppa.The increase
2212    in code size is of course negligible
2213    
2214    ----------------------------------------------------------------------
2215    Name: Allen Leung
2216    Date: 2002/03/03 13:20:00 EST
2217    Tag: leunga-20020303-mlrisc-tools
2218    
2219      Added #[ ... ] expressions to mlrisc tools
2220    
2221    ----------------------------------------------------------------------
2222    Name: Matthias Blume
2223    Date: 2002/02/27 12:29:00 EST
2224    Tag: blume-20020227-cdebug
2225    Description:
2226    
2227    - made types in structure C and C_Debug to be equal
2228    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
2229    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
2230    
2231    ----------------------------------------------------------------------
2232    Name: Matthias Blume
2233    Date: 2002/02/26 12:00:00 EST
2234    Tag: blume-20020226-ffi
2235    Description:
2236    
2237    1. Fixed a minor bug in CM's "noweb" tool:
2238       If numbering is turned off, then truly don't number (i.e., do not
2239       supply the -L option to noweb).  The previous behavior was to supply
2240       -L'' -- which caused noweb to use the "default" line numbering scheme.
2241       Thanks to Chris Richards for pointing this out (and supplying the fix).
2242    
2243    2. Once again, I reworked some aspects of the FFI:
2244    
2245       A. The incomplete/complete type business:
2246    
2247       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
2248         gone!
2249       - ML types representing an incomplete type are now *equal* to
2250         ML types representing their corresponding complete types (just like
2251         in C).  This is still safe because ml-nlffigen will not generate
2252         RTTI for incomplete types, nor will it generate functions that
2253         require access to such RTTI.   But when ML code generated from both
2254         incomplete and complete versions of the C type meet, the ML types
2255         are trivially interoperable.
2256    
2257         NOTE:  These changes restore the full generality of the translation
2258         (which was previously lost when I eliminated functorization)!
2259    
2260       B. Enum types:
2261    
2262       - Structure C now has a type constructor "enum" that is similar to
2263         how the "su" constructor works.  However, "enum" is not a phantom
2264         type because each "T enum" has values (and is isomorphic to
2265         MLRep.Signed.int).
2266       - There are generic access operations for enum objects (using
2267         MLRep.Signed.int).
2268       - ml-nlffigen will generate a structure E_foo for each "enum foo".
2269         * The structure contains the definition of type "mlrep" (the ML-side
2270         representation type of the enum).  Normally, mlrep is the same
2271         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
2272         then mlrep will be defined as a datatype -- thus facilitating
2273         pattern matching on mlrep values.
2274         ("-ec" will be suppressed if there are duplicate values in an
2275          enumeration.)
2276         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
2277         will be generated for each C enum constant xxx.
2278         * Conversion functions m2i and i2m convert between mlrep and
2279         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
2280         * Coversion functions c and ml convert between mlrep and "tag enum".
2281         * Access functions (get/set) fetch and store mlrep values.
2282       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
2283         enumerations are merged into one single enumeration represented by
2284         structure E_'.
2285    
2286    ----------------------------------------------------------------------
2287    Name: Allen Leung
2288    Date: 2002/02/25 04:45:00 EST
2289    Tag: leunga-20020225-cps-spill
2290    
2291    This is a new implementation of the CPS spill phase.
2292    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
2293    In case of problems, replace it with the old file spill.sml
2294    
2295    The current compiler runs into some serious performance problems when
2296    constructing a large record.  This can happen when we try to compile a
2297    structure with many items.  Even a very simple structure like the following
2298    makes the compiler slow down.
2299    
2300        structure Foo = struct
2301           val x_1 = 0w1 : Word32.int
2302           val x_2 = 0w2 : Word32.int
2303           val x_3 = 0w3 : Word32.int
2304           ...
2305           val x_N = 0wN : Word32.int
2306        end
2307    
2308    The following table shows the compile time, from N=1000 to N=4000,
2309    with the old compiler:
2310    
2311    N
2312    1000   CPS 100 spill                           0.04u  0.00s  0.00g
2313           MLRISC ra                               0.06u  0.00s  0.05g
2314              (spills = 0 reloads = 0)
2315           TOTAL                                   0.63u  0.07s  0.21g
2316    
2317    1100   CPS 100 spill                           8.25u  0.32s  0.64g
2318           MLRISC ra                               5.68u  0.59s  3.93g
2319              (spills = 0 reloads = 0)
2320           TOTAL                                   14.71u  0.99s  4.81g
2321    
2322    1500   CPS 100 spill                           58.55u  2.34s  1.74g
2323           MLRISC ra                               5.54u  0.65s  3.91g
2324              (spills = 543 reloads = 1082)
2325           TOTAL                                   65.40u  3.13s  6.00g
2326    
2327    2000   CPS 100 spill                           126.69u  4.84s  3.08g
2328           MLRISC ra                               0.80u  0.10s  0.55g
2329              (spills = 42 reloads = 84)
2330           TOTAL                                   129.42u  5.10s  4.13g
2331    
2332    3000   CPS 100 spill                           675.59u  19.03s  11.64g
2333           MLRISC ra                               2.69u  0.27s  1.38g
2334              (spills = 62 reloads = 124)
2335           TOTAL                                   682.48u  19.61s  13.99g
2336    
2337    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
2338           MLRISC ra                               4.96u  0.27s  2.72g
2339              (spills = 85 reloads = 170)
2340           TOTAL                                   2375.26u  57.21s  48.00g
2341    
2342    As you can see the old cps spill module suffers from some serious
2343    performance problem.  But since I cannot decipher the old code fully,
2344    instead of patching the problems up, I'm reimplementing it
2345    with a different algorithm.  The new code is more modular,
2346    smaller when compiled, and substantially faster
2347    (O(n log n) time and O(n) space).  Timing of the new spill module:
2348    
2349    4000  CPS 100 spill                           0.02u  0.00s  0.00g
2350          MLRISC ra                               0.25u  0.02s  0.15g
2351             (spills=1 reloads=3)
2352          TOTAL                                   7.74u  0.34s  1.62g
2353    
2354    Implementation details:
2355    
2356    As far as I can tell, the purpose of the CPS spill module is to make sure the
2357    number of live variables at any program point (the bandwidth)
2358    does not exceed a certain limit, which is determined by the
2359    size of the spill area.
2360    
2361    When the bandwidth is too large, we decrease the register pressure by
2362    packing live variables into spill records.  How we achieve this is
2363    completely different than what we did in the old code.
2364    
2365    First, there is something about the MLRiscGen code generator
2366    that we should be aware of:
2367    
2368    o MLRiscGen performs code motion!
2369    
2370       In particular, it will move floating point computations and
2371       address computations involving only the heap pointer to
2372       their use sites (if there is only a single use).
2373       What this means is that if we have a CPS record construction
2374       statement
2375    
2376           RECORD(k,vl,w,e)
2377    
2378       we should never count the new record address w as live if w
2379       has only one use (which is often the case).
2380    
2381       We should do something similar to floating point, but the transformation
2382       there is much more complex, so I won't deal with that.
2383    
2384    Secondly, there are now two new cps primops at our disposal:
2385    
2386     1. rawrecord of record_kind option
2387        This pure operator allocates some uninitialized storage from the heap.
2388        There are two forms:
2389    
2390         rawrecord NONE [INT n]  allocates a tagless record of length n
2391         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2392                                     and initializes the tag.
2393    
2394     2. rawupdate of cty
2395          rawupdate cty (v,i,x)
2396          Assigns to x to the ith component of record v.
2397          The storelist is not updated.
2398    
2399    We use these new primops for both spilling and increment record construction.
2400    
2401     1. Spilling.
2402    
2403        This is implemented with a linear scan algorithm (but generalized
2404        to trees).  The algorithm will create a single spill record at the
2405        beginning of the cps function and use rawupdate to spill to it,
2406        and SELECT or SELp to reload from it.  So both spills and reloads
2407        are fine-grain operations.  In contrast, in the old algorithm
2408        "spills" have to be bundled together in records.
2409    
2410        Ideally, we should sink the spill record construction to where
2411        it is needed.  We can even split the spill record into multiple ones
2412        at the places where they are needed.  But CPS is not a good
2413        representation for global code motion, so I'll keep it simple and
2414        am not attempting this.
2415    
2416     2. Incremental record construction (aka record splitting).
2417    
2418        Long records with many component values which are simulatenously live
2419        (recall that single use record addresses are not considered to
2420         be live) are constructed with rawrecord and rawupdate.
2421        We allocate space on the heap with rawrecord first, then gradually
2422        fill it in with rawupdate.  This is the technique suggested to me
2423        by Matthias.
2424    
2425        Some restrictions on when this is applicable:
2426        1. It is not a VECTOR record.  The code generator currently does not handle
2427           this case. VECTOR record uses double indirection like arrays.
2428        2. All the record component values are defined in the same "basic block"
2429           as the record constructor.  This is to prevent speculative
2430           record construction.
2431    
2432    ----------------------------------------------------------------------
2433    Name: Allen Leung
2434    Date: 2002/02/22 01:02:00 EST
2435    Tag: leunga-20020222-mlrisc-tools
2436    
2437    Minor bug fixes in the parser and rewriter
2438    
2439    ----------------------------------------------------------------------
2440    Name: Allen Leung
2441    Date: 2002/02/21 20:20:00 EST
2442    Tag: leunga-20020221-peephole
2443    
2444    Regenerated the peephole files.  Some contained typos in the specification
2445    and some didn't compile because of pretty printing bugs in the old version
2446    of 'nowhere'.
2447    
2448    ----------------------------------------------------------------------
2449    Name: Allen Leung
2450    Date: 2002/02/19 20:20:00 EST
2451    Tag: leunga-20020219-mlrisc-tools
2452    Description:
2453    
2454       Minor bug fixes to the mlrisc-tools library:
2455    
2456       1.  Fixed up parsing colon suffixed keywords
2457       2.  Added the ability to shut the error messages up
2458       3.  Reimplemented the pretty printer and fixed up/improved
2459           the pretty printing of handle and -> types.
2460       4.  Fixed up generation of literal symbols in the nowhere tool.
2461       5.  Added some SML keywords to to sml.sty
2462    
2463    ----------------------------------------------------------------------
2464    Name: Matthias Blume
2465    Date: 2002/02/19 16:20:00 EST
2466    Tag: blume-20020219-cmffi
2467    Description:
2468    
2469    A wild mix of changes, some minor, some major:
2470    
2471    * All C FFI-related libraries are now anchored under $c:
2472        $/c.cm      --> $c/c.cm
2473        $/c-int.cm  --> $c/internals/c-int.cm
2474        $/memory.cm --> $c/memory/memory.cm
2475    
2476    * "make" tool (in CM) now treats its argument pathname slightly
2477      differently:
2478        1. If the native expansion is an absolute name, then before invoking
2479           the "make" command on it, CM will apply OS.Path.mkRelative
2480           (with relativeTo = OS.FileSys.getDir()) to it.
2481        2. The argument will be passed through to subsequent phases of CM
2482           processing without "going native".  In particular, if the argument
2483           was an anchored path, then "make" will not lose track of that anchor.
2484    
2485    * Compiler backends now "know" their respective C calling conventions
2486      instead of having to be told about it by ml-nlffigen.  This relieves
2487      ml-nlffigen from one of its burdens.
2488    
2489    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
2490    
2491    * Export C_DEBUG and C_Debug from $c/c.cm.
2492    
2493    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
2494      subtyping relationship between incomplete pointers and their complete
2495      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
2496      with the convention of instantiating 'o with ('t, 'c) obj whenever
2497      the pointer target type is complete.  In the incomplete case, 'o
2498      will be instantiated with some "'c iobj" -- a type obtained by
2499      using one of the functors PointerToIncompleteType or PointerToCompleteType.
2500    
2501      Operations that work on both incomplete and complete pointer types are
2502      typed as taking an 'o ptr while operations that require the target to
2503      be known are typed as taking some ('t, 'c) obj ptr.
2504    
2505      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
2506      where void is an eqtype without any values.  This makes it possible
2507      to work on voidptr values using functions meant to operate on light
2508      incomplete pointers.
2509    
2510    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
2511      been vastly simplified.
2512    
2513    ----------------------------------------------------------------------
2514    Name: Matthias Blume
2515    Date: 2002/02/19 10:48:00 EST
2516    Tag: blume-20020219-pqfix
2517    Description:
2518    
2519    Applied Chris Okasaki's bug fix for priority queues.
2520    
2521    ----------------------------------------------------------------------
2522    Name: Matthias Blume
2523    Date: 2002/02/15 17:05:00
2524    Tag: Release_110_39
2525    Description:
2526    
2527    Last-minute retagging is becoming a tradition... :-(
2528    
2529    This is the working release 110.39.
2530    
2531    ----------------------------------------------------------------------
2532    Name: Matthias Blume
2533    Date: 2002/02/15 16:00:00 EST
2534    Tag: Release_110_39-orig
2535    Description:
2536    
2537    Working release 110.39.  New bootfiles.
2538    
2539    (Update: There was a small bug in the installer so it wouldn't work
2540    with all shells.  So I retagged. -Matthias)
2541    
2542    ----------------------------------------------------------------------
2543    Name: Matthias Blume
2544    Date: 2002/02/15 14:17:00 EST
2545    Tag: blume-20020215-showbindings
2546    Description:
2547    
2548    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
2549    the latter can be useful for exploring what bindings are available at
2550    the interactive prompt.  (The first function returns only the list
2551    of symbols that are really bound, the second prints those but also the
2552    ones that CM's autoloading mechanism knows about.)
2553    
2554    ----------------------------------------------------------------------
2555    Name: Matthias Blume
2556    Date: 2002/02/15 12:08:00 EST
2557    Tag: blume-20020215-iptrs
2558    Description:
2559    
2560    Two improvements to ml-nlffigen:
2561    
2562      1. Write files only if they do not exist or if their current contents
2563         do not coincide with what's being written.  (That is, avoid messing
2564         with the time stamps unless absolutely necessary.)
2565    
2566      2. Implement a "repository" mechanism for generated files related
2567         to "incomplete pointer types".   See the README file for details.
2568    
2569    ----------------------------------------------------------------------
2570    Name: Matthias Blume
2571    Date: 2002/02/14 11:50:00 EST
2572    Tag: blume-20020214-quote
2573    Description:
2574    
2575    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
2576    because of the new and improved tag generation scheme.  (Thanks to Allen
2577    Leung for pointing it out.)
2578    
2579    ----------------------------------------------------------------------
2580    Name: Lal George
2581    Date: 2002/02/14 09:55:27 EST 2002
2582    Tag: george-20020214-isabelle-bug
2583    Description:
2584    
2585    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2586    of Isabelle on the x86.
2587    
2588    From Allen:
2589    -----------
2590     I've found the problem:
2591    
2592         in ra-core.sml, I use the counter "blocked" to keep track of the
2593         true number of elements in the freeze queue.  When the counter goes
2594         to zero, I skip examining the queue.  But I've messed up the
2595         bookkeeping in combine():
2596    
2597             else ();
2598             case !ucol of
2599               PSEUDO => (if !cntv > 0 then
2600                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2601                                        ^^^^^^^^^^^^^^^^^^^^^^^
2602                      moveu := mergeMoveList(!movev, !moveu)
2603                     )
2604                  else ();
2605    
2606         combine() is called to coalesce two nodes u and v.
2607         I think I was thinking that if the move counts of u and v are both
2608         greater than zero then after they are coalesced then one node is
2609         removed from the freeze queue.  Apparently I was thinking that
2610         both u and v are of low degree, but that's clearly not necessarily true.
2611    
2612    
2613    02/12/2002:
2614        Here's the patch.  HOL now compiles.
2615    
2616        I don't know how this impact on performance (compile
2617        time or runtime).  This bug caused the RA (especially on the x86)
2618        to go thru the potential spill phase when there are still nodes on the
2619        freeze queue.
2620    
2621    
2622    
2623    
2624    ----------------------------------------------------------------------
2625    Name: Matthias Blume
2626    Date: 2002/02/13 22:40:00 EST
2627    Tag: blume-20020213-fptr-rtti
2628    Description:
2629    
2630    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2631    updates.
2632    
2633    ----------------------------------------------------------------------
2634    Name: Matthias Blume
2635    Date: 2002/02/13 16:41:00 EST
2636    Tag: blume-20020213-cmlpq
2637    Description:
2638    
2639    Added new priority queue export symbols (which have just been added to
2640    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2641    not compile and the installer would choke.)
2642    
2643    ----------------------------------------------------------------------
2644    Name: Matthias Blume
2645    Date: 2002/02/13 16:15:00 EST
2646    Tag: blume-20020213-various
2647    Description:
2648    
2649    1. More tweaks to ml-nlffigen:
2650    
2651       - better internal datastructures (resulting in slight speedup)
2652       - "-match" option requires exact match
2653       - "localized" gensym counters (untagged structs/unions nested within
2654         other structs/unions or within typedefs get a fresh counter; their
2655         tag will be prefixed by a concatenation of their parents' tags)
2656       - bug fixes (related to calculation of transitive closure of types
2657         to be included in the output)
2658    
2659    2. Minor Basis updates:
2660    
2661       - added implementations for List.collate and Option.app
2662    
2663    ----------------------------------------------------------------------
2664    Name: Matthias Blume
2665    Date: 2002/02/11 15:55:00 EST
2666    Tag: blume-20020211-gensym
2667    Description:
2668    
2669    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2670    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2671    names (ML structure names that correspond to unnamed C structs, unions,
2672    and enums), so that separate runs of ml-nlffigen do not clash.
2673    
2674    ----------------------------------------------------------------------
2675    Name: Matthias Blume
2676    Date: 2002/02/11 12:05:00 EST
2677    Tag: blume-20020211-gensml
2678    Description:
2679    
2680    A quick fix for a problem with GenSML (in the pgraph-util library):
2681    Make generation of toplevel "local" optional.  (Strictly speaking,
2682    signature definitions within "local" are not legal SML.)
2683    
2684    Other than that: updates to INSTALL and cm/TODO.
2685    
2686    ----------------------------------------------------------------------
2687    Name: Matthias Blume
2688    Date: 2002/02/08 15:00:00 EST
2689    Tag: blume-20020208-uniquepid
2690    Description:
2691    
2692    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2693    
2694    1. The installer (config/install.sh) has gotten smarter:
2695    
2696         - Configuration options are a bit easier to specify now
2697           (in config/targets).
2698         - Bug in recognizing .tar.bz2 files fixed.
2699         - Installer automatically resolves dependencies between
2700           configuration options (e.g., if you ask for eXene, you will
2701           also get cml -- regardless whether you asked for it or not).
2702         - Installer can run in "quieter mode" by setting the environment
2703           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2704           "completely silent", though.
2705         - Build HashCons library as part of smlnj-lib.
2706    
2707    2. A new scheme for assigning persistent identifiers to compilation
2708       units (and, by extension, to types etc.) has been put into place.
2709       This fixes a long-standing bug where types and even dynamic values
2710       can get internally confused, thereby compromising type safety
2711       (abstraction) and dynamic correctness.  See
2712    
2713         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2714    
2715       for an example of how things could go wrong until now.
2716    
2717       The downside of the new scheme is that pids are not quite as
2718       persistent as they used to be: CM will generate a fresh pid
2719       for every compilation unit that it thinks it sees for the first
2720       time.  That means that if you compile starting from a clean, fresh
2721       source tree at two different times, you end up with different
2722       binaries.
2723    
2724       Cutoff recompilation, however, has not been compromised because
2725       CM keeps pid information in special caches between runs.
2726    
2727    ----------------------------------------------------------------------
2728    Name: Lal George
2729    Date: 2002/02/07 15:34:13 EST 2002
2730    Tag: <none>
2731    Description:
2732    
2733    Compilers that generate assembly code may produce  global labels
2734    whose value is resolved at link time. The various peephole optimization
2735    modules did not take this in account.
2736    
2737    TODO. The Labels.addrOf function should really return an option
2738    type so that clients are forced to deal with this issue, rather
2739    than an exception being raised.
2740    
2741    ----------------------------------------------------------------------
2742    Name: Lal George
2743    Date: 2002/02/06 13:55:02 EST
2744    Tag: george-20020206-ra-breakup
2745    Description:
2746    
2747    1. A bug fix from Allen.
2748    
2749        A typo causes extra fstp %st(0)'s to be generated at compensation
2750        edges, which might cause stack underflow traps at runtime.  This
2751        occurs in fft where there are extraneous fstps right before the 'into'
2752        trap instruction (in this case they are harmless since none of the
2753        integers overflow.)
2754    
2755    2. Pulled out various utility modules that were embedded in the modules
2756       of the register allocator. I need these modules for other purposes, but
2757       they are not complete enough to put into a library (just yet).
2758    ----------------------------------------------------------------------
2759    Name: Matthias Blume
2760    Date: 2002/01/31 16:05:00 EST
2761    Tag: blume-20020131-sparc-ccalls
2762    Description:
2763    
2764    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2765       of extra stack space by mistake.  Fixed.
2766    
2767    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2768    
2769    ----------------------------------------------------------------------
2770    Name: Allen Leung
2771    Date: 2002/01/30
2772    Tag: leunga-20020130-nowhere-bug-fix
2773    Description:
2774    
2775       MLRISC bug fixes:
2776       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2777       2. MachineInt.fromString was negating its value.
2778    
2779    ----------------------------------------------------------------------
2780    Name: Matthias Blume
2781    Date: 2002/01/29
2782    Tag: blume-20020129-INSTALL
2783    Description:
2784    
2785    - Added somewhat detailed installation instructions (file INSTALL).
2786    - Fixed curl-detection bug in config/install.sh.
2787    - It is now possible to select the URL getter using the URLGETTER
2788      environment variable:
2789    
2790          not set / "unknown"      --> automatic detection (script tries wget,
2791                                       curl, and lynx)
2792          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2793                                       how to properly invoke them)
2794          other                    --> use $URLGETTER directly, it must take
2795                                       precisely two command-line arguments
2796                                       (source URL and destination file name)
2797    
2798    ----------------------------------------------------------------------
2799    Name: Matthias Blume
2800    Date: 2002/01/28
2801    Tag: blume-20020128-sparc-ccalls
2802    Description:
2803    
2804    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2805    - Make use of the allocParam argument in sparc-c-calls.
2806    
2807    ----------------------------------------------------------------------
2808    Name: Matthias Blume
2809    Date: 2002/01/28
2810    Tag: blume-20020128-allocParam
2811    Description:
2812    
2813    John Reppy:  Changes c-calls API to accept client-callback for
2814    allocating extra stack space.
2815    me: Corresponding changes to mlriscGen (using a dummy argument that
2816        does not change the current behavior).
2817    
2818    ----------------------------------------------------------------------
2819    Name: Matthias Blume
2820    Date: 2002/01/28 12:00:00
2821    Tag: Release_110_38
2822    Description:
2823    
2824    This time for real!!!
2825    
2826    ----------------------------------------------------------------------
2827    Name: Matthias Blume
2828    Date: 2002/01/28 10:56:00 EST
2829    Tag: blume-20020128-retraction
2830    Description:
2831    
2832    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
2833       with blume-Release_110_38-retracted.)
2834    
2835    1. Fixed a problem with incorrect rounding modes in real64.sml.
2836       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2837    
2838    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2839       and enums fixed.  The naming of corresponding ML identifiers should
2840       now be consistent again.
2841    
2842    ----------------------------------------------------------------------
2843    Name: Allen Leung
2844    Date: 2002/01/27
2845    Tag: leunga-20020127-nowhere
2846    Description:
2847    
2848       Added a target called nowhere in the configuration scripts.
2849       Enabling this will build the MLRISC 'nowhere' tool (for translating
2850       programs with where-clauses into legal SML code) during installation.
2851    
2852    ----------------------------------------------------------------------
2853    Name: Matthias Blume
2854    Date: 2002/01/25 21:27:00 EST
2855    Tag: blume-Release_110_38-retracted
2856    Description:
2857    
2858    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
2859    
2860    README will be added later.
2861    
2862    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
2863    (Release_110_38) removed.  Reason: Last-minute bug fixes.
2864    
2865    ----------------------------------------------------------------------
2866    Name: Matthias Blume
2867    Date: 2002/01/25
2868    Tag: blume-20020125-ffi
2869    Description:
2870    
2871    A large number of tweaks and improvements to ml-nlffi-lib and
2872    ml-nlffigen:
2873    
2874       - ML represenation types have been streamlined
2875       - getter and setter functions work with concrete values, not abstract
2876         ones where possible
2877       - ml-nlffigen command line more flexible (see README file there)
2878       - some bugs have been fixed (hopefully)
2879    
2880    ----------------------------------------------------------------------
2881    Name: Lal George
2882    Date: 2002/01/24
2883    Tag: george-20020124-risc-ra-interface
2884    Description:
2885    
2886       There is a dramatic simplification in the interface to the
2887       register allocator for RISC architectures as a result of making
2888       parallel copy instructions explicit.
2889    
2890    ----------------------------------------------------------------------
2891    Name: Matthias Blume
2892    Date: 2002/01/22
2893    Tag: blume-20020122-x86-ccalls
2894    Description:
2895    
2896    Bug fix for c-calls on x86 (having to do with how char- and
2897    short-arguments are being handled).
2898    
2899    ----------------------------------------------------------------------
2900    Name: Matthias Blume
2901    Date: 2002/01/21
2902    Tag: blume-20020121-ff
2903    Description:
2904    
2905    Another day of fiddling with the FFI...
2906    
2907    1. Bug fix/workaround:  CKIT does not complain about negative array
2908       dimensions, so ml-nlffigen has to guard itself against this possibility.
2909       (Otherwise a negative dimension would send it into an infinite loop.)
2910    
2911    2. Some of the abstract types (light objects, light pointers, most "base"
2912       types) in structure C are now eqtypes.
2913    
2914    3. Added constructors and test functions for NULL function pointers.
2915    
2916    ----------------------------------------------------------------------
2917    Name: Matthias Blume
2918    Date: 2002/01/18
2919    Tag: blume-20020118-ready-for-new-release
2920    Description:
2921    
2922    Made config/srcarchiveurl point to a new place.  (Will provide boot
2923    files shortly.)
2924    
2925    Maybe we christen this to be 110.38?
2926    
2927    ----------------------------------------------------------------------
2928    Name: Matthias Blume
2929    Date: 2002/01/18
2930    Tag: blume-20020118-more-ffifiddle
2931    Description:
2932    
2933    Today's FFI fiddling:
2934    
2935      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
2936        C.Set.* that use concrete (MLRep.*) arguments and results instead
2937        of abstract ones.
2938    
2939      - Provided word-style bit operations etc. for "int" representation
2940        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
2941        and Long.
2942    
2943    ----------------------------------------------------------------------
2944    Name: Matthias Blume
2945    Date: 2002/01/18
2946    Tag: blume-20020118-use-x86-fp
2947    Description:
2948    
2949    Now that x86-fast-fp seems to be working, I turned it back on again
2950    by default.  (Seems to work fine now, even with the FFI.)
2951    
2952    Other than that, I added some documentation about the FFI to
2953    src/ml-nlffigen/README and updated the FFI test examples in
2954    src/ml-nlffi-lib/Tests/*.
2955    
2956    ----------------------------------------------------------------------
2957    Name: Allen Leung
2958    Date: 2002/01/17
2959    Tag: leunga-20020117-x86-fast-fp-call
2960    Description:
2961    
2962       1. Fixed a problem with handling return fp values when x86's fast fp
2963          mode is turned on.
2964    
2965       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
2966          of %f32.
2967    
2968       3. Added a constructor INT32lit to the ast of MLRISC tools.
2969    
2970    ----------------------------------------------------------------------
2971    Name: Matthias Blume
2972    Date: 2002/01/16
2973    Tag: blume-20020116-ffifiddle
2974    Description:
2975    
2976    More fiddling with the FFI interface:
2977    
2978     - Make constness 'c instead of rw wherever possible.  This eliminates
2979       the need for certain explicit coercions.  (However, due to ML's
2980       value polymorphism, there will still be many cases where explicit
2981       coercions are necessary.  Phantom types are not the whole answer
2982       to modeling a subtyping relationship in ML.)
2983    
2984     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
2985    
2986     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
2987       clumsy workaround for varargs problem.)
2988    
2989    ----------------------------------------------------------------------
2990    Name: Lal George
2991    Date: 2002/01/15
2992    Tag: <none>
2993    Description:
2994    
2995    1. Since COPY instructions are no longer native to the architecture,
2996       a generic functor can be used to implement the expandCopies function.
2997    
2998    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
2999       TEXT segment.
3000    
3001    ----------------------------------------------------------------------
3002    Name: Matthias Blume
3003    Date: 2002/01/15
3004    Tag: blume-20020115-ffiupdates
3005    Description:
3006    
3007    1. Fix for bug resulting in single-precision float values being returned
3008       incorrectly from FFI calls.
3009    
3010    2. Small modifications to C FFI API:
3011    
3012        - memory-allocation routines return straight objects (no options)
3013          and raise an exception in out-of-memory situations
3014        - unsafe extensions to cast between function pointers and pointers
3015          from/to ints
3016        - added structure C_Debug as an alternative to structure C where
3017          pointer-dereferencing (|*| and |*!) always check for null-pointers
3018        - added open_lib' to DynLinkage;  open_lib' works like open_lib
3019          but also takes a (possibly empty) list of existing library handles
3020          that the current library depends on
3021    
3022    ----------------------------------------------------------------------
3023    Name: Matthias Blume
3024    Date: 2002/01/10
3025    Tag: blume-20020110-newffigen
3026    Description:
3027    
3028    1. Updates to portable graph code.
3029    
3030    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
3031       more scalable now so that even huge interfaces such as the one
3032       for GTK compile in finite time and space. :-)
3033       See src/ml-nlffigen/README for details on what's new.
3034    
3035    ----------------------------------------------------------------------
3036    Name: Lal George
3037    Date: 2001/01/09 14:31:35 EST 2002
3038    Tag: george-20011206-rm-native-copy
3039    Description:
3040    
3041            Removed the native COPY and FCOPY instructions
3042            from all the architectures and replaced it with the
3043            explicit COPY instruction from the previous commit.
3044    
3045            It is now possible to simplify many of the optimizations
3046            modules that manipulate copies. This has not been
3047            done in this change.
3048    
3049    ----------------------------------------------------------------------
3050    Name: Lal George
3051    Date: 2001/12/06 16:50:13 EST 2001
3052    Tag: george-20011206-mlrisc-instruction
3053    Description:
3054    
3055    Changed the representation of instructions from being fully abstract
3056    to being partially concrete. That is to say:
3057    
3058      from
3059            type instruction
3060    
3061      to
3062            type instr                              (* machine instruction *)
3063    
3064            datatype instruction =
3065                LIVE of {regs: C.cellset, spilled: C.cellset}
3066              | KILL of {regs: C.cellset, spilled: C.cellset}
3067              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
3068              | ANNOTATION of {i: instruction, a: Annotations.annotation}
3069              | INSTR of instr
3070    
3071    This makes the handling of certain special instructions that appear on
3072    all architectures easier and uniform.
3073    
3074    LIVE and KILL say that a list of registers are live or killed at the
3075    program point where they appear. No spill code is generated when an
3076    element of the 'regs' field is spilled, but the register is moved to
3077    the 'spilled' (which is present, more for debugging than anything else).
3078    
3079    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
3080    We used to generate:
3081    
3082            DEFFREG f1
3083            f1 := f2 + f3
3084            trapb
3085    
3086    but now generate:
3087    
3088            f1 := f2 + f3
3089            trapb
3090            LIVE {regs=[f1,f2,f3], spilled=[]}
3091    
3092    Furthermore, the DEFFREG (hack) required that all floating point instruction
3093    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
3094    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
3095    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
3096    this architecture).
3097    
3098    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
3099    all the architectures. This will result in further simplification of the
3100    register allocator that must be aware of them for coalescing purposes, and
3101    will also simplify certain aspects of the machine description that provides
3102    callbacks related to parallel copies.
3103    
3104    ANNOTATION should be obvious, and now INSTR represents the honest to God
3105    machine instruction set!
3106    
3107    The <arch>/instructions/<arch>Instr.sml files define certain utility
3108    functions for making porting easier -- essentially converting upper case
3109    to lower case. All machine instructions (of type instr) are in upper case,
3110    and the lower case form generates an MLRISC instruction. For example on
3111    the alpha we have:
3112    
3113      datatype instr =
3114         LDA of {r:cell, b:cell, d:operand}
3115       | ...
3116    
3117      val lda : {r:cell, b:cell, d:operand} -> instruction
3118        ...
3119    
3120    where lda is just (INSTR o LDA), etc.
3121    
3122    ----------------------------------------------------------------------
3123    Name: Matthias Blume
3124    Date: 2001/11/22 21:40:00 EST
3125    Tag: Release_110_37
3126    Description:
3127    
3128    Release 110.37.  This time for real.
3129    
3130    ----------------------------------------------------------------------
3131    Name: Matthias Blume
3132    Date: 2001/11/21 16:35:00 EST
3133    Tag: blume-20011121-foot-in-mouth
3134    Description:
3135    
3136    Removed the "Release_110_37" tag because of a serious bug.
3137    This will be re-tagged once the bug is fixed.
3138    
3139    ----------------------------------------------------------------------
3140    Name: Matthias Blume
3141    Date: 2001/11/21 16:14:00 EST
3142    Tag: blume-20011121-forgottenfile
3143    Description:
3144    
3145    Forgot to add a file.  (Just a .tex-file -- part of
3146    the CM manual source.)
3147    
3148    ----------------------------------------------------------------------
3149    Name: Matthias Blume
3150    Date: 2001/11/21 16:10:00 EST
3151    Tag: blume-20011121-invalid_110_37
3152    Description:
3153    
3154    Note: I removed the original tag "Release_110_37" from this commit
3155          because we found a serious bug in all non-x86 backends.
3156          - Matthias
3157    
3158    1. Modifications to the SML/NJ code generator and to the runtime system
3159       so that code object name strings are directly inserted into code
3160       objects at code generation time.  The only business the runtime system
3161       has with this is now to read the name strings on occasions.
3162       (The encoding of the name string has also changed somewhat.)
3163    
3164    2. CM now implements a simple "set calculus" for specifying export lists.
3165       In particular, it is now possible to refer to the export lists of
3166       other libraries/groups/sources and form unions as well as differences.
3167       See the latest CM manual for details.
3168    
3169    3. An separate notion of "proxy" libraries has again be eliminated from
3170       CM's model.  (Proxy libraries are now simply a special case of using
3171       the export list calculus.)
3172    
3173    4. Some of the existing libraries now take advantage of the new set
3174       calculus.
3175       (Notice that not all libraries have been converted because some
3176       of the existing .cm-files are supposed to be backward compatible
3177       with 110.0.x.)
3178    
3179    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
3180       "General.exnMessage"!  The former relies on a certain hook to be
3181       initialized, and that often does not happen in the stand-alone case.)
3182    
3183    ----------------------------------------------------------------------
3184    Name: Lal George
3185    Date: 2001/11/21  13:56:18 EST
3186    Tag: george-2001121-pseudo-ops
3187    Description:
3188    
3189      Implemented a complete redesign of MLRISC pseudo-ops. Now there
3190      ought to never be any question of incompatabilities with
3191      pseudo-op syntax expected by host assemblers.
3192    
3193      For now, only modules supporting GAS syntax are implemented
3194      but more should follow, such as MASM, and vendor assembler
3195      syntax, e.g. IBM as, Sun as, etc.
3196    
3197    ----------------------------------------------------------------------
3198    Name: Matthias Blume
3199    Date: 2001/11/14 11:52:00 EST
3200    Tag: blume-20011114-srcname
3201    Description:
3202    
3203    1. Routed the name of the current source file to mlriscgen where it
3204       should be directly emitted into the code object.  (This last part
3205       is yet to be done.)
3206    
3207    2. Some cleanup of the pgraph code to make it match the proposal that
3208       I put out the other day.  (The proposal notwithstanding, things are
3209       still in flux here.)
3210    
3211    ----------------------------------------------------------------------
3212    Name: Lal George
3213    Date: 2001/11/14 09:44:04 EST
3214    Tag:
3215    Description:
3216    
3217      Fix for a backpatching bug reported by Allen.
3218    
3219      Because the boundary between short and long span-dependent
3220      instructions is +/- 128, there are an astounding number of
3221      span-dependent instructions whose size is over estimated.
3222    
3223      Allen came up with the idea of letting the size of span
3224      dependent instructions be non-monotonic, for a maxIter
3225      number of times, after which the size must be monotonically
3226      increasing.
3227    
3228      This table shows the number of span-dependent instructions
3229      whose size was over-estimated as a function of maxIter, for the
3230      file Parse/parse/ml.grm.sml:
3231    
3232         maxIter            # of instructions:
3233            10                      687
3234            20                      438
3235            30                      198
3236            40                        0
3237    
3238      In compiling the compiler, there is no significant difference in
3239      compilation speed between maxIter=10 and maxIter=40. Actually,
3240      my measurements showed that maxIter=40 was a tad faster than
3241      maxIter=10! Also 96% of the  files in the compiler reach a fix
3242      point within 13 iterations, so fixing maxIter at 40, while high,
3243      is okay.
3244    
3245    ----------------------------------------------------------------------
3246    Name: Matthias Blume
3247    Date: 2001/10/31 15:25:00 EST
3248    Tag: blume-20011031-pgraph
3249    Description:
3250    
3251    CKIT:
3252    * Changed the "Function" constructor of type Ast.ctype to carry optional
3253      argument identifiers.
3254    * Changed the return type of TypeUtil.getFunction accordingly.
3255    * Type equality ignores the argument names.
3256    * TypeUtil.composite tries to preserve argument names but gives up quickly
3257      if there is a mismatch.
3258    
3259    installation script:
3260    * attempts to use "curl" if available (unless "wget" is available as well)
3261    
3262    CM:
3263    * has an experimental implementation of "portable graphs" which I will
3264      soon propose as an implementation-independent library format
3265    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
3266    
3267    NLFFI-LIB:
3268    * some cleanup (all cosmetic)
3269    
3270    NLFFIGEN:
3271    * temporarily disabled the mechanism that suppresses ML output for
3272      C definitions whose identifiers start with an underscore character
3273    * generate val bindings for enum constants
3274    * user can request that only one style (light or heavy) is being used;
3275      default is to use both (command-line arguments: -heavy and -light)
3276    * fixed bug in handling of function types involving incomplete pointers
3277    * generate ML entry points that take record arguments (i.e., using
3278      named arguments) for C functions that have a prototype with named
3279      arguments
3280      (see changes to CKIT)
3281    
3282    ----------------------------------------------------------------------
3283    Name: Allen Leung
3284    Date: 2001/10/27 20:34:00 EDT
3285    Tag: leunga-20011027-x86-fast-fp-call
3286    Description:
3287    
3288       Fixed the bug described in blume-20010920-slowfp.
3289    
3290       The fix involves
3291          1. generating FCOPYs in FSTP in ia32-svid
3292          2. marking a CALL with the appropriate annotation
3293    
3294    ----------------------------------------------------------------------
3295    Name: Matthias Blume
3296    Date: 2001/10/16 11:32:00 EDT
3297    Tag: blume-20011016-netbsd
3298    Description:
3299    
3300    Underscore patch from Chris Richards (fixing problem with compiling
3301    runtime system under recent NetBSD).
3302    
3303    ----------------------------------------------------------------------
3304    Name: Allen Leung
3305    Date: 2001/10/12 17:18:32 EDT 2001
3306    Tag: leung-20011012-x86-printflowgraph
3307    Description:
3308    
3309    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
3310    
3311    ----------------------------------------------------------------------
3312    Name: Lal George
3313    Date: 2001/10/11 23:51:34 EDT
3314    Tag: george-20011011-too-many-instrs
3315    Description:
3316    
3317    The representation of a program point never expected to see more
3318    than 65536 instructions in a basic block!
3319    
3320    ----------------------------------------------------------------------
3321    Name: Lal George
3322    Date: 2001/10/09 09:41:37 EDT
3323    Tag: george-20011008-mlrisc-labels
3324    Description:
3325    
3326    Changed the machine description files to support printing of
3327    local and global labels in assembly code, based on host assembler
3328    conventions.
3329    
3330    ----------------------------------------------------------------------
3331    Name: Matthias Blume
3332    Date: 2001/09/25 15:25:00 EDT
3333    Tag: blume-20010925-exninfo
3334    Description:
3335    
3336    I provided a non-hook implementation of exnName (at the toplevel) and
3337    made the "dummy" implementation of exnMessage (at the toplevel) more
3338    useful: if nothing gets "hooked in", then at least you are going to
3339    see the exception name and a message indicating why you don't see more.
3340    
3341    [For the time being, programs that need exnMessage and want to use
3342    ml-build should either use General.exnMessage (strongly recommended) or
3343    refer to structure General at some other point so that CM sees a
3344    static dependency.]
3345    
3346    [Similar remarks go for "print" and "use":  If you want to use their
3347    functionality in stand-alone programs generated by ml-build, then use
3348    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
3349    
3350    ----------------------------------------------------------------------
3351    Name: Matthias Blume
3352    Date: 2001/09/20 17:28:00 EDT
3353    Tag: blume-20010920-slowfp
3354    Description:
3355    
3356    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
3357    
3358    ----------------------------------------------------------------------
3359    Name: Matthias Blume
3360    Date: 2001/09/20 17:20:00 EDT
3361    Tag: blume-20010920-canonicalpaths
3362    Description:
3363    
3364    0. Updated the BOOT file (something that I forgot to do earlier).
3365    
3366    1. Small internal change to CM so that it avoids "/../" in filenames
3367       as much as possible (but only where it is safe).
3368    
3369    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
3370       that arguments that contain delimiters are passed through correctly.
3371       This change also means that all "special" arguments of the form
3372       @SMLxxx... must come first.
3373    
3374    3. Changed install script to put relative anchor names for tool commands
3375       into pathconfig.
3376    
3377    ----------------------------------------------------------------------
3378    Name: Matthias Blume
3379    Date: 2001/09/18 15:35:00 EDT
3380    Tag: blume-20010918-readme11036
3381    Description:
3382    
3383    Added README files.
3384    
3385    ----------------------------------------------------------------------
3386    Name: Matthias Blume
3387    Date: 2001/09/18 11:45:00 EDT
3388    Tag: Release_110_36 (retag)
3389    Description:
3390    
3391    Fixed mistake in config/preloads. Retagged as 110.36.
3392    
3393    ----------------------------------------------------------------------
3394    Name: Matthias Blume
3395    Date: 2001/09/18 09:40:00 EDT
3396    Tag: Release_110_36_orig (tag changed)
3397    Description:
3398    
3399    New version (110.36).  New bootfiles.
3400    
3401    ----------------------------------------------------------------------
3402    Name: Matthias Blume
3403    Date: 2001/09/14 16:15:00 EDT
3404    Tag: blume-20010914-x86fastfp
3405    Description:
3406    
3407    John committed some changes that Allen made, in particular a (hopefully)
3408    correctly working version of the x86-fp module.
3409    
3410    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
3411    flag to "true".  Everything seems to compile to a fixpoint ok, and
3412    "mandelbrot" speeds up by about 15%.
3413    
3414    ----------------------------------------------------------------------
3415    Name: Matthias Blume
3416    Date: 2001/09/13 11:20:00 EDT
3417    Tag: blume-20010913-minimal
3418    Description:
3419    
3420    1. Stefan Monnier's patch to fix a miscompilation problem that
3421       was brought to light by John Reppy's work on Moby.
3422    
3423    2. Implemented a minimal "structure Compiler" that contains just
3424       "version" and "architecture".  The minimal version will be
3425       available when the full version is not.  This is for backward-
3426       compatibility with code that wants to test Compiler.version.
3427    
3428    ----------------------------------------------------------------------
3429    Name: Matthias Blume
3430    Date: 2001/08/28 14:03:00 EDT
3431    Tag: blume-20010828-ml-lex
3432    Description:
3433    
3434    Fix for bug 1581, received from Neophytos Michael.
3435    
3436    ----------------------------------------------------------------------
3437    Name: Matthias Blume
3438    Date: 2001/08/27 11:20:00 EDT
3439    Tag: blume-20010827-readme11035
3440    Description:
3441    
3442    Fleshed out the README file for 110.35.
3443    
3444    ----------------------------------------------------------------------
3445    Name: Matthias Blume
3446    Date: 2001/08/24 17:10:00 EDT
3447    Tag: Release_110_35
3448    Description:
3449    
3450    New version number (110.35).  New bootfiles.
3451    
3452    ----------------------------------------------------------------------
3453    Name: Lal George
3454    Date: 2001/08/24 13:47:18 EDT 2001
3455    Tag: george-20010824-MLRISC-graphs
3456    Description:
3457    
3458     removed clusters from MLRISC completely and replaced with graphs.
3459    
3460    ----------------------------------------------------------------------
3461    Name: Matthias Blume
3462    Date: 2001/08/23 17:50:00 EDT
3463    Tag: blume-20010823-toplevel
3464    Description:
3465    
3466    - some reorganization of the code that implements various kinds of
3467      environments in the compiler (static, dynamic, symbolic, combined)
3468    - re-implemented the EnvRef module so that evalStream works properly
3469      (if the stream contains references to "use", "CM.make", etc.)
3470    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
3471    
3472    ----------------------------------------------------------------------
3473    Name: Matthias Blume
3474    Date: 2001/08/20 15:50 EDT
3475    Tag: blume20010820-slipup
3476    Description:
3477    
3478    I forgot to commit a few files.  Here they are...
3479    
3480    ----------------------------------------------------------------------
3481    Name: Matthias Blume
3482    Date: 2001/08/20 15:35:00 EDT
3483    Tag: blume-20010820-debugprof
3484    Description:
3485    
3486    !!!! NEW BOOTFILES !!!!
3487    
3488    This is another round of reorganizing the compiler sources.  This
3489    time the main goal was to factor out all the "instrumentation"
3490    passes (for profiling and backtracing) into their own library.
3491    The difficulty was to do it in such a way that it does not depend
3492    on elaborate.cm but only on elabdata.cm.
3493    
3494    Therefore there have been further changes to both elaborate.cm and
3495    elabdata.cm -- more "generic" things have been moved from the former
3496    to the latter.  As a result, I was forced to split the assignment
3497    of numbers indicating "primtyc"s into two portions: SML-generic and
3498    SML/NJ-specific.  Since it would have been awkward to maintain,
3499    I bit the bullet and actually _changed_ the mapping between these
3500    numbers and primtycs.  The bottom line of this is that you need
3501    a new set of bin- and bootfiles.
3502    
3503    I have built new bootfiles for all architectures, so doing a fresh
3504    checkout and config/install.sh should be all you need.
3505    
3506    The newly created library's name is
3507    
3508        $smlnj/viscomp/debugprof.cm
3509    
3510    and its sources live under
3511    
3512        src/compiler/DebugProf
3513    
3514    ----------------------------------------------------------------------
3515    Name: Matthias Blume
3516    Date: 2001/08/15 17:15:00 EDT
3517    Tag: blume-20010815-compreorg
3518    Description:
3519    
3520    This is a first cut at reorganizing the CM libraries that make up the
3521    core of the compiler.  The idea is to separate out pieces that could
3522    be used independently by tools, e.g., the parser, the typechecker, etc.
3523    
3524    The current status is a step in this direction, but it is not quite
3525    satisfactory yet.  Expect more changes in the future.
3526    
3527    Here is the current (new) organization...
3528    
3529        What used to be $smlnj/viscomp/core.cm is now divided into
3530        six CM libraries:
3531    
3532             $smlnj/viscomp/basics.cm
3533                           /parser.cm
3534                           /elabdata.cm
3535                           /elaborate.cm
3536                           /execute.cm
3537                           /core.cm
3538    
3539        The CM files for these libraries live under src/system/smlnj/viscomp.
3540        All these libraries are proxy libraries that contain precisely
3541        one CM library component.  Here are the locations of the components
3542        (all within the src/compiler tree):
3543    
3544             Basics/basics.cm
3545             Parse/parser.cm
3546             ElabData/elabdata.cm
3547             Elaborator/elaborate.cm
3548             Execution/execute.cm
3549             core.cm
3550    
3551         [This organization is the same that has been used already
3552         for a while for the architecture-specific parts of the visible
3553         compiler and for the old version of core.cm.]
3554    
3555         As you will notice, many source files have been moved from their
3556         respective original locations to a new home in one of the above
3557         subtrees.
3558    
3559         The division of labor between the new libraries is the following:
3560    
3561             basics.cm:
3562                - Simple, basic definitions that pertain to many (or all) of
3563                  the other libraries.
3564             parser.cm:
3565                - The SML parser, producing output of type Ast.dec.
3566                - The type family for Ast is also defined and exported here.
3567             elabdata.cm:
3568                - The datatypes that describe input and output of the elaborator.
3569                  This includes types, absyn, and static environments.
3570             elaborator.cm:
3571                - The SML/NJ type checker and elaborator.
3572                  This maps an Ast.dec (with a given static environment) to
3573                  an Absyn.dec (with a new static environment).
3574                - This libraries implements certain modules that used to be
3575                  structures as functors (to remove dependencies on FLINT).
3576             execute.cm:
3577                - Everything having to do with executing binary code objects.
3578                - Dynamic environments.
3579             core.cm:
3580                - SML/NJ-specific instantiations of the elaborator and MLRISC.
3581                - Top-level modules.
3582                - FLINT (this should eventually become its own library)
3583    
3584    Notes:
3585    
3586    I am not 100% happy with the way I separated the elaborator (and its
3587    data structures) from FLINT.  Two instances of the same problem:
3588    
3589        1. Data structures contain certain fields that carry FLINT-specific
3590           information.  I hacked around this using exn and the property list
3591           module from smlnj-lib.  But the fact that there are middle-end
3592           specific fields around at all is a bit annoying.
3593    
3594        2. The elaborator calculates certain FLINT-related information.  I tried
3595           to make this as abstract as I could using functorization, but, again,
3596           the fact that the elaborator has to perform calculations on behalf
3597           of the middle-end at all is not nice.
3598    
3599        3. Having to used exn and property lists is unfortunate because it
3600           weakens type checking.  The other alternative (parameterizing
3601           nearly *everything*) is not appealing, though.
3602    
3603    I removed the "rebinding =" warning hack because due to the new organization
3604    it was awkward to maintain it.  As a result, the compiler now issues some of
3605    these warnings when compiling init.cmi during bootstrap compilation. On
3606    the plus side, you also get a warning when you do, for example:
3607       val op = = Int32.+
3608    which was not the case up to now.
3609    
3610    I placed "assign" and "deref" into the _Core structure so that the
3611    code that deals with the "lazy" keyword can find them there.  This
3612    removes the need for having access to the primitive environment
3613    during elaboration.
3614    
3615    ----------------------------------------------------------------------
3616    Name: Matthias Blume